perm filename W78.IN[LET,JMC] blob
sn#554842 filedate 1981-01-09 generic text, type C, neo UTF8
COMMENT ⊗ VALID 00022 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00004 00002 ∂16-Jan-78 1618 FTP:Levin at SUMEX-AIM phone call to Michael Flynn
C00015 00003 ∂19-Jan-78 2156 FTP:MINSKY at MIT-AI (Marvin Minsky)
C00031 00004 ∂24-Jan-78 1100 JMC*
C00050 00005 ∂27-Jan-78 1943 DCL
C00068 00006 ∂30-Jan-78 1520 BS Research Support, A.Yao
C00226 00007 ∂07-Feb-78 2113 LES Pub problem
C00257 00008 ∂11-Feb-78 1337 FTP:Levinthal at SUMEX-AIM Advanced Automation
C00317 00009 ∂16-Feb-78 1757 FTP:Wiederhold at SUMEX-AIM (Response to message)
C00359 00010 ∂22-Feb-78 1345 FTP:Feigenbaum at SUMEX-AIM is anyone handling this for the department?
C00393 00011 ∂26-Feb-78 1731 JLH
C00430 00012 ∂03-Mar-78 0946 RWW NEWFOL
C00458 00013 ∂09-Mar-78 0958 FTP:PRATT at MIT-AI (Vaughan Pratt)
C00510 00014 ∂15-Mar-78 1609 JBR S1 Meeting
C00538 00015 ∂21-Mar-78 1357 Levinthal at SUMEX-AIM EPRI meeting
C00572 00016 ∂24-Mar-78 1249 DPB Aspvall's candidacy
C00616 00017 ∂03-Apr-78 0910 Levin at SUMEX-AIM tour of duty officially ended!
C00670 00018 ∂12-Apr-78 1105 CCG fellowship
C00706 00019 ∂20-Apr-78 2039 REM Working Interval-Compression program for files
C00740 00020 ∂24-Apr-78 1202 Levin at SUMEX-AIM brief biographical info on referees for manna
C00776 00021 ∂27-Apr-78 1116 WINOGRAD at PARC-MAXC Re: External Employment
C00822 00022 ∂01-May-78 1552 DCL
C00873 ENDMK
C⊗;
∂16-Jan-78 1618 FTP:Levin at SUMEX-AIM phone call to Michael Flynn
Date: 16 JAN 1978 1614-PST
From: Levin at SUMEX-AIM
Subject: phone call to Michael Flynn
To: mccarthy at SAIL
There are two M. Flynns; the CSD M. Flynn's number is: 7-1450...laurie
-------
∂17-Jan-78 0132 LES DIAL line
It has been called to my attention that you not only abandoned your job
this evening, as usual, but you left DIAL in control of TTY11, so that
it couldn't be used by others. I hope that it isn't counting message
units all this time.
∂17-Jan-78 1233 SSO Donahue's seminar
To: DCO
CC: JMC
I can7t reschedule Jim's talk easily, but he will
also be speaking at Xerox on Friday. I asked Jim Horning for the
time, and I will pass it along.
Sue
∂17-Jan-78 1450 DCO Jim Donahue's talk
To: DCL, ZM, RWW, JMC
Jim Donahue is also talking at Xerox on Friday at 10:00 in the
CSL Commons - anyone welcome. (This mainly for those who want to hear
both the AI colloquium thursday and also Donahue.)
∂17-Jan-78 1916 FTP:CARL at MIT-AI (Carl Hewitt)
Date: 17 JAN 1978 2202-EST
From: CARL at MIT-AI (Carl Hewitt)
To: jmc at SU-AI
Jean Sammet has received a garbled telegram from Stoyan asking for an
invitation to the conference. What is the best current address to
try to send another invitation.
Sincerely,
Carl
The following is my only address. I suggest both another letter and
a telegram.
Dr. Herbert Stoyan
DDR 806 Dresden
Togliattistr. 40
East Germany
∂18-Jan-78 1929 FTP:Boyer at SRI-KL (Bob Boyer) Prime factorization
Date: 18 Jan 1978 1704-PST
From: Boyer at SRI-KL (Bob Boyer)
Subject: Prime factorization
To: jmc at SU-AI
In your article "A Basis for a Mathematical Theory of Computation"
you remark upon the distance one can get into elementary number
theory using recursion induction. You suggest that the
the theorem that if a prime p divides ab, then it divides
either a or b is within reach, but the statement of the
unique factorization theorem requires a notation for dealing
with sets of integers.
Recently J Moore and I have got our theorem prover up to
the unique factorization theorem; at least we so regard our
work. What we have proved are two theorems:
(1) if prime.list(l1) and prime.list(l2) and
times.list(l1) = times.list(l2)
then perm(l1,l2),
where prime.list is the function that takes a list
and checks that every element is a prime, times.list
multiplies up the members of a list, and perm computes
whether its first argument is a permutation of
its second argument, and
(2) if x is a positive integer
then times.list(prime.factors(x))=x
and prime.list(prime.factors(x)).
where prime.factors is a recursive function for
explicitly computing a list of the prime factors
of x (or NIL if x is 1).
The hard part of the proof is getting up to the
theorem mentioned above about p dividing ab. The route
the proof takes involves the use of the greatest
common divisor function including the theorem
GCD(ab, ac) = a*GCD(b, c).
"Prime" is defined to check whether any of the
numbers less than the given argument divide the given
argument.
The proofs involve about 30 subsidiary lemmas being
proved (on top of our library of previously proved
theorems about TIMES, QUOTIENT, REMAINDER, LESSP, and PLUS).
The only axioms used were those of (our version of) Peano arithmetic
and our principles of induction and recursive definition.
All the functions involved were proved total.
We are really enthusiastic about getting this far
and would be delighted to come show you or send you
the proofs.
Bob & J
I would much like to be shown the proofs. How about next week some time?
∂18-Jan-78 1935 FTP:Levin at SUMEX-AIM A&P meeting
Date: 18 JAN 1978 1642-PST
From: Levin at SUMEX-AIM
Subject: A&P meeting
To: mccarthy at SAIL
reminder: A&P meeting thur. at 1:00, Polya 204.
-------
∂18-Jan-78 1930 HPM udp
I would miss it, since music tends to monopolize the packs now, and
I often need it for hours at a time going through picture sequences.
With one udp it will be harder. On the other hand, I know we need
the file space.
∂18-Jan-78 2235 FTP:Boyer at SRI-KL (Bob Boyer) Showing our prime factorization proofs
Date: 18 Jan 1978 2236-PST
From: Boyer at SRI-KL (Bob Boyer)
Subject: Showing our prime factorization proofs
To: JMC at SU-AI
J and I could come talk about our proofs anytime next Friday
the $20th. We have visitors earlier in the week. Could you indicate
roughly the number of people who will attend?
-------
This Friday is not feasible for me. A week from now, the 27th would be good,
and Robin Milner will probably be here too. I suppose the audience will
be at least 10.
∂18-Jan-78 2238 FTP:Boyer at SRI-KL (Bob Boyer) A correction
Date: 18 Jan 1978 2239-PST
From: Boyer at SRI-KL (Bob Boyer)
Subject: A correction
To: jmc at SU-AI
Sorry, I meant to say: Friday the 27th instead of the 20th.
-------
27th looks good to me. How about 1pm AI Lab. I would expect about
10 people including probably Robin Milner who will be visiting from
Edinburgh.
∂19-Jan-78 0802 BPM*
Please remember to talk to Prof. McCluskey today about splitting the Computer
Forum finder fee for Tektronix. Prof. Van Cleemput wants it split 3-1 in
his favor, and I think it should be 1-1. Thanks. --Brian
McCluskey didn't come to the meeting today. I'll try him at the meeting
tomorrow.
∂19-Jan-78 1258 REF
I'm not sure that Friday the 27th 1pm is that good a time. For one thing
Zohar seems to have already invited someone for 2pm:
∂16-Jan-78 1457 ZM
Event of FRI 27-Jan-78 at 2:00pm
Special A.I. Lab Colloquium. John Reynolds (Un. of Syracuse) will talk on
"Syntactic Control of Interference"
For another, I'm personally not that enthusiastic about Friday afternoons;
Terry Winograd's seminar is at that time, and about 10-15 lab people have
been attending.
Post script: The schedule for colloquium is on page 5 of WHO[COL,REF].
Local speakers (RWW, DBG, etc.) can usually be "pushed" around.
OK, call Boyer and propose a different time.
∂19-Jan-78 1554 RWW BUG
I BELIEVE IT IS FIXED. It was introduced in Nov by REF when he changed
tauteq for efficiency sake. Your core image is irreparably damaged
but a new one should be ok.
rww
∂19-Jan-78 1906 FTP:Boyer at SRI-KL (Bob Boyer) Confirmation
Date: 19 Jan 1978 1907-PST
From: Boyer at SRI-KL (Bob Boyer)
Subject: Confirmation
To: JMC at SU-AI
Good, then: 1pm at the AI Lab on Friday the 27th.
I hope Robin attends.
-------
∂19-Jan-78 2144 FTP:MINSKY at MIT-AI (Marvin Minsky)
Date: 20 JAN 1978 0045-EST
From: MINSKY at MIT-AI (Marvin Minsky)
To: JMC at SU-AI
Green? Is there a tenure thing?
∂19-Jan-78 2156 FTP:MINSKY at MIT-AI (Marvin Minsky)
Date: 20 JAN 1978 0057-EST
From: MINSKY at MIT-AI (Marvin Minsky)
To: JMC at MIT-AI
yes.
∂20-Jan-78 0030 LES Phone bills for home terminals
In the discussion with Phil Surra this morning, he stated that
after discussions with ARPA, he now agrees that it is reasonable for some
of our staff to have home terminals, provided that we maintain close
property control, but that we should not pay for home telephone service.
I restated the reasons why I think it is a business-related expense. He
said that since there is a convenience factor for the individual, in that
some travel from home to work can be avoided, the individual should share
part of the cost. He further stated that the ARPA office agrees with this
policy and that it is the policy under which they operate. He proposed that
we convert to this policy as soon as possible.
It appears to me that we can do one of at least three things.
1. Acquiesce.
2. Check with someone in the ARPA office to be sure that the story is
straight, since he has done a bit of flim-flamming in the past.
3. Continue to assert that this is a legitimate cost of doing research.
Acquiescing has a small beneficial side effect of avoiding a possible
hassle with the phone company on the issue of these being non-business
lines. In fact, if we make a strong defense of these payments on grounds
of business-relatedness, we may be making a case against ourselves for
Ma Bell.
Incidentally, this change would also apply to your phone lines, so you
might wish to consider changing to a UDS modem with a dial-up line since
that is substantially cheaper than the lines you have now.
My current inclination is choice 1, but if you wish we can follow one
of the other routes.
I favor 2. You need to get acquainted with the new Blue, however.
What fraction must be paid by the individual? If the ARPA bureaucrats
are doing it to themselves, I suppose we'll have to acquiesce.
∂20-Jan-78 0051 LES
The new Blue seems to be Carlson. Of course, such responsibility mappings
are seldom one-to-one. I'll call him tomorrow.
All ordinary telephone charges would have to be paid by the individual,
but Surra agreed that extra charges, such as having to get a foreign
exchange line or extended service in order to reach the lab on a
no-message-unit basis could reasonably be reimbursed.
We are not bound by ARPA internal policies, but we do need to have a good
defense in order to go our own way. In this case, I'm afraid that we are
in a slightly awkward position. We could probably get out of the
awkwardness by changing back to ordinary business lines, but that would be
very expensive.
∂20-Jan-78 0257 FTP:CARL at MIT-AI (Carl Hewitt)
Date: 20 JAN 1978 0558-EST
From: CARL at MIT-AI (Carl Hewitt)
To: jmc at SU-AI
CC: CARL at MIT-AI
Thanks for sending me the hard copy of your paper. I have passed on
copies to Barbara Liskov and Jean Sammet.
Have you had a chance to glance at the MACLISP interpreters yet?
Have you had a chance to send copies of your paper to your colleagues
in the development of LISP?
Cheers,
Carl
I have just written a letter to the people mentioned in lisp.le1[let,jmc],
and I suppose it will go out today or Monday. Ah, if I could only get away
with just glancing at the interpreters.
∂20-Jan-78 0927 FTP:CARL at MIT-AI (Carl Hewitt)
Date: 20 JAN 1978 1229-EST
From: CARL at MIT-AI (Carl Hewitt)
To: jmc at SU-AI
CC: CARL at MIT-AI
Dear John,
I have looked over your mailing list and it looks fairly complete to me.
As for the interpreters, you are the "grandfather" of them all and so must
look after your descendants.
Cheers,
Carl
∂20-Jan-78 0954 FTP:CARL at MIT-AI (Carl Hewitt)
Date: 20 JAN 1978 1256-EST
From: CARL at MIT-AI (Carl Hewitt)
To: jmc at SU-AI
CC: CARL at MIT-AI
John--
If you have any test cases that you would like to try on the
MACLISP evaluators you can send them to me and I will try them out for you.
Sincerely,
Carl
∂20-Jan-78 0956 FTP:Levin at SUMEX-AIM reminder of meeting
Date: 20 JAN 1978 0957-PST
From: Levin at SUMEX-AIM
Subject: reminder of meeting
To: mccarthy at SAIL, buchanan, feigenbaum
SENIOR FACULTY MEETING TODAY at 4:30, Polya 204.
-------
∂20-Jan-78 1031 JC UDP
To: JMC, LES, JBR
CAN WE DISCUSS THE UDP SITUATION AT 1530?
∂20-Jan-78 1109 TOB Arnold
John:
Dave Arnold listed me as principal advisor and was
turned down by Denny Brown. I had contacted Feigenbaum
before about Gennery and was told that on an exception
basis, i could be principal advisor even though I am
not academic council. According to Denny, that is not
settled in Gennery's case. I want to be principal
advisor and would like to do something in this case.
What should I do next?
Tom
I will talk to Denny.
∂20-Jan-78 1249 LES
To: JC
CC: JMC
∂20-Jan-78 1031 JC UDP
To: JMC, LES, JBR
CAN WE DISCUSS THE UDP SITUATION AT 1530?
Yes, as far as I am concerned.
∂21-Jan-78 1537 LES Computer use complaints
To: DCL
CC: RAK, JMC
You and Dick are doing a lot of complaining about what you regard as
abuse of the computer. I do not object to receiving specific complaints
about individuals who appear to be doing something evil and will continue
to respond to such gripes.
On the other hand, I object to continual carping (spelled with a "c", please
note) about general overload and the existance of untouchables on this
machine. If you are trying to make the point that our admittance policy
is incorrect, then you had better take this up in a larger forum. I do
not single-handedly set that policy and I regard complaints in this area
as an intrusion on my work.
In my view, this has passed the annoyance stage and has become upsetting.
Therefore, I plan to ignore further communications of this kind. I am
at least as far behind in my work as you are.
∂21-Jan-78 1607 REF SAIL Talk
To: boyer at SRI-KL
CC: REF, JMC
It appears that there is another talk scheduled at the Lab for
Friday the 27 (at 2:00 pm), so John has asked me to negotiate another time
for your presentation with you.
We are having a regular lab colloquium this quarter, Mondays at
4:00, and would be glad to have you speak on Feb. 6. Please let me know
if you find this time agreeable. If not, we can arrange a different time.
I look forward to hearing from you.
Bob Filman.
∂21-Jan-78 1626 DCL policy and bureaucracy
To: LES
CC: JMC
We are in the process of compiling a list of complaints and constructive
suggestions for remedies. Hopefully this will be the agenda for LES' "wider
forum" discussion.
∂21-Jan-78 1937 FTP:Boyer at SRI-KL (Bob Boyer) Rescheduling of my presentation
Date: 21 Jan 1978 1939-PST
From: Boyer at SRI-KL (Bob Boyer)
Subject: Rescheduling of my presentation
To: REF at SU-AI
cc: JMC at SU-AI
Monday February 6 at 4pm is a convenient time for making my presentation.
I suspect that a colloquim is a more formal forum than I had in mind,
but two weeks notice is sufficient to get a more polished presentation
ready (I hope). I will send you an abstract if you want, and I would
like a crude estimate on the probable attendance,
since I'll probably have a hand-out.
-------
∂23-Jan-78 0902 FTP:Levin at SUMEX-AIM faculty dinner
Date: 23 JAN 1978 0903-PST
From: Levin at SUMEX-AIM
Subject: faculty dinner
To: mccarthy at SAIL
would you like to pre-order? Faculty club expects that night
to be a busy one...laurie
-------
I think not. We h`x6)βC3.sSeβ&yβ∪'≡≠WOMbβO=βO!β←?r;Qβ7∂#S↔IεkW∂ hS'→βO!βSπ↑+Mβ¬π;#'3*βS=β>+QβSF)β≠?}!84({⊃M6+∞q5]aα↓E]AJ∞⊗Qα4*S{Q↓↓α≤~≥1αTj
1αTj 1α]*51α6+'∨↔v∪πW5ε QαN,j⊗a6 J5↓↓α4*C⊗{≠↔O≡{KMα6+'∨↔v∪πW5bα∨K↔.q1α7≤≠πKSGI1α ∂∪S!1∧[W7π∪P4*Cf+πO∃π≠↔;⊃πK?WIε∪??-ε{K∪↔↔→βS=εk∃βπ~βO??rβπMβε{OO'⊗c∃9↓¬##↔eπ;↔K∀hS∪W∃ε QβSF)β ?}[OS?⊗)β?9∧3K'∪∂I1α+∞q9↓Iαq↓αSF)βO?}s↔Iβ>)β?K&+I1β&C∃β /#S↔HhS?WIε≠#π;≡)β←'faβ ∃ε{→β#∂3';≥ε∪??/~βπQβ&C∃α }{/OS␈∪∃βSF)β≠'↔≠Qβ∪∂Iβ?→ε≠3πO~p4*SF ;/Mbα∂πK}cg9α& +;πHh(4({⊃M6+∞q5]aα↓EeEHJ*6
R4+∨/!αQ7≡C'KQε{K∪↔∩β≠?Ko→1β↔∨β↔∂'∞c3eβ6{Iα#/∪O∂#(h(4({⊃Q6+∞q5]aα↓A]AJ*6
R4+≠Ns'O!εCπW∨.`4(4PyIQ6V 95]B↓↓Aa λ&~RβR#πK"βπQα≥∩%6.b↓"C↔&+Iα#∂∪Q%M#↔∂#v{3?∨Jβ≠?K.≠πOQα↓4*& S∃iβ⊃Qα+∞q↓Ee;A↓Aa ⊃6BN h*≠K}iiα#∂∪Qβπ"αNJ%lZ1↓"ε+S↔I∧CπKQHh*OW⊗S↔∂QRαS↔∂Fs?3??Iβ≠?⊗+∂πO h*S=R↓↓α*l→βπQ¬~ε&0hS∂
iα↓α#π↔ 4(4TS?#9`h(&π~βg?Uε[;?]bα%∨[*β ↔↔rβ←?K↑K;≥βf S↔3Jβ?9β
βCπC/⊃βCW↔β?KSNs≥βSxh+'∪.sS'≠Jαε%β␈βC?K'+;'SN+Mβ?6+IβSF)β;↔G!↓IAπK↔πK~q↓αC∂∪Qβ?2βS#∃ε≠#πK&+H4+O→βS=π≠πeβ∂!β3↔∂≠Qβ¬ε3↔]β>{K∪Mε ?W"αO?[N+QβC␈≠O' Nc'S'/→9↓αJ;5βO'+∂,4V{9βSF Qβ?v)55β≡ 9βg␈)β#↔g↓⎇↓αJ;⊃β3N[∃βSzβ∨↔Qπ#?∨↔&C↔Iβ>KS!βN{Uβ≠␈⊂4+¬ε≠?WCf)β?→εC?WK~βS=β& 3-β∞∪?WQπ##∃β∨+ +↔∨!84(hRC↔S/⊂4)5ji555hh*>-bβ#?]ε ?W"βS?7␈∪K?]εcW;∂BβπQα≥∩%9↓∧Iβ#π6+9∨Qπ3'O'&+⊃αN∀Iβ'9π W'S(h+O?n)βS'n)84({⊃Q6+∞q5]aα↓AaM(J~RAT3↔'∨.s πWjβπQα≥*6⊗al
&5NK;π∪6+KSπw!β∪?.∪3∃β≡≠#↔∪.c';≥ε3?I↓
yME=;@4*∪∂#∃i↓∪!α*εr↓Ee]B↓AaM2jBNPhR≠K?kQα≠↔N;↔; ∂+5βπ"αNV6-A6ε&hh*OW⊗S↔∂QRβ';π'3↔KS∞sQβ∪␈+ 3∃π≠∂#↔'+3';:β≠?Iβ =ME{9`4*&yi↓↓εS7
β∂!αNεLa1β∂/!βπQ¬~ε&0hP4)#n+OOπ>)βS=∧≠πK?gK9βSzβ∪↔3O3↔Iβ&yα ?∩α≠3?N!1βONs∂∃βF)β#π⊗#3eβ/3↔H4Vc??/~β'9βFKMβ↔f+∂SK}s'
βn '1β⊗{a$4Ph*%α<z>~⊗" α%β∞cK↔π'Iβ#π"β¬β∂}k7'Sn+;Qβ&yβSπ↑)βS#⊗+∃β[O≠'S'v84+O≡K↔;SO≠SMβ␈+QβSzβ∪';v+Iβ?rαSW↔≡#πeβvK∨#Qβ =ME{9aβ←F+9α%ε ∨K↔.!βS<hSS#∃εk↔↔SNs≥βSNk∃β≠␈⊃βS#*αO↔;N{Iα≠∞≠W3SJβ7↔↔&K;≥9∧≠?W3"β←∃βn{[∃βO 4+Wαβ'9β&C∃β∪∂I57O∂I↓EIS→A55∪QEU9¬##πQπ;?W3"βOπSO≠≠eβ⊗{S!αTj
≡Mε ;⊂4U#πK+∞q∨Mβ≡cπOMε≠?;O'∪π';'→9α?∩β7?[*β'Qβ⊗ ∂-β&yα7?v#πeβvK∨#QαC WPhRSπKV 9β∪}+Mβ;␈!β3'↑)βS=ε≠?7∃εK9β≠⊗{5α3␈→α∨π&{M%8hP4*πBβ←↔3baα∂π⊗{3g9bβO↔∃π;#πQπK?Uβ≡ 9β∪zλ4(4T+⊂4)ji555jh4(4PyIQ6V 95]B↓↓EA ⊂&~RβR#πK"βπQα≥∩%6.b↓"C↔&+Iα#∂∪Q%NcW;∂B↓4*& S∃iβ⊃Qα+∞q↓Ee;A↓EA ⊃6BN h*≠K}iiα#∂∪Qβπ"αNJ%lZ1↓"ε+S↔I∧CπKQHh*OW⊗S↔∂QRβ3W;≡@4*S{Q↓↓βVk
βπ"αNU6 H4(4T;??⊃rq9β#␈9βπ ␈+Qβ∂}k';≥εC↔K∃ε K?Wv!↓EES!U84Ri555ji44*≡+∃βg␈)βS#.q84( ∂24-Jan-78 1100 JMC*
honig.le2 and specu
∂24-Jan-78 1121 FTP:Levin at SUMEX-AIM meeting with Prof. Lieberman of Provost office
Date: 24 JAN 1978 1121-PST
From: Levin at SUMEX-AIM
Subject: meeting with Prof. Lieberman of Provost office
To: mccarthy at SAIL
He would like to meet with you, Osserman (Math), Veinott (OR)
and Effrin (Statistics) to discuss: Absences from campus
while on grants. What is the best half hour for you next
monday, 30th between 2 - 3:30?
John Linville called...return call: 7-2930.
...laurie
-------
∂24-Jan-78 1436 LES
∂24-Jan-78 1426 FTP:RUSSELL at USC-ISI Demonstration on 25 Jan
Date: 24 JAN 1978 1406-PST
Sender: RUSSELL at USC-ISI
Subject: Demonstration on 25 Jan
From: RUSSELL at USC-ISI
To: Les at SU-AI
Cc: Russell
Message-ID: <[USC-ISI]24-JAN-78 14:06:53.RUSSELL>
Les,
I am planning a demonstration of ARPANET capabilites
tomorrow morning to a group from the President's Task Force
on the reorganization of the government. One of the capabilities
I would like to show is the SU-AI News system. Will
the SU-AI computer be up tomorrow morning between 0830 and 1130 EST
for a small demonstration of the news system?
Will appreciate your help.
dave
-------
∂24-Jan-78 1433 LES Demonstration on 25 Jan
To: Russell at USC-ISI
Dave,
Yes, we expect to be up tomorrow morning. If there is anything we can
do to help, such as keeping the load down, please let me know.
Les
∂24-Jan-78 1454 CJS phone message
John Linvill of Electrical Engineering would like you to call him at 497-2930
∂24-Jan-78 1630 FTP:Taynai at SUMEX-AIM Senior Faculty Meeting
Date: 24 JAN 1978 1630-PST
From: Taynai at SUMEX-AIM
Subject: Senior Faculty Meeting
To: EJM at SAIL, JMC at SAIL
cc: Buchanan, Feigenbaum, Taynai, TAR at SAIL, RWF at SAIL,
cc: JGH at SAIL
Because of conflict of schedules, the senior faculty meeting will be
held
Tuesday January 31 9-11 a.m. Polya 204
Carolyn Tajnai
-------
∂24-Jan-78 1908 FTP:Levin at SUMEX-AIM forest baskett
Date: 24 JAN 1978 1727-PST
From: Levin at SUMEX-AIM
Subject: forest baskett
To: mccarthy at SAIL
Names suggested for comparison with forest:
Brinch-Hansen
Batter Lampson
Peter Wiener
Bernice Galler
Sam Fuller
Peter Denning
Corbato, Saltzer
-------
I apologize for my handwriting, but two names (which
may be referred to in the future) were
Butler Lampson, a computer scientist at Xerox PARC, and
Bernard (Bernie) Galler, at the University of Michigan.
∂25-Jan-78 0311 REM Single-character crunch on tty-output from pdp-10 to home-terminal
To: PMF, TVR, JMC, LCS, WD
Using a single-character near-Huffman code for output, it only takes 4.64 bits
per character. Thus a 100-->57.9 compression if we assume 7-bit input
characters and IMAGE-mode output which I am using, thus getting effectively
about 50 characters per second on 300-baud dialup line. Errors are very
few thus can be ignored or corrected easily in such an interactive use.
(Note, part of the compression involves a pseudo-Baudot conversion, not
present in my 1971 and 1976 data-compression work at SU-AI.) Next, left
context like in my data-compression work, to get even better compression...
∂25-Jan-78 1129 FTP:Feigenbaum at SUMEX-AIM Call from John Linvill re Baskett case
Date: 25 JAN 1978 1130-PST
From: Feigenbaum at SUMEX-AIM
Subject: Call from John Linvill re Baskett case
To: jmc at SAIL
John Linvill just called and asked if the promotion committee work re Baskett
could be "joint" with his department in the sense that we invited Mike Flynn to
work with our A&P Committee on the consideration of tenure for Baskett.
I think this is a good idea, for political reasons of promoting long-term good
relations and cooperative spirit with EE--and Mike will certainly be helpful
in theevaluation. Linvill expressed a STRONG interest in Baskett's getting
tenure at Stanford.
Ed
-------
∂25-Jan-78 1148 FTP:Feigenbaum at SUMEX-AIM some straws in the wind re building
Date: 25 JAN 1978 1149-PST
From: Feigenbaum at SUMEX-AIM
Subject: some straws in the wind re building
To: dek at SAIL, rwf at SAIL, jmc at SAIL, les at SAIL,
To: bs at SAIL
Mail from SU-AI rcvd at 25-JAN-78 0950-PST
Date: 25 Jan 1978 0948-PST
From: JJK at SU-AI (Jonathan King)
To: feigenbaum at SUMEX-AIM
Ed -- At a meeting to discuss building matters, Bill Bussey let slip the
fact that the provost was considering using M. Jacks Hall space for the
CS Library rather than expanding the Math Library. I pursued this with
Roger Buckhout, who told me that costs for Math Library expansion, principally
mechanical alterations, "shocked" the provost, who then asked for cost
estimates of using 2000 sq ft of Jacks Hall for the library. CSD had not
been informed because no decision had been reached. Roger suggested that
since we now know of this possibility, we might wish to lobby against it;
he claims he is doing so. I don't know when a decision will be made.
-------
--------------------------------------------------------------------------
the enclosed message from King gives important news. Do we want to encourage
or discourage the plcement of our library in our own building? We lose lots of
square footage, but we get our library back! Would you please give me your
opinion? Better yet, give it to McCarthy so he can take necessary action!
Actually, the whole faculty should discuss it but I wanted to take
preliminary soundings.
Ed
---------------------------------------------------------------------------------
-------
∂25-Jan-78 1442 REM via AMET Improved PDP-10 --> 6502 code results
To: PMF, REM, JMC
The code table is located in HOTER.FAI[1,REM] at the label CRUCOD:
It is a single-character no-context almost-Huffman-code with
pseudo-Baudot coding. Here are results for the new code installed
last night:
RUN1 RUN2
CHARS 13683 33396
TOKENS 13836 34005
<FLUSH> 1137 3642 (Generated when line goes idle to flush out
bits from last few characters.)
BITS 74779 192721 (Includes <FLUSH>s)
BITS 63409 156301 (Discounting the ten bits for each <FLUSH>)
BIT/CHR 4.63 4.68
C.R. .579 .585
SPEED 51.8 51.3 (Chars/sec at 300 baud)
(all ratios ignore the <FLUSH> tokens, which only occur when both
incoming and outgoing lines go idle, hence wouldn't appear at all
if output was using the phone line to capacity - hence ratios apply
to any output-limited task, which is what the compression is all about)
∂25-Jan-78 2059 DEK library
To: FEIGENBAUM at SUMEX-AIM, JMC
Putting CS library into Jacks Hall would be such a
horrible decision, I can't imagine our enlightened administration ever
allowing it to happen. talk about saving a few dollars one year
to be faced with many many more later!
being on sabbaatical, I can't write down my hundred reasons why the math
and cs library should be combined (i have studied this carefully for
several years, esp as head of the lib. committee), but i can say that
i can't think of any good reason to give up our precious space for something
that will cause us to have a split-up librrary anyway and will cost the
library a lot more besides.
the math library has to be expanded anyway, even if just for math alone,
so any change in the plan to merge both is a loss for everybody.
if there was any reason i would consider leaving stanford, it would be
(a) if i had to be dept chairman, or (b) if our cs library or space for
visitors were blundered away!
I agree with Don that it would be bad to lose the space in Jacks Hall,
and it is important to merge the CS Library into the Math Library. I
will send a memo to the Administration to that effect.
∂26-Jan-78 1046 RWW fol
I know you won't believe this but caroline and myself tested that
fol just minutes before it failed for you so as not to be
embarrassed and it worked perfectly. I don't now know what the problem
is.
rww
∂26-Jan-78 1309 FTP:Feigenbaum at SUMEX-AIM (Response to message)
Date: 26 JAN 1978 0917-PST
From: Feigenbaum at SUMEX-AIM
Subject: (Response to message)
To: JMC at SU-AI
cc: DEK at SU-AI
In response to your message sent 25 Jan 1978 2215-PST
Thanks, Don, for the timely and strong comment re library space. Can
John quote some of your comments in his strrongly worded response to
Miller? It would hhelp, I think.
Ed
-------
∂26-Jan-78 1310 RWW fol
It seems that there was a confusion in the way IL treats its upper
when more then one version of FOL exists. This has all been fixed and
I believe everything is back to normal!!
rww
∂26-Jan-78 1401 DEK library
To: feigenbaum at SUMEX-AIM
CC: JMC
sure, what I said can be quoted ... I lay awake last night thinking
of some even stronger things, which maybe shouldn't be quoted!
One point seems obvious, that the additional cost of staffing another
library will surely make any short-term building cost savings inoperative.
Coupling that with the fact that CSD library won't really fit in
2000 square feet, considering present growth rates, and needless
duplication of equipment as well as staff, etc. etc. makes the whole
idea a losing proposition. Surely Stanford won't be so shortsighted.
State colleges and even some universities don't mind economizing on
quality in the name of "efficiency". But I didn't think Stanford
was like that, I thought we wanted to have a high quality library.
What is shocking about that?
∂26-Jan-78 1409 BS CSLibrary in M.Jacks Hall
CC: Feigenbaum at SUMEX-AIM, LES, RWF, JJK, JMC
My comments probably are not constructive, but in my several surveys of
the M.Jacks plans (including placing people in offices) I found that there is
not room for a library inM.Jacks Hall if we plan to house CS faculty and staff,
DSL faculty and 100 graduate students. If a library is included, then we
lose some office and seminar/conference room space. Also, it is not clear
to me why there would be sufficient money to build a library in M.Jacks Hall
but not enough to remodel the Math Library. This sounds like funny politics
to me.
Betty
∂26-Jan-78 1522 DCL Dynamic Logic
To: DCO, CGN, WLS, WP, FWH, RAK, SMG, RLD, JMC, RWW, ZM, ND
I am trying to organize a time for David Harel to give an informal
seminar on DYNAMIC LOGIC. It will have to be on Tuesday sometime
between 11.00 am and 2.00pm. If you are interested and have a time
preference, please send msg. to DCL.
I am interested in Harel seminar, because he has been recommended for
a position in the department of which I am acting chairman. I teach
from 11:00 to 12:15 on Tuesday. Apart from that I will come.
∂26-Jan-78 1603 PAM
JOHN .. HERE ARE WAYS TO GET IN TOUCH WITH SANDRA ...Paul
Blakeslee,Sandy 65E Esc Vill ( 7-1000 or 7-2558) 323-5308
∂26-Jan-78 2320 SMG success with prover
To: GROUP.DIS[VCG,DCL]:
The simplifier has just done a fairly difficult problem involving
DEFRANGE. This required forwards and backwards rules and
involved a lot of arithmetic inequalities. This is the most
complicated example from runtime error checking that has yet
worked with the new simplifier.
The example probably would have worked several weeks ago, but one
of the rules was too restricitve.
s
∂27-Jan-78 0944 FTP:CARL at MIT-AI (Carl Hewitt)
Date: 27 JAN 1978 1245-EST
From: CARL at MIT-AI (Carl Hewitt)
To: Jmc at SU-AI
CC: CARL at MIT-AI
Dear John--
I will be in Palo Alto next Tuesday and Wed.
Would you like to have dinner on one of these evenings?
Sincerely,
Carl
Sure, let me consult Vera, and I'll get back to you.
∂27-Jan-78 0949 FTP:Levin at SUMEX-AIM your polya office
Date: 27 JAN 1978 0951-PST
From: Levin at SUMEX-AIM
Subject: your polya office
To: jmc at SAIL
cc: levin
recruiters from IBM yorktown will be here next week. Is it
O.K. if they use your office mon. morning?
...laurie
-------
Yes, it's ok.
∂27-Jan-78 1943 DCL
To: GROUP.DIS[VCG,DCL]:
***********************************************************************
SPECIAL SEMINAR ON PROGRAM VERIFICATION
TUESDAY 31st. JANUARY
PLACE: ERL 237
TIME: 11.00 am
TITLE: AN OVERVIEW OF DYNAMIC LOGIC
****************************
SPEAKER: DAVID HAREL
M.I.T.
This is an informal introduction to Dynamic Logic. It will be of interest
to those concerned with logical systems for specifying programs and
proving correctness. David Harel will talk for about 45 mins. and then
answer questions.
***************************************************************************
∂27-Jan-78 2019 RCM MRC
To: LES, JMC
I have had just about all I can take of MRC. Tonight he refused to help
when the KA10 refused to restart, although he was the only systems person
at the lab, because of some dispute he apparently has with LES, JBR, ME, etc.
I realize that competent systems programmers are hard to find, but MRC seems
to be making a negative contribution.
∂27-Jan-78 2023 RCM Sato's thesis
Today I received from Sato a copy of his thesis. If you still cannot find
your copy, I can lend you mine to Xerox.
∂27-Jan-78 2319 MRC
To: LES, JMC
∂27-Jan-78 0705 Ryland at RUTGERS-10 Dialnet
Mark: This is CHris Ryland at Columbia University. I've been
following the Dialnet development with a great deal of interest,
and am tentatively offering to help. Would you be interested in
having me work with you on the final protocol definitions, and
on an implementation on, say a DEC-20 and a UNIX system? We
have both here, and I'd really like to see Dialnet running on
them, since we have no other way to connect 'em, except a
hackified single-wire link. The reason it's tentative, is that
I might be able to work on it for credit; if I can't, it might be
too much work to tackle right now, though I certainly would at
least plan to work on the uNIX or DEC-20 implementation this summer.
Let me know what you think....Thanks,
--Chris
mrc - Lets's take him up on this, at least for Unix. I still want
to do the Twenex implementation.
That's fine, but can Rutgers-10 equip itself with a pair of compatible
modems?
∂27-Jan-78 2358 MRC
To: JMC, LES
∂27-Jan-78 2352 JMC
That's fine, but can Rutgers-10 equip itself with a pair of compatible
modems?
mrc - Oh, he's not Rutgers; he's Unix and Twenex at Columbia. I think
he's more interested in Dialnet between the two of them more than to us.
I presume the idea is that we'd exchange sources and all that. I think
that's fully in keeping with the idea of a communication protocol. I
will encourage him to have hardware to talk to us, but I think at this
stage of development it is more important to get people talking Dialnet
than worrying about setting up a network.
My personal suspicion is that there are many places interested in
Dialnet for solely in-house purposes and have are not interested in
either talking to anybody or having anybody talk to them. I consider
it essential to Dialnet propagation that we reach these people too.
Same question w/r Columbia?
∂27-Jan-78 2359 MRC what happened with RCM tonight
To: LES
CC: JMC, RCM
Sigh. I guess you are getting tired of being asked to arbitrate this
sort of thing. Let me try to explain this one.
When RCM asked me to look at the KA, I informed him that I was NOT a
"systems person" and was expressly forbidden to do systems work by
the god squad. This includes fixing things when they are broken, other
than the normal things that users are allowed to do. There is nothing
I may do other than just restart the KA in the normal way, which RCM
had already tried. If RCM expected me to do something wizardish, he
was expecting me to do something I am not allowed to do.
I'm not overreacting; look at all the "Goddamn it, MRC, keep your hands
off ..." comments in the log and which I've gotten in my mail and so on.
I simply decided that since I've been denied initiative to do systems
work, I won't take it upon myself.
Now, my question is, why wasn't there a member of the god squad covering
the system at the time? Why did one user (RCM) feel he had to harass
another user (MRC) to fix the system because there weren't any wizards
around? And, why is MRC being blamed for doing as he's been told?
∂28-Jan-78 0004 MRC
∂28-Jan-78 0003 JMC
Same question w/r Columbia?
mrc - I honestly don't know. I think he's interested only in in-house
applications for Dialnet. I don't see how it affects his offer to help
either way though. It would be nice to have a coast-to-coast link at
this stage of development though.
∂28-Jan-78 0020 LES MRC
To: RCM
CC: JMC
Indeed, MRC is not a system programmer, though he has had some aspirations
in that direction. You should expect no more (or less) help from him than
any other random user.
∂28-Jan-78 0041 LES Misunderstanding
To: MRC
CC: JMC
I have explained the situation to RCM. He has no basis for making demands
on you.
∂28-Jan-78 0031 MRC Dialnet
To: Ryland at RUTGERS-10
CC: LES, JMC
Chris,
Thanks for your offer. I talked with JMC and he agreed it would
be a good idea. I have a personal interest in my writing the version
for Twenex, but any assistance you're willing to offer for that and
especially for the Unix version and with the protocols would be quite
welcome.
Are you intending to buy modems compatable with ours, or is your
interest in Dialnet strictly in house? This will, of course, greatly
affect the nature of the cooperation we can have, although it shouldn't
affect software work at all.
I know some people who would be interested in a Unix implementation
of Dialnet. That's really great. We'll take over the world yet.
If you can, catch me on-line sometime so we can talk. I don't know
how to get in touch with you via the phone (there's some magic number
you have to dial to go long-distance which I haven't been given yet or
something like that), but if you can call, I'm at (415)497-1407.
∂28-Jan-78 0516 MRC clarification about December
I realized that I may have unduly alarmed you by what I said over
the phone about the possibility of my leaving. So I hope this will
clarify how I feel and hopefully will reassure you.
I've been a system hacker (elsewhere) for a while now and I know
it's my "thing". Just designing protocols (without programming,
waiting for hardware) and some illicit programming on the side (such
as TELNET) is alright for a while, but eventually I stagnate and lose
my enjoyment in what I'm doing; which in turn hurts both the quality
of my work and my relations with the people I work with.
I need an environment where I am turned loose on everything and
have a free hand to work on whatever happens to hold my interest at
the time. It doesn't appear that SAIL can provide me with that.
Recently more restrictions appeared; now I can't even read (much less
edit) some system programs which I wrote!
However, please be assured from me personally that I will not
vanish; I couldn't live with myself if I did. I haven't even decided
whether or not to leave to start off, and if I did, it would be at a
point where somebody else could take over with epsilon difficulty; and
I would do my best to ensure a smooth changeover. December is when
the grant runs out and the move to the new building happens.
December is still quite a way off, and a lot may happen before
then. I just feel you have a right to know what's up. The "problem"
is mostly with me; I've been told enough times that I'm high-strung
and overly-sensitive. But I've lived with it have done alright; I
have (almost) everything I've ever wanted out of life right now and
not many people my age can say that.
∂28-Jan-78 0838 JSL via SUMX
:JOHN you may want to see ieee.txt[1,jsl] or move it to a more accessible area.
∂28-Jan-78 0922 FTP:Brian K. Reid <BR10 at CMU-10A> Re: letter to justice department
Date: Saturday, 28 Jan 1978 1220-EST
From: Brian K. Reid <BR10 at CMU-10A>
Subject: Re: letter to justice department
To: JMC at SU-AI (John McCarthy)
CC: Minsky at MIT-AI
In-Reply-To: John McCarthy's message of 27 Jan 78 21:06
Good. I trust that you both saw the article in the Christian Science
Monitor last week?
∂28-Jan-78 1314 BH
∂28-Jan-78 0049 JMC mail
It would be nice if the message telling a user that he has mail would
distinguish whether he is a primary recipient or is merely getting a copy.
BH: No sooner said than done! ";; Mail for JMC (CC) from ..." is the format.
∂28-Jan-78 1645 FTP:MINSKY at MIT-AI (Marvin Minsky)
Date: 28 JAN 1978 1945-EST
From: MINSKY at MIT-AI (Marvin Minsky)
To: MINSKY at MIT-AI, jmc at SU-AI, reid at CMU-10A
No, I didn't see CSN article. Don't know how letter mismailed, by temp
secy has bugs...
It's just another fairly reasonable article, except for the fact that
it assumes that we went after Quasar before we knew they had faked
their store demos. In fact we intended to investigate before going
public.
∂28-Jan-78 2033 RWW dinner
In case you've forgotten Monday night at 7:00.
rww
$
∂29-Jan-78 1333 REF
Fine. That's the harder part. The placement of the other blocks, however
is non trivial. It's also easier to figure out if you have the blocks to play
with. I'll try to remember to bring them in tomorrow.
∂30-Jan-78 1024 FTP:RYLAND at RUTGERS-10 Dialnet Help
Date: 30 Jan 1978 (Monday) 1321-EST
From: RYLAND at RUTGERS-10
Subject: Dialnet Help
To: MRC at SU-AI
cc: les at SU-AI, jmc at SU-AI
Mark:
Yes, our interest includes getting the appropriate modems, as well
as in-house (hard wire) use. I would be interestd in getting going
on protocol help as soon as possible, and would be willing to
help on either the Twenex or UNIX implementations. Have you
thought at all how to get the standard -20 Frontend -11 to do
dialing? Could you send me appropriate hard-copy versions of
your protocol documents, via US mail? We're not on the net, so
it's hard for me to FTP things (have to go via tape from Rutgers).
A good address is: Chris Ryland, watson Labs, Columbia U.,
612 W. 115th St., New York, NY 10025. Cheers!
--Chris
Les, make sure Ryland knows what we have ordered from Vadic.
∂30-Jan-78 1223 LES Blakeslee meeting
To: JMC, PAM
Reminder: Sandy is supposed to show up at 4pm today.
∂30-Jan-78 1223 RPG Maclisp
To: LES, JMC
The results of the LISP survey are in, and I have a
suggestion based on those results. The suggestion is that Maclisp
be given the status of the official Lisp at the lab. In particular,
I would like to update the documentation so that people looking
for "LISP" get routed to Maclisp more directly. This involves
a few changes to existing files plus changing the file that
is opened with "READ LISP" from LISP.WD to the Maclsp Manual.
Finally I would propose that the Monitor command "LISP"
runs Maclsp instead of L16.
According to the survey, 11 people use Maclisp, 3 use L16,
and 3 use Ucilisp.
-rpg-
∂30-Jan-78 1520 BS Research Support, A.Yao
CC: JMC, DPB
Dr. Richard Lau, ONR-PASADENA, called me today to let me know that the
Yaos'research proposal to ONR has been turned down. This means that
Andy has no research support for the remainder of the year--as far as I
know--and that his salary will be 100% from theDepartment budget. I thought
this might make a difference in teaching assignments. Frances' salary is not
affected.
Betty
∂30-Jan-78 2129 REF BLOCKS PUZZLE
The puzzle seems first attributable to John Horton Conway, or
so claims Martin Gardner in the Feb 1976 Sci. Amer.
∂31-Jan-78 0649 FTP:RYLAND at RUTGERS-10 DIALnet
Date: 31 Jan 1978 (Tuesday) 0947-EST
From: RYLAND at RUTGERS-10
Subject: DIALnet
To: mrc at SU-AI
cc: jmc at SU-AI, les at SU-AI
Mark: I've arranged to get credit for this work, and thus expect
to bite off a piece that I can finish i three months; however,
since this network promises to be quite valuable for us, I also
expect to keep on working on it at least until the end of the year.
Can you suggest a workable chunk of the project? For example,
the mail protocol, or design of the Twenex NCP, or whatever
you call it? I've had a good deal of ARPA experience, and thus
would feel comfortable with most of the different protocols.
To tell you the truth, I'd rather work on a Twenex imple-
mentation rather than a UNIX, since that's the machine I work
on mostly, but I would be quite interested in both. Any ideas?
Thanks again,
--Chris
∂31-Jan-78 0837 BS Audit
I have a DCAA auditor coming this afternoon to go over salary allocations
for "some faculty and staff." I don't know the names of the individuals
to be audited. I don't anticipate any problems, but will let you know
if any arise. I wonder whether this is the audit which Jerry Lieberman
mentioned to you?
Betty
That would be it, I think.
∂31-Jan-78 1505 100 : cr Phone Message
Please call Bill Musgrave, Datamation Magazine, at (213)559-5111.
Please send Bill Musgrave at Datamation a copy of the Stanford
Press release on the Quasar robot. There is a copy in my Quasar
file you can xerox.
∂01-Feb-78 1103 PAT Computer Forum
I've put the times and places of the various meals in your calendar. I checked
about spouses for the dinner and, regretfully, they are not included.
∂01-Feb-78 2037 BH MAIL
To: MRC
CC: JMC, LES, JBR, ME
I have already told you once to keep your hands off MAIL.
I am now telling you again, for the last time. What have you
done to it?
∂01-Feb-78 2324 MRC
To: BH, JMC, LES, JBR, ME
∂01-Feb-78 2037 BH MAIL
To: MRC
CC: JMC, LES, JBR, ME
I have already told you once to keep your hands off MAIL.
I am now telling you again, for the last time. What have you
done to it?
mrc - GODDAMN IT, BH; YOU WERE ONE OF THE PEOPLE WHO TOLD ME ABOUT
THE MICROCODE CHANGE WHICH FLUSHED EXTEND. I FIXED NETWRK TO NOT
USE EXTEND AND JBR FLUSHED IT. I ALSO TOLD YOU ABOUT IT TO RECOMPILE
IT. YOU DID ***********NOT************** RECOMPILE MAIL. THIS CAUSED
MAIL TO GET ILLEGAL INSTRUCTIONS.
NEXT, YOUR SLOPPY PROGRAMMING IN MAIL PREVENTED THE NEW ROUTINES FROM
WORKING. IN PARTICULAR, YOU PASSED STRINGS TO THE ROUTINE WITH A HIGH
PROBABILITY OF LOWERORDER BYTES BEING NON-NULL. THE STANDARD KA-STYLE
ALGORITHM DOES MASKED WORD COMPARES WHICH WOULD NOT WORK WITH YOUR CALL.
BECAUSE OF THIS, I WAS FORCED TO SPEND HALF AN HOUR OF DESPERATELY FAST
HACKING TO NETWRK FIRST TO FIND OUT WHY IT WAS LOSING THEN TO FIX IT
IN NETWRK.
I DID **********NOT********** ALTER YOUR MAIL SOURCE IN ANY WAY AT ALL.
NETWRK NOW ASSUMES THE CALLER CAN DO SLOPPY THINGS LIKE MAIL DOES, SINCE
I DECIDED IT SHOULD.
I AM FED UP WITH THIS SORT OF CRAP, THIS "WHAT DID YOU DO NOW" BUSINESS.
I MERELY REPAIRED SOMETHING WHICH WAS LOSING, EVEN THOUGH IT WAS YOUR
SLOPPINESS, BECAUSE I DECIDED THAT MY ROUTINES SHOULD BE CHANGED TO BE
FLEXIBLE ENOUGH.
HAD I NOT ACTED MAIL TO NETWORK SITES WOULD STILL BE LOSING NOW, AND
YOU WOULD HAVE TO COME HERE AND FIX IT.
∂02-Feb-78 2251 BH apology
To: MRC
CC: JMC, LES, JBR, ME
The reason I yelled at you is that I logged in and saw a system message
about MAIL being broken with no other information about what was going
on, and thought you had done God knows what without telling me. But it
turns out I was done in by my own new feature--I somehow acquired a
MSG.MSG file without knowing it and therefore didn't get any mail
for two days until I (just now) figured it out. I'm sorry.
∂03-Feb-78 2135 WD
To: JMC, WD
Sure! I think I'll be down there tomorrow.
Thanks. The slides and a note will be on the planter at the top of the
stairs if the convenient time for you to drop by is when we aren't home.
∂05-Feb-78 1520 LES Demonstrations on Thursday & Friday
To: TED, JBR, ME
CC: JMC
The demonstrations described in the attached message are likely to use NS
here, so lets try to be up.
∂05-Feb-78 0505 FTP:RUSSELL at USC-ISI Demonstration Schedule for 9, 10 Feb 78
Date: 5 FEB 1978 0455-PST
Sender: RUSSELL at USC-ISI
Subject: Demonstration Schedule for 9, 10 Feb 78
From: RUSSELL at USC-ISI
To: Mckinley, Sacerdoti at SRI-KL, Heart at BBN-TENEX, Nickerson at
BBN-TENEX, Tom at CCA, AV at MIT-DMS, Santos at BBN, Forsdick at BBN, Les
at SU-AI Cc: Uncapher at ISIB, Ellis at ISIB, Kahn, Carlson, Cerf, Rothnie
at CCA, Help at CCA, russell, Fhollister Message-ID: <[USC-ISI] 5-FEB-78
04:55:51.RUSSELL>
Two demonstrations of the ARPANET and the packet radio net are
scheduled this week, both to be held at SRI in Menlo Park.
A group of senior Army officials are scheduled from 0800-1300
PST on Thursday, 9 Feb 78.
The new Director of ARPA is scheduled for this briefing from
0800-1400 PST on Friday, 10 Feb 78.
Will appreciate your efforts to keep the net and hosts involved
as stable as possible during this period.
Thanks,
Dave Russell
-------
∂05-Feb-78 1751 MRC Dialnet documents
To: Ryland at RUTGERS-10
CC: LES, JMC
I have prepared a care package of the protocol documents as they exist
today, along with a price quotation we got for the equipment we have
ordered for our system. It will be in the mail tommorrow.
Mark
∂06-Feb-78 1036 TOB reactor
John
What is the name of that reactor which was shut down
because of a legging dropped into it?
How long was it down?
Tom
It was one of TVA's reactors at Brown's Ferry. It was shut down
for reloading or something like that when the accident happened.
I vaguely remember that it was down for 17 additional days, and
remember more precisely that the cost of the extra shutdown
was $2,800,000 for coal as an alternate source.
∂06-Feb-78 1121 FTP:Hart at SRI-KL (Peter Hart) AI Forecast for DoD
Date: 6 Feb 1978 1119-PST
From: Hart at SRI-KL (Peter Hart)
Subject: AI Forecast for DoD
To: Duda, Nilsson, Sacerdoti, Tenenbaum,
To: Feigenbaum at SUMEX-AIM, JMC at SU-AI, Newell at CMU-10B,
To: Reddy at CMU-10B, Minsky at MIT-AI, PHW at MIT-AI,
To: Amarel at RUTGERS-10, Uncapher at USC-ISIB, Winograd at SAIL,
To: Rick at RAND-UNIX, Denicoff at USC-ISI
The next msg from me is an undebugged first draft of the AI "forecast"
that I have been working on. Comments at all levels-- bibliographic updates,
tone, content, etc.-- are invited.
(The paper is about 25 pages, single-spaced; you'll probably want to list
it off-line.)
Comments arriving later than the third week of February are unlikely to
affect the written material, but may affect my oral presentation.
Peter
-------
∂06-Feb-78 1316 FTP:Hart at SRI-KL (Peter Hart) Draft of AI Forecast
Date: 6 Feb 1978 1132-PST
From: Hart at SRI-KL (Peter Hart)
Subject: Draft of AI Forecast
To: Duda, Nilsson, Sacerdoti, Tenenbaum,
To: Feigenbaum at SUMEX-AIM, JMC at SU-AI, Newell at CMU-10B,
To: Reddy at CMU-10B, Minsky at MIT-AI, PHW at MIT-AI,
To: Amarel at RUTGERS-10, Uncapher at USC-ISIB,
To: Winograd at SAIL, Rick at RAND-UNIX, Denicoff at USC-ISI
ARTIFICIAL INTELLIGENCE AND NATIONAL SECURITY--
Some Opportunities
Peter E. Hart
Abstract
Artificial Intelligence (AI) is a field of inquiry
within computer science whose aim is to increase the
reasoning and perceptual abilities of computers. Experimental
AI systems have attempted to demonstrate these abilities in
the context of specific problems. Typical results of current
efforts are computer programs that deduce logical consequences
of facts stored in a database; that visually inspect workpieces to
identify flaws; that tutor an apprentice electronic
technician; or that significantly supplement the diagnostic
abilities of a trained physician.
It has been only within the past few years that advances
in AI research, coupled with continuing large reductions in the
cost of computation, have enabled AI to address problems of
practical significance. In this paper I attempt to clarify
the potential of artificial intelligence for contributing materially,
over the next two decades, to the national security of the
United States. I identify several prototypical information processing
functions, and describe-- using illustrations drawn from a variety
of national security applications-- how new capabilities would result
from incorporating projected advances in AI. I have assessed,
where possible, the anticipated technological risk and effort
associated with pursuing the opportunities discussed, and comment
on a few critical problems where technological breakthroughs appear
to be required. Finally, I attempt to anticipate
the general trends in artificial intelligence in other parts
of the world.
!
ACKNOWLDGEMENTS
!
EXECUTIVE OVERVIEW
!
I INTRODUCTION
Artificial Intelligence (AI) is a field of inquiry within
computer science that attempts to increase the reasoning
and perceptual abilities of computer systems. Because of the
ambiguity inherent in its name-- no one has offered an entirely
satisfactory definition of "intelligence" whether in a person or
in a machine-- AI is most easily characterized by example. Based
on work in AI, experimental systems have been demonstrated that:
- provide consultation to physicians on problems of medical
diagnosis [Shortliffe, 1976; Pople, 1975]
- analyze aerial imagery to count ships in a harbor or freight
cars on a railroad track [Barrow, 1977]
- answer database queries posed in natural, ungrammatical
English [Waltz, 1977; Sacerdoti, 1977]
- provide consultation to geologists on problems of
mineral exploration [Hart and Duda, 1978]
- determine the structure of large organic molecules from their
mass spectra [Buchanan and Lederberg, 1971]
- understand continuous human speech about restricted domains
of discourse [Reddy, 1976; Walker, 1976]
- control robot systems on the basis of video and other sensory
input [Winston, 1972; Nitzan and Rosen, 1976; Hart, 1975]
Philosophically, AI is founded on an appreciation of the fact
that computers are able to process abstract symbols in general, and
are not limited merely to numerical manipulations. They can therefore
deal with broad classes of situations, and with logical or other
relations among these classes. This viewpoint contrasts sharply with
the myth that "computers can do only, and exactly, what they have been
programmed to do." The examples above illustrate that computer
systems can be given broad strategies enabling them to deal usefully
with situations that have not been defined or anticipated in detail.
During the past several years, there has been an accumulation
of convincing evidence that AI has moved from a purely research phase
to an ability to engage problems of real-world significance. My
chief purpose in this paper is to identify ways in which this
growing ability can be addressed, over the next two decades,
to significant problems of national security. I shall be
particularly concerned with such issues as threat assessment,
status of own assets, flexible autonomous control, and life
cycle and personnel costs. In these and allied areas,
where adaptability, extreme flexibility, and survivability are
required, AI appears to be the critical technology.
In approaching this task, I have attempted to avoid
predictions based on direct extrapolation of current trends, and
have instead tried to identify technological opportunities: technical
goals achievable, on the given time scale, if appropriate resources
were devoted to the purpose. I have, however, made one major
assumption: I assume that the spectacular advances in large-scale
integrated circuits will continue. Specifically, I shall accept
current estimates that the equivalent of today's large scale
processors and memories will be possible on a few chips for a
few hundred dollars by about 1990 [Noyce, 1977].
The remainder of this paper is organized as follows. The
next section establishes a simple discussion framework in terms of
information processing functions and national security applications.
Sections III and IV describe a set of illustrative applications based
on AI that would lead to a very substantial multiplication of human
effectiveness. Section V briefly discusses the technological
foundation for these applications, and identifies two technical issues
where significant advances are still needed. Comments on Soviet
developments and some general observations conclude the paper.
!
II THE INFORMATION PROCESSING CYCLE AND NATIONAL SECURITY
The interactions between an information processing system
and its environment may be very broadly characterized as shown in
Figure 1.
*****************
* *
|----<-----<------* WORLD *--------<-------<----↑
| * * ↑
| ***************** ↑
| ↑
| ↑
| ================================================== ↑
| * Information Processing System * ↑
| * * ↑
| * ------------ ----------- ------- * ↑
|->*===>|Interpret/|====>| Plan |=====>| Act |==>*==>↑
* | Analyze | | | | | *
* ------------ ----------- ------- *
* *
* *
=================================================
FIGURE 1
A set of inputs is first accepted by the system for analysis
and interpretation. On the basis of this analysis, a plan is formed
for achieving a specified set of goals. This plan is then put
into action, and the system cycles to monitor and evaluate the
effects of its actions.
This elementary view describes the information processing
functions of a military commander at an abstract level: he must
analyze a situation in terms of his assets and an adversary's
assets and intentions; he must form a plan of action to achieve
a set of specified goals; he must initiate actions in accordance
with his plan; and he must iterate the process in order to monitor
the situation as it unfolds.
Computerized systems also follow this general pattern, but
one or more of the functions may be either trivialized or not present
at all. In conventional computerized systems, the problem of analying
input data is simplified as far as possible through insistence on
the use of formal programming languages like Fortran, or through the
use of highly stylized formats for submitting input data. Artificial
intelligence systems-- by which I mean conventional computers
whose programs incorporate principles of AI-- can present a sharp
contrast. The input may be ungrammatical English or photographic
imagery, and the task of analyzing and interpreting this input
can be complex indeed.
The planning function in conventional computer programs
ordinarily amounts to following a fixed algorithm specified by
the programmer; standard numerical computations are an example
of this, and are not usually thought of as "planning" at all.
By contrast, AI systems can generate plans to meet situations
that have been forseen in only broad outline by the system designer;
planning a sequence of databases accesses to answer an unanticipated
query from a user is an example of this [Sacerdoti, 1977].
It is worth pausing for a moment to reflect on this ability
of AI systems to accept such enormous variability in their input
compared with more conventional styles of computer programs. The
source of this ability does not reside in a few clever programming
"tricks," but in the incorporation within the system of a substantial
body of knowledge about the problem domain. For example, a system
for analyzing aerial photographs might know about much more than
the visual appearance of a few distinguished objects like trucks. It
might also know about "common sense" relations between trucks
and roads, about relations between roads, rivers, and
bridges, about relations between buildings and roads, and so forth.
A principal preoccupation of work in AI has therefore been to
determine just what knowledge is required to deal with a class of
problems, to invent methods for representing this knowledge in a
computer, and to develop computer procedures for effectively
applying this knowledge to any particular problem in the class.
I shall return to this theme a number of times subsequently.
Computer systems act on their environments by utilizing
standard terminal devices or by controlling other devices like
motors or signal generators; conventional and AI systems differ
little in this respect. They differ markedly, however, in their
ability to monitor the effects of their actions-- that is, to
complete the cycle shown in Figure 1. (I exclude from this
discussion conventional control systems, whether digitally implemented
or not.) Conventional computer systems may, for example, print
their output, but they do not normally have a means of determining
whether that output achieved a desired goal, much less have the
means for altering their strategy if need be. Some AI systems,
most notably those dealing with the control of autonomous robots
[Fikes, et al., 1972; Sacerdoti, 1977], do in fact close the loop
in a meaningful way.
In the following sections, I shall describe some hypothetical AI
systems in terms of these basic functions of interpreting and
analyzing the input data, forming a plan, acting, and monitoring
the effects of actions. I have also elected to describe hypothetical
systems under several areas of national security applications,
specifically the following: intelligence systems applications,
command and control systems applications, weapons systems applications,
and support system applications. Accordingly, I shall be discussing
systems using the following table as a guide.
!
\ |
\ Function | Interpret/ Plan Act Monitor
\ | analyze
\ |
\ |
Application \ |
area \ |
--------------- \-------------------------------------------------
|
Intelligence | ... ... ... ...
|
Command & | ... ... ... ...
Control |
|
Weapons | ... ... ... ...
|
Support | ... ... ... ...
|
-------------------------------------------------------------------
National security applications areas and information
processing functions
Table 1
Opportunities exist to develop AI-based systems that have
substantial capability in each of the information processing functions
named. Some systems will integrate all of these functions within a
single framework; a system exercising supervisory control of a
cruise missile would be an example of this. Integrated systems of
this sort are discussed in the next section.
Opportunities also exist, on perhaps a more modest scale, for
AI-based systems that emphasize only one of the information processing
functions named; a system for processing aerial imagery to assess
traffic movements would be an example of this. Systems of this sort
are discussed in Section IV.
!
III EXAMPLES OF INTEGRATED SYSTEMS
I shall attempt in this section to identify hypothetical
AI-based systems that have an ability to analyze input data,
to form a plan for achieving a stated goal, for acting
on the world to achieve that plan, and to monitor the effects of
their actions on a continuing basis. Accordingly, of the hypothetical
systems described can be thought of as occupying one of the rows
in Table 1.
Before describing these hypothetical systems, a word of
caution is appropriate. My choice of systems should by no means
be taken as either predictive or prescriptive. I do not forecast
that development of these systems will inevitably follow, nor
do I assert that these systems necessarily have claim to a high
priority. Instead, I have attempted to use examples to illustrate
how an evolving technology could be applied to problems of national
security.
A. Real-time Control of EW Reconnaissance Functions
The problem of attempting to create and maintain a useful
assesment of the electronic order of battle of a potential adversary
is an enormous one, whether measured in terms of signal variability
or sheer volume of data. Signal environments are very dense,
and the proportion of so-called exotic signals is increasing.
Furthermore, it is generally assumed that the signal environment
encountered during peacetime is likely to change markedly in
a wartime situation; locations of emitters, signal parameters, and
operating doctrine are all subject to change.
In view of these considerations, requirements for future
EW reconnaissance systems are likely to place a premium on flexibility
and on an ability to process a large volume of data. It may be
possible, within the next 20 years, to deploy reconnaissance systems
that make use of AI techniques to help meet these stringent
requirements.
Consider, then, a hypothetical system whose primary function
was to maintain an up-to-date electronic order of battle. The system
might involve a number of sensor platforms, each with multi-modal
sensors (perhaps RF, optical, and IR), and each with a large
amount of computational capability. A principal computational function
would be to analyze and interpret the significance of sensor data.
This analysis would have to be done using many different kinds of
knowledge about the EW problem domain. Certainly, the system would
need to know about the signal parameters and estimated locations
of all known emitters. In addition, however, the system could also
be given knowledge about deployment doctrine (e.g., complementary
artillery and missile defenses); about operational doctrine
(e.g., acquisition radar operates before tracking radar); about radar
system characteristics (e.g., alternative pulse repetition frequencies
of some emitters); about logistics requirements (e.g., ground access
to emitter sites); and even about the weather (e.g., optical trackers
may be inoperable.) It is worth noting that this knowledge would
not have to be used in very deep and subtle ways, analogous to the
use of mathematical knowledge to construct deep mathematical proofs.
Instead, so-called shallow reasoning would probably suffice.
For example, a few inferences based on the appropriate knowledge
may enable the system to anticipate a signal before it is received;
plausibly anticipated, the signal would be far less difficult to
analyze and classify than it would otherwise be.
The computational functions on board the platform would not
necessarily be limited to analyzing and interpreting the output
of the sensor array. A planning function could enhance the
effectiveness of the system in several ways. One of the more
straightforward uses of planning would be to allocate sensor
resources. This would certainly be desirable if the system to verify
the anticipated presence of particular emitters with particular
characteristics. It would also be desirable if the system recognized,
on broad grounds, the existence of gaps in the current electronic
order of battle, and planned the use of sensor resources to search
for emitters that would normally be expected. Carried one step
further this planning function might suggest real-time course
changes, subject to approval by the crew, that would result in more
complete coverage of an area.
The hypothetical reconnaissance system I have outlined
is an example of an integrated system that can be thought of as
occupying the first row in Table 1. It offers the potential
of enhancing the effectiveness of current systems by making
better use of a given complement of sensor resources, both
in their real-time deployment and in the analysis of their output.
It rests on technical capabilities that are only now being
studied, but that may well be adequately explored in the next
ten years, leading to a deployable system perhaps ten years
thereafter. The single most significant technical obstacle
may center on speed of response of a sequential computer. This
and related issues will be taken up subsequently.
B. Command and Control Management Aid
Command and control of military forces, at both the tactical
and strategic levels, is unquestionably a key information processing
issue affecting national security. Especially in crisis situations,
military outcomes may depend not so much on the gross balance of
forces as on the timely application of particular forces to carefully
specified objectives; the Mayaguez incident provides a good example of
this at the tactical level.
As the defense establishment prepares for the 1980's and
beyond, it appears certain that a major share of the entire information
base supporting command and control decisions will reside on digital
computers. Methods for dealing effectively with this mass of
computerized information will become critically important. Advances
in artificial intelligence are likely to play a key role in filling
this need.
I envisage an integrated computer-based system that supports
a military commander by serving him in a variety of ways. The first
important function it provides would be to enable the commander rapidly
to form a coherent picture of his own resources: his physical assets,
their geograpical deployment, their state of readiness, and their
logistic support. The raw data for preparing this assessment would
be contained in a set of databases. For reasons of security
and survivability, the computers supporting these databases will
be dispersed over a large geographical area and will be connected
by secure computer network communications. For reasons having to
do with the evolution of both hardware and software designs, the
computer databases will take on a variety of logical forms, and will
be managed by a non-uniform set of local database management systems.
Under these conditions, answering even a single "elementary"
query-- for example, a query of the form "What is the composition,
location, and readiness of Unit X?"-- will be in general a difficult
technical problem. The query will in principal be
answerable on the basis of available data. It will, however,
be necessary to construct the answer piece by piece, through the
process of accessing specific data items residing on different
computers. Carrying out this process will require a mass of detailed
information about such matters as the general contents of the various
databases; the names and logical formats of particular files and
file components; the formal query language provided by each local
database management system; the network protocols for establishing
communication channels; the availability of backup files and machines,
and so forth.
All these technical details are unrelated to the military
picture sought by the commander, and will unavoidably distract him
and his staff from their principal concerns. There is little doubt,
however, that AI systems could be implemented in the next 20 years
that would embody much of this detailed knowledge. Such a system
would at a minimum use this knowledge to insulate the commander and
his staff from the technical minutiae of the information processing
systems they rely on. Beyond this minimum, the AI-based command and
control management aid could perform additional valuable functions
such as the following.
A second important function of the command and control
management aid would be to allow a commander to test strategies
by exploring "what if" situations on a computer. Facilities could
be provided that would allow him to specify hypothetical scenarios
and engagements. The system would explore the consequences of the
various contingencies in terms of effects on manpower, weapons,
vehicles, and so forth.
After a commander had defined a course of action, the
hypothetical system could play a critical role in helping him implement
it. One important system function would be to construct a detailed
plan of actions that must be taken to support the general plan
outlined by the commander. For example, carrier missions require
careful scheduling of many functions, including aircraft maintenance,
fuel and armaments, pilot availability and briefing, and so forth.
In order to perform these scheduling functions, the system would
need a considerable amount of knowledge about such matters as
normal amounts of time and resources needed for each plan component,
alternative strategies for satisfying a particular plan component,
and interactions among plan components.
Once a plan has been put into effect, the hypothetical system
could help the commander monitor the outcome. Of course, he would
normally pay close attention to an evolving situation himself,
but while he was monitoring major aspects of a deployment or
an engagement the system would be examining logistics
data flowing into the information processing system. At an
elementary level, the system would prepare exception reports
or sound alerts simply because a critical variable fell outside
an acceptable range-- for example, fuel supplies may be
unacceptably low. At a more sophisticated level, the system would
use its knowledge of pro forma plan execution, and its understanding
of the interactions among plan components, to sound an alert whenever
a potentially dangerous situation developed. For example, fuel
supplies may be dangerously low only because of a particularly
unfavorable combination of meteorological conditions, sea state,
and deployment of supply ships.
The hypothetical system I have outlined could be elaborated
or simplified along several dimensions. For reasons of brevity,
I have discussed only the assessment of friendly assets; a
complementary picture could be drawn for estimating the state of
enemy resources. I have also omitted important details of how
the hypothetical system would communicate with a commander and
his staff; that issue will be discussed in a later section.
To a limited extent, conventional information processing
systems-- that is, non-AI oriented systems-- can deal with a few of
the technical issues raised above through the expedient of
first attempting to forsee all forms of questions that will
ever be asked, and then using preprogrammed methods for
constructing the answers. That this expedient is doomed to
failure should be obvious. Crisis situations inevitably lead
to unforseen questions; failure or destruction of system components
lead to unforseen system configurations. Requirements of flexibility
and survivability will demand AI methods for computing answers
dynamically to virtually any question within the general scope of
the system as it exists at the moment.
The hypothetical system I have described is an example of an
integrated system that can be thought of as occupying the second row
of Table 1. Before leaving the topic, it is worth commenting
generally on the knowledge that the system would need
and the style of reasoning that would use. In common
with the EW reconnaissance system already described, the
command and control aid would need a broad base of knowledge but
would not normally need to engage in deep reasoning about that
knowledge. This observation has important implications for the
kinds of capabilities needed in the technology base supporting this
application.
C. Semi-Autonomous Cruise Missiles
Recent Congressional and Presidential decisions make it likely
that cruise missiles will be an important aspect of our defense
posture over the next two decades. Here I take up the question of
how the effectiveness of a fleet of cruise missiles could be
multiplied through the installation of sophisticated on-board
computational capabilities.
It should be possible, over the next 20 years, to equip
cruise missiles with an AI-based control computer that accomplished
some of the functions normally performed by the pilot of a manned
aircraft. The computer system would need to be furnished at the
outset with knowledge about many of the aspects of the specified
mission. At a minimum, it would need to know about the physical
characteristics of primary and secondary targets, about the
terrain to be overflown, about enemy defenses, about the composition
of friendly forces (including other cruise missiles), and about
the flight characteristics of the missile being controlled.
The control computer would use this knowledge to select
particular targets, and to create a general plan for reaching them.
The chief advantage of such a system over conventional approaches
is flexibility. Particularly when operating in squadrons,
there will be a critical need to coordinate the activities of
individual missiles in order to achieve specified objectives.
The required level of organization will have to be accomplished
in real time, based on the course of a particular engagement.
Communications between a deployed cruise missile and a remote
human pilot are likely to be limited by constraints of security,
bandwidth, and enemy jamming. Under these circumstances some
degree of missile autonomy will be required, but a rigid,
pre-programmed flight plan will result in lowered effectiveness
of any fixed amount of missile resources. Artificial intelligence
methods, including the possibility of having missile controllers
"negotiate" among themselves using line-of-sight communication
channels, are likely to provide the required flexibility.
The hypothetical control computer I have outlined can
be thought of as occupying the third row of Table 1. Advanced
versions of this system-- by which I mean systems embodying a
substantial fraction of the abilities of a human pilot-- probably
lie beyond the the two decade horizon of this discussion. More modest
versions, achievable within two decades, are likely nonetheless to
be very valuable; a moderate ability to reorganize a missile squadron
dynamically would pose a serious psychological and real threat to
an adversary. A worthwhile interim goal would be to explore
the dimensions of this tradeoff between system complexity and
system effectiveness; thus far, it as received little attention.
D. Automated Maintenance Instructor and Operational Aid
Equipment maintenance is one of the least glamorous yet most
important issues facing the armed services today. Over half the
enlisted men in the Air Force are in maintenance roles; similarly,
the most labor-intensive activity in the Navy is associated with
diagnosing and repairing equipment failures. From the standpoint
of both cost and operational readiness, the maintenance issue is
already a critical one. Unfortunately, all current trends-- the
increasing complexity of equipment, the decreasing availability
of 18 year old men, the decreasing education levels of enlistees,
and the increasing rate of turnover-- all these trends will make the
maintenance problem even more acute over the next 20 years.
There are many ways in which AI can alleviate maintenance
problems by providing maintenance personnel with the right knowledge
in the right form at the right time. One family of approaches
would be to augment or supplant the present system of technical
manuals with interactive computer systems. A relatively straightforward
variant of this approach would be to implement an "active index"
to existing maintenance manuals. Unlike a conventional, passive
index, the active index would first accept a problem statement from the
user and would then direct him to appropriate parts of the maintenance
documentation. The user-- i.e., the maintenance man-- would pose
the question in a natural way using a combination of voice,
touch-screen, and keyboard devices. The system would need both to
analyze the user's problem and to assess his sophistication
in order to direct him to the appropriate documentation.
A slightly more advanced system than the one just described
would contain the maintenance documentation itself, not just the
active index, in computer-readable and computer-displayable form.
At the cost of an increased amount of computer memory, the system
would rapidly be able to display the appropriate documentation
to the user. The system would also be able to engage in a more direct
dialog with the user. For example, the user might point to a
diagrammed part and inquire about its name, part number, function,
and current inventory level.
More advanced versions of a maintenance aid would rely not
so much on retrieving selected portions of previously written
documentation as on creating the documentation dynamically
to satisfy a particular need. Consider, for example, the problem
of failure diagnosis for a complex piece of equipment. Pre-planned
diagnostic procedures like conventional decision trees have severe
limitations that are well-appreciated by those who have written or
used them. Alternatively, AI systems-- like those used to furnish
consultation on problems of medical diagnosis-- can suggest sequences
of tests or observations to resolve the particular problem of the
moment. These consultation systems form their diagnostic strategy
dynamically on the basis of a complex set of associations between
observables and underlying conditions; strategies so constructed
are typically far more effective than any fixed sequence of tests.
The flexibility provided by AI systems would also be valuable
in furnishing instructions about assembly and disassembly
of a piece of equipment. By exploiting knowledge of the basic
construction of a piece of equipment, a hypothetical maintenance
aid could suggest a procedure for transforming the equipment from
any state of partial assembly to any other state of partial (or
complete) assembly. Equivalently, the system would be able to
ascertain whether or not a sequence of assembly steps executed by the
maintenance man was valid.
I have described a hypothetical sequence of increasingly
sophisticated maintenance aids that are examples of integrated
systems occupying the fourth row of Table 1. The emphasis in
any of these systems could be placed either on its pedagogic
value in an educational setting or on its operational value in
other settings. It is also worth observing that each hypothetical
system could be upgraded to incorporate features of more advanced
versions. In view of the cost of existing technical manuals-- one
1974 estimate put that cost at $2.5 billion for the Navy alone--
this evolutionary aspect is important for reasons beyond those
of technical risk.
!
IV EXAMPLES OF APPLICATION MODULES
In the preceding section I suggested several ways in which
artificial intelligence could be employed in integrated information
processing systems. Here I will illustrate some potential
applications of AI as a modular component. Each of these
modules emphasizes one of the processing functions heading the
columns in Table 1, and can so be thought of as occupying one or
more of the cells in the appropriate column. As before, I caution
that the hypothetical application modules are intended only to serve
as examples of technological opportunities, and are not necessarily
to be taken as predictions.
A. Natural Language Access to Computers
As computers become more nearly omnipresent in military life,
and as the proportion of the military having direct contact with
computers grows, there is an increasing need to make interacting
with computers as simple and natural as possible. One approach
to satisfying this need is to provide computers with an ability
to accept and understand ordinary, or "natural," languages like
English.
It will certainly be possible, over the next 20 years,
to construct useful natural language interpreters for a variety
of tasks; indeed, such interpreters or "natural language front
ends," are beginning to appear already. Among the characteristics
we can expect to have are the following.
Front end modules will accept ungrammatical, colloquial
English text as input and translate that text into an
unambiguous command, statement, or query to the computer
system. It will be possible to personalize the front end for
individual users either by providing it with personal preference
profiles directly, or by allowing it to form personal models from an
accumulation of interactive dialogs with the user.
Over a period of time, a user typically becomes more familiar
with the computer and with his specific task, and will therefore
wish gradually to replace verbose forms of ordinary language with
terse abbreviations; front end modules will accomodate this shift
unobtrusively.
More advanced front end modules are possible that would
allow the user to augment natural language text with convenient
graphics input and/or with voice input. With these additional
channels, the user would experience a more comfortable computer
interface as well as a faster communication rate.
The hypothetical front ends I have sketched have the
interpretation and analysis of input data as their principal
function, and so would fall in the first column of Table 1. The
possibilities for applying such front ends appear to be almost
limitless, especially in view of rapidly decreasing hardware
costs. Users wishing to access databases for a wide range
of applications would find them to be effective aids. Students
interacting with computers for instructional purposes would find
them invaluable. Interestingly, even computer programmers, who
have daily professional contact with computers, would quickly
appreciate the utility of these front ends for invoking seldom-used
capabilities of the machine.
B. Analysis of Reconnaissance Imagery
The defense establishment daily gathers large amounts of
raw photographic material, much of it taken from satellites and
high altitude aircraft. The timeliness of intelligence products
prepared from this material would be improved if AI aids were
developed to automate the more labor-intensive, but routine, aspects
of photointerpretation.
It should be possible, over the next two decades, to give
a photointerpreter an interactive computer-based aid that
at a minimum performed counting tasks (freight cars, aircraft on
the ground, storage tanks, etc.) and mensuration tasks (length of
runway, height of building, width of roadway, etc.) Mensuration
tasks would be entirely straightforward to implement if the user
indicated, on a touch-screen device, the objects to be measured.
More difficult, but still feasible, would be capability allowing the
user first to indicate a particular object in a picture and
then to request that all similar objects be measured or counted.
The ability to find all objects similar to a designated
object also leads to a monitoring or alerting capability. For
example, the photointerpretation aid might be instructed to count
the number of ships in a harbor and asked to notify the user whenever
that number changed by some pre-specified amount. Alternatively,
the system might be instructed to monitor vehicular traffic on a road
or in an area, and asked to alert the user when some volume of traffic
was exceeded.
The hypothetical interpretation aid I have outlined emphasizes
the analysis and interpretation of relatively unconstrained input
data for intelligence purposes, and so would fall in the top left cell
in Table 1. An attractive technical aspect of the hypothetical
system is that a relatively high error rate could be tolerated.
In an alerting role, for example, an incorrect alert would
cause additional work for the person using the system,
but would not ordinarily lead to any more serious a consequence.
The system could therefore be biased to incur a relatively high
probability of false alarm, and would therefore achieve a relatively
low probability of missing a significant development.
C. Assurance of Database Validity
A traditional problem of database management is to assure that
new data entering the system is valid. This problem is likely to
become more acute as reliance on computer databases increases, and
as geographical dispersion of databases becomes more common. Errors
in entering new data will take many forms, ranging from typographical
errors to deliberate misrepresentation or falsification of data
by the responsible authority.
Existing methods for dealing with this problem are primitive,
and usually involve elementary checks on individual data items. For
example, the name of an airfield will not satisfy even a simple
check on the allowable entries for "type of aircraft," nor will
a value of 3000 satisfy a numerical check on "number of pilots
on board."
To achieve a greater assurance of validity than is possible
by these elementary means, it will be necessary to develop
methods that exploit more extensive general knowledge of the
activities and situations recorded in the database. For example,
overland troop movements involve certain typical relations
among number of troops, armament, support equipment, distance and
terrain, and so forth. In general, it will not be possible to capture
these relations using only algebraic relations. Instead, a richer
model that (in this example) reflects the logical organization of
a troop movement will be needed, and AI methods for representing
such situations and processes will be called for.
A data stream purporting to encode the status of a troop
movement would need to be in harmony with the general model of
such situations as represented in the validity-checking module.
The module would check not only for internal consistency at a
given moment, but would also check that the time sequence of
database updates satisfied plausible constraints. Where these
checks fail to be satisfied, exceptions would be reported.
Possible excuses-- by which I mean alternative explanations of
the exceptions-- could also be furnished.
The hypothetical module I have outlined functions as a
monitor of externally generated events, and so can be thought of
as an example of an application module in the fourth column of
Table 1. There has been to date very little work aimed at
developing modules along the lines suggested, but the growing
reliance on computer databases points to a growing need for this
capability.
D. Automatic Programming
The cost of computer software is a multi-billion dollar
expense in defense procurements and operations. Attempts are
made to control these costs through what may be termed conventional
approaches such as standardizing languages and improving
the management of software development efforts. These attempts
should continue, and no doubt will, but there is nonetheless
a need to effect quantum improvements in the efficiency of the
software enterprise.
The cost of producing software can be usefully separated into
two major components: the cost of specifying in detail what a
piece of software is intended to do, and the cost of writing a
program satisfying that specification. It appears that AI methods
could be developed to facilitate both the specification and
the program writing processes. Here I shall concentrate only
on the second activity, the translation of a specification into
a computer program, because it is the better defined technically
of the two.
It is possible to specify the intended function of
even complicated computer programs in a precise and complete
fashion; indeed, this has already been done in isolated experimental
cases. Given such a specification, there are formal logical
methods for proving mathematically whether or not a given computer
program in fact satisfies its specifications. These
logical methods can be embodied in an AI module that verifies the
consistency of a computer program with its specification.
It should be possible, within the next 20 years, to extend
the abilities of existing AI program verification modules to the point
where they can verify even large programs. Operating in either a fully
automatic mode or in an interactive, user-guided mode, such a module
would insure that an expensive or critical program was free of
errors. The reliability of software would then become comparable
to the reliability of hardware; developments in either arena would
proceed from an exhaustive and precise specification to a realization
satisfying it.
The availability of a program verification module also has
important implications for the task of modifying an existing program
to meet a new need. It is widely appreciated that modifying any large
program is likely to introduce new errors; these errors could be
identified and controlled through the use of a verification module.
In addition to furnishing the means for verifying the
consistency of a program with its specification, AI can contribute
methods for actively synthesizing a program to satisfy a given
specification. As in the case of program verification, a module
could be developed to operate in either a fully automatic mode or
in a user-guided mode, depending upon the difficulty of the task.
Program verifying modules and program synthesizing modules
emphasize an ability to do logical planning, and so can be
thought of as occupying the second column of Table 1. They are
particularly attractive because they establish the reliability
of software by a process of mathematical proof, rather than
by the conventional expedient of testing a finite number of cases.
!
V TECHNOLOGICAL FOUNDATIONS
The preceding sections describe a few of the potential
applications of artificial intelligence that may be possible
over the next two decades. Here I shall try to sketch very briefly
the AI technology that underlies these and many
other such applications. I shall also attempt to identify a
small number of critical technical problems where quantum
advances in the underlying technology are possible, and I will
suggest some relevant considerations for the formulation of research
policy.
A. Fundamental Issues
The central preoccupation of artificial intelligence is with
knowledge:
- what kinds and what breadth of knowledge will a system need
in order to deal effectively with a given class of problems?
- how may this knowledge be represented in a computer?
- what computer procedures will be required to apply this
knowledge to a particular problem?
This preoccupation has resulted in the evolution
of a body of techniques of considerable generality
for representing and manipulating many different types of knowledge.
As a research strategy, these AI techniques have usually been
developed in the context of application-oriented problems-- for example,
problems of interpreting English sentences or problems of analyzing
photographic imagery. Each problem class has its own unique
characteristics, but each also affords an opportunity to explore
central issues of general interest.
I shall follow generally the outline contained in
[Nilsson, l974], and first discuss some of these central scientific
issues. Subsequently, I will review briefly the exploration of
such issues within the context of particular classes of problems.
A great deal of knowledge about classes of problems can be
structured as collections of facts. Facts may be simple-- "The
USS Deepsix is commanded by J.J. McCoy"-- or may be more complex--
"All crashes of USAF aircraft between 1974 and 1976 that were
manufactured by either McDonnell-Douglas or Lockheed had at least
one landing gear component manufactured by Rohr."
The representation of knowledge as a structured collection
of facts is called an assertional representation. Assertional
representations generally stress static aspects of knowledge,
or knowledge that obtains at a particular point in time. Research
in artificial intelligence has lead to the development of a variety
of techniques for the assertional representations of knowledge.
Some of these representations are based directly on the foundations
of formal logic, and so have a mathematical heritage extending many
decades into the past. Others, invented only within the past few
years, represent knowledge as networks of facts whose associations
and relations are specified by the linkages in the network.
Assertional representations, while indispensable for
representing many static aspects of knowledge, are awkward for
representing dynamic aspects of knowledge or knowledge emphasizing
repeated or conditional actions. For example, a contingency plan
may specify a number of alternative actions to be attempted under
varying conditions, or may call for repeated attempts to achieve
a goal unless or until some other condition holds. Knowledge of this
sort is more naturally represented in the form of logical or
temporal processes, rather than as static assertions. Such
representations are called procedural representations of knowledge.
Artificial intelligence research has lead to the development of
a number of ways in which knowledge may be represented procedurally.
Procedural representations typically share many of the characteristics
of conventional programming languages (which are themselves formalisms
for representing procedures.) However, the procedural representations
used in AI incorporate many special constructs that greatly enhance
their usefulness. These constructs include, for example, mechanisms
for specifying general conditions for invoking procedures-- conditions
which need not be defined precisely ahead of time by the system
designer.
Current AI systems frequently use a combination of assertional
and procedural representations of knowledge to achieve their
goals. All representational schemes, however, need to address certain
underlying design issues. One important issue concerns the degree
of detail with which any given body of knowledge is to be represented.
It is an evident truism that finely detailed representations will
be more unwieldy than coarse representations and, correspondingly,
that the coarsest representation adequate for the task is to
be preferred. Unfortunately, the range and variability of tasks
confronting any single AI system may be very large, and specifying
the requisite level of detail is likely to be difficult; in fact,
a variable level of detail is often needed. One of the contributions
of AI research has been the development of hierarchical methods
for representing both static and dynamic knowledge at variable
levels of detail, and for shifting attention smoothly from one
level to another.
In order for a body of knowledge to be useful, a means
must be established to apply that knowledge to a particular problem--
that is, to reason from the knowledge to an acceptable problem solution.
Artificial intelligence research has explored many ways of implementing
reasoning processes on computers. Some of the methods are motivated
by the studies of cognitive psychologists on human problem-solving
performance. Other methods are based on formal proof procedures
developed for problems in mathematical logic. Still others,
adapted from the methods of statistical decision theory, are used
where plausible inferences rather than strict logical deductions
are needed.
Regardless of the philosophical motivations underlying them,
reasoning processes are needed to deduce relevant consequences of that
knowledge explicitly represented in the computer. This ability
is critically important, because explicitly represented knowledge
typically is only a very small fraction of the knowledge needed
to solve any specific problem-- the remaining knowledge is implicit,
and therefore must be deduced. Reasoning processes are also used in AI
systems to construct plans of actions. The ability to construct plans
has obvious importance when the task is, say, the control of a
remote vehicle like a cruise missile. Less obviously perhaps, but
equally important, is the use of planning capabilities to control
the operation of the computer itself. For example, a system for
analyzing a sequence of photographs to monitor equipment movement
may depend on an internal planning function to specify customized
processing steps for each frame.
One of the achievements of AI has been to develop
systems that make simultaneous use of general planning and
deductive abilities as well as deductive abilities that are
specialized to particular classes of problems. This has led
to systems whose competence in defined areas even today approaches
that of skilled human specialsts, but whose internal mechanisms
do not have to be designed entirely anew for every problem class.
Artificial intelligence systems, as the preceding remarks
suggest, are typically large and exceedingly complex programs.
Furthermore, the relatively broad applicability of the core ideas
means that similar mechanisms are likely to be employed in
a range of programs. There is thus the need and the means to
simplify AI experimentation by developing advanced programming
languages that facilitate the use of the core mechanisms.
The possibility of following this research strategy was recognized
very early, and since the late l950's there has been a succession
of increasingly advanced languages specifically designed to aid
AI research. These so-called "AI languages" are extremely important
tools for developing new systems; they reduce the programming
effort for creating new systems by large factors compared to the
effort that would be required if more conventional programming
languages were employed.
B. Applications as Research Vehicles
Most research in artificial intelligence has been
pursued not by examining the core issues abstractly, but by
confronting problems in particular areas of human intellectual
activity. Many of these areas have been alluded to already;
among the more prominent are:
- analysis of photographic imagery
- interpretation of English text
- interpretation of continuous speech
- verification and synthesis of computer programs
- control of remote vehicles
- interactive, professional-level consultation on
specialized areas in medicine and other sciences
- deduction of implicit facts from explicit databases
Much effort has been directed toward developing AI systems that
function in these and other application areas. Historically,
projects to design and implement such systems have been intended more
to further understanding of basic principles than to achieve a
practical goal. These projects are sometimes termed first level
applications of artificial intelligence.
The first level applications of AI have, in the past several
years, begun to reach rather impressive levels of performance in their
own right. For example, the INTERNIST medical consultation system
[Pople, 1977] has successfully diagnosed multiple disease cases
taken from the medical literature. The MYCIN medical consultation
system [Shortliffe, 1976] can diagnose bacterial infections at a level
of skill comparable to that of a general practitioner. The
DENDRAL system for inferring chemical structure from mass spectra
operates more accurately and more quickly than skilled human
specialists [Buchanan and Lederberg, 1971]. A program verification
system developed by Boyer and Moore automatically verified the
correctness of a small compiler for LISP expressions
[Boyer and Moore, 1977]. The LADDER system [Sacerdoti, 1977]
is able to analyze English questions, and automatically
deduce answers from data contained in a number of remote
computerized databases. The SOPHIE system [Brown et al., 1974]
uses the medium of restricted English to teach a technician how
to troubleshoot regulated power supplies. The HAWKEYE system
[Barrow, et al., 1977] uses stored maps to guide it in
such tasks as tracing roads or counting rail cars in aerial
photographs. The HARPY system [Reddy, 1976] achieved over
90% accuracy in interpreting continuous spoken English questions
about limited subject matter, but with a total vocabulary of one
thousand words. A number of experimental systems for automating
industrial manufacturing processes use visual and other sensors
to guide the acquisition of parts from a moving belt or to facilitate
the assembly of simple mechanical components [Nitzan and Rosen, 1976;
Binford, et al., 1975; Nevins and Whitney, 1975].
Limitations of space prevent a fuller discussion of these
and other first level application systems. A more complete survey,
with many references, is given in [Duda, et al., 1978].
Within the past few years, some AI-based capabilities
have become viable in practical terms. For example, some recently
introduced manufacturing methods rely on the computer analysis of
imagery to guide automatic tools. Such an application is sometimes
termed a second-level application; the principle objective is a purely
practical one. It is to be expected that when AI technology is
embedded in a second-level application system, the emphasis shifts
away from AI and towards immediate practical concerns. As more
second-level application systems appear, it will be important for
research policy considerations to appreciate the intellectual
ancestry of the technology being capitalized upon.
In the face of the great diversity in first-level
applications, it is worth pausing to comment on the internal
coherence of artificial intelligence as a field of inquiry. A
major reason for its coherence is the extent to which the core
scientific issues, and the steps taken to address them, cut
across the application areas. As an illustration of this,
the identical representational mechanism has been used in distinct
systems that analyze English text, interpret photographic imagery,
deduce facts from databases, and consult on a class of scientific
problems. Conversely, ostensibly unrelated application efforts often
contribute complementary advances to the core technology.
C. Opportunities for Technological Breakthroughs
A brief review of artificial intelligence can do no more than
hint at the extremely complex issues underlying the design and
implementation of AI systems. I hope, however, that one point is
evident: artificial intelligence is to be achieved not by patching
together a few clever programming tricks, but by employing a
relatively large body of knowledge to address a relatively
unconstrained set of problems. Accordingly, breakthroughs in AI
are likely to center explicitly on the acquisition, representation,
and use of knowledge. Here I select three technical problems
where quantum advances are possible within the next ten years,
and where these advances would qualitatively improve our ability to
achieve practical goals. While not all workers in AI will necessarily
agree with my choices, I believe that most would have these or
closely related problems high on their own lists.
The problems I highlight are:
- acquisition and management of large knowledge bases
- representational mechanisms for extended classes of
knowledge
- exploitation of multi-processor computer architectures
for AI systems
I shall briefly comment on each of these.
As AI systems attempt to exploit increasing amounts of
knowledge, the methods for acquiring and managing knowledge
assume critical importance. Large knowledge bases by definition
cannot be provided to the system in a short time; they
must be obtained incrementally, frequently from many individuals
or other sources, over a protracted period of time. Means for
facilitating this process might include interactive modules for
interviewing the knowledge source (often called the "expert");
for prompting him in a natural way (akin to how another person might
prompt an expert); for detecting inconsistencies; and for
unobtrusively managing the storage of knowledge in several levels
of computer memory. Some advances in knowledge acquisition and
management, chiefly within the confines of a single first level
application problem, have already been made; much greater advances
appear possible.
I earlier mentioned some of the general approaches to the
very broad problem of developing computer representations of
knowledge. The very difficulty in defining precisely what is meant by
"knowledge" gives a strong hint at the difficulty encountered in
developing mechanisms that would suffice to represent any particular
"piece of knowledge." Many AI workers believe that the field is
still in a relatively primitive state with respect to knowledge
representation, the impressive demonstrations of first and second
level application systems in the past several years notwithstanding.
For example, there is currently no fully satisfactory way in which
to represent and reason about such seemingly elementary statements
as "Captain Jones believes that Admiral Smith knows that X is true."
My third choice of problems-- the exploitation of multiprocessor
architectures-- is motivated by the large amounts of computation that
AI systems typically require, coupled with the stunning advances
in large scale integrated circuits that can be expected to continue
for many years to come. To date, virtually all AI programs have been
developed for traditional sequential computers. A few experiments
have been conducted with novel architectures involving on the order
of ten parallel, general purpose processors. A few additional
experiments have been conducted using arrays of very simple logic
units for specialized problems in picture analysis. There has
been almost no serious exploration of designs involving on the order
of hundreds of general purpose processors, or designs involving on
the order of millions of very simple logic units (other than for
restricted problems in picture analysis.) Yet decreasing hardware
costs will render both of these approaches feasible in coming years.
The three problems I have outlined are moderately well
defined, and qualitative advances over the coming decade are
possible. It should be pointed out, however, that the record of
predictions in artificial intelligence is not an especially bright
one. Some early predictions of achievements, made at a time when
little concrete experience was yet available, have not been
satisfied. On the other hand, current capabilities for interpreting
English questions are far beyond what most knowledgable workers
would have predicted as recently as three years ago.
D. Research Strategies
Artificial intelligence as a field of inquiry has deep
theoretical concerns, but in its daily affairs AI operates chiefly
as an empirical discipline. New ideas for intellectual mechanisms
must be made precise enough to permit computer implementation, and
once implemented these mechanisms must be experimented with to reveal
their properties. Indeed, a preoccupation with rigorous definition
and testing is one of the main distinctions between AI research
and more general philosophical inquiries into the nature of
intelligence.
Any research strategy for pursuing possibilities such as
those outlined earlier will need to recognize the empirical aspects
of AI research and development. Virtually all current AI programs
represent an investment of typically between three and fifteen man-years
on program implementation and experimentation. Serious attempts
to create programs of very high performance entail perhaps two or
three times this amount of effort. All these figures would
be several times larger if computers having AI languages and
other special software were not available to facilitate program
development. The investment would also be increased several
times if research teams, representing a blend of many highly
specialized skills, were not largely in place at the initiation
of a project. The implications of all this for R & D policy are
evident: Significant AI undertakings require a critical mass of
people and computational facilities that can be assembled only with
considerable time and effort. A decision to realize some of
the potential of AI to contribute more heavily to national
security involves a committment to maintain centers of excellence
with the requisite critical mass.
A second important issue for research policy is the question
of the appropriate blend of applications-oriented development with
more fundamental research. There are of course many conflicting
points of view on this issue, and no single suggestion will find
approval in all quarters. I believe, however, that an especially
productive research strategy would be to establish projects whose goals
are reasonably well-defined in practical terms, but whose reward
structures do not notably penalize failure to achieve those practical
goals. Given thoughtfully selected goals of an appropriate degree of
difficulty, this strategy is likely to produce either practical
results, significant scientific advances, or both. Furthermore,
any scientific advances would be made in a setting that facilitates
their eventual application to practical problems. (Naturally,
research efforts producing neither practical solutions nor scientific
advances would be judged accordingly.)
I do not, of course, recommend this R & D strategy to
the exclusion of all others. Fundamental research efforts
will always be needed to expand the conceptual base of AI; without
question, many of the most important technical avenues in AI today
were opened as a result of undirected basic research.
Engineering efforts will always be needed to translate proven
designs into practice; products of AI research institutions are
unlikely to be fieldable. The strategy recommended does, however,
offer a critical advantage worth re-emphasizing: The freedom to
fail in an attempt to achieve a practical goal means the freedom
to incur technological risk. Without that freedom, what are
nominally research efforts will result at best in only incremental
advances in the underlying technology.
!
VI ARTIFICIAL INTELLIGENCE ABROAD
Most of the research in artificial intelligence has
historically been performed in the United States and, at least insofar
as the open, English-language literature is concerned, US dominance
in AI continues to the present time. Other countries, however, have
shown an increasing commitment to AI research and development.
For example, one of the oldest and largest centers of AI research
is in Scotland at the University of Edinburgh, under the direction
of Professors Bernard Meltzer and Donald Michie. Another important
center of European activity in AI is at Linkoping, Sweden, under
the direction of Professor Eric Sandewall. Smaller AI groups have
been in operation for many years in France, Italy, and West Germany.
Japanese interest in AI has heightened very noticeably in
the past five years, with serious AI efforts now established in
industrial, university, and government laboratories. Significantly,
the principal, biennial international AI conference is scheduled
to meet next in Japan. Japanese work appears to be aimed chiefly
at problems of industrial automation, and emphasizes advances in
such tasks as the use of computer vision to control manipulators.
Soviet activity in AI is difficult to assess with any degree
of precision, but available evidence points unmistakably to a
substantial national committment to pursuing artificial intelligence.
Perhaps the first real indication of the extent of Soviet interest
came in 1975 at the biennial AI technical conference. The conference
was held in the USSR at the strong request of Soviet researchers;
surprisingly, approximately 200 Soviet workers attended the conference,
notwithstanding the fact that it was held in Tbilisi, Georgia, many
hours of flying time from other research centers. It was apparent
that the Soviet workers had followed the Western AI literature extremely
closely.
Since 1975, it appears that Soviet efforts have concentrated
in two technical areas: command and control issues, such
as natural language data retrieval, and robotics research, such as
the development of computer-controlled manipulators and mobile
vehicles. Developments in these areas appear to be continuing in
many places: at the USSR Academy of Sciences in Moscow there is a
group in the Computer Center under Professor V. Briabin and another
group in the Institute of Applied Mathematics under Professor
D. Okhotsimsky. In Leningrad there are groups under Professor
Yurevich at the Leningrad Technical Institute and under Professor
A. Timofeev at Leningrad University. Other AI work is under way
in Kiev, Novosibirsk, Vladivostok, and Tbilisi.
There is no question that Soviet AI is hampered by a
shortage of computing power, but they continue to study the best
Western work very carefully and can be expected to maintain their
position of following closely behind the West in those sub-specialties
of greatest interest to them. They are training a large number of
students in artificial intelligence, and particularly in robotics
appear to be moving ahead very vigorously. Given better computer
facilities, and given a little time for the diffusion of software
from their premier laboratories, they will be in a position to challenge
the technological dominance of Western institutions in artificial
intelligence.
!
VII CONCLUSIONS
I have tried in this paper to indicate ways in which AI can
materially increase our national security over the next two decades.
The hypothetical systems and modules described address specific
classes of problems, but should suggest more generally the spectrum
of possibilities that is open. Achieving practical success in many
of these areas will require more than just "engineering advances"
beyond existing technology, but appears attainable given a national
will to do so. Any serious attempt to realize these advances,
however, will need to recognize that there will be competition from
the private sector for the limited supply of skilled AI technologists.
My own view is that artificial intelligence will in time
become integrated within a wide range of defense systems. Wherever
adaptibility, extreme flexibility, and survivability are required,
artificial intelligence appears to be the key technology. These
requirements are nowhere at a greater premium than in the national
security enterprise.
!
SELECTED BIBLIOGRAPHY
Barrow, H. G., et al., "Interactive Aids for Cartography and
Photointerpretation: Progress Report," Proc. Image Understanding
Workshop, Oct. 1977.
Binford, T. O., et al., "Exploratory Study of Computer Integrated
Assembly Systems," Progress Rept. Stanford AI Lab., Stanford
Univ., Stanford, Cal., Nov. 1975.
Boyer, R. S., and J. S. Moore, "A Lemma-Driven Automatic Theorem
Prover for Recursive Function Theory," Proc. Fifth Int. Jt.
Conf. on AI, Cambridge, Mass., Aug. 1977.
Brown, J. S., et al., "SOPHIE: A Sophisticated Instructional
Environment for Teaching Elctronic Troubleshooting (an Example
of CAI in AI)," BBN Report No. 2790, Bolt, Beranek, and Newman, Inc.,
Cambridge, Mass., Mar. 1974
Buchanan, B. G., and J. Lederberg, "The Heuristic DENDRAL Program
for Explaining Empirical Data," Proc. IFIP Congress, Vol. 71,
Ljubljana, Yugoslavia, 1971.
Duda, R. O., N. J. Nilsson, and B. Raphael, "State of Technology
in Artificial Intelligence," to appear in ********* REF NEEDED ****
Fikes, R. E., et al., "Learning and Executing Generalized Robot
Plans," Artificial Intelligence, Vol. 1, 1970.
Hart, P. E., "Progress on a Computer-Based Consultant," Proc.
Fourth Int. Conf. on AI, Tbilisi, Georgia, USSR, Sept. 1975.
Hart, P. E., and R. O. Duda, "PROSPECTOR-- A Computer-Based
Consultation System for Mineral Exploration," to appear in
J. Int. Assoc. for Mathematical Geology, 1978.
Nevins, J. L., and D. E. Whitney, "Adaptable-Programmble Assembly
Systems: An Information and Control Problem," Proc. Fifth Int.
Symp. Industrial Robots, IIT Research Institute, Chicago, Ill.,
Sept. 1975.
Nilsson, N. J., "Artificial Intelligence," Proc. IFIP Congress 74,
Stockholm, Sweden, 1974.
Nitzan, D., and C. A. Rosen, "Programmable Industrial Automation,"
IEEE Trans. on Computers, Vol. C-25, No. 12, Dec. 1976.
Noyce, R. N., "Large-Scale Integration: What is Yet to Come?"
Science, Vol. 195, No. 4283, 18 Mar. 1977.
Pople, H. E., Jr., et al., "DIALOG: A Model of Diagnostic Logic for
Internal Medicine," Proc. Fourth Int. Jt. Conf. on AI, Tbilisi,
Georgia, USSR, Sept. l975.
Pople, H. E., Jr., "The Formation of Composite Hypotheses in
Diagnostic Problem Solving: An Exercise in Synthetic Reasoning,"
Proc. Fifth Int. Jt. Conf. on AI, Cambridge, Mass., Aug. 1977
Reddy, D. R., "Speech Understanding Systems: Summary of Results of
of the Five-Year Research Effort," Dept. of Comp. Sci.,
Carnegie-Mellon University, Pittsburgh, Penn., Sept. 1976.
Sacerdoti, E. D., A Structure for Plans and Behavior, Elsevier,
New York, 1977.
Sacerdoti, E. D., "Language Access to Distributed Data with Error
Recovery," Proc. Fifth Int. Jt. Conf. on AI, Cambridge, Mass.,
Aug. 1977.
Shortliffe, E. H., Computer-based Medical Consultation: MYCIN,
Elsevier, New York, 1976.
Walker, D. E., (ed.), "Speech Understanding Research," Final
Technical Report, Stanford Research Institute, Oct. 1976.
Winston, P. H., "The MIT Robot," Machine Intelligence, Vol. 7,
B. Meltzer and D. Michie, eds., American Elsevier Publishing
Company, New York, 1972.
-------
∂06-Feb-78 1656 BCM lisp
I found an interesting mistake in m. newey's thesis. It relates to his
description of lcom0 and, in particular, combool.
Also, I think using the intensional/extensional stuff gives a nice way
of converting a lisp function to a new boolean function which is true iff
the original function is total. Perhaps, this and related properties
can be usedto simplify some parts of program verification.
Well, I talk to you tomorrow.
uvidimcya pozhe
Ben Moszkowski
∂06-Feb-78 1706 JMB Richard Shuey address
Due to power failure today, I didn't get your note till now. Will
mail you the address in the AM tomorrow.
....Jeff
∂06-Feb-78 2011 DCL
To: GROUP.DIS[VCG,DCL]:
***********************************************************************
VERIFICATION GROUP SEMINAR WEDNESDAY 8th. FEBRUARY
PLACE: ERL 237
TIME: 2.30 pm
TITLE: Verification in Languages with the Class Construct
SPEAKER: David Luckham and Wolf Polak
This is an informal discussion of the use of classes and the resulting verification
problems. Some proof rules for implementation in the verifier will be suggested
and applied to examples. Relation with Guttag's work and our current verifier rules
for arrays and pointers will be discussed.
***************************************************************************
∂06-Feb-78 2028 RAK via AMES-TIP#50 Overreaction
To: LES, JMC
I think I over reacted with respect to MRC today due to circumstances
not related to actual incidents. I think there may be a problem there,
but I was too strong in stating it.
Dick Karp
∂06-Feb-78 2145 FTP:MRC at MIT-MC (Mark Crispin) Look what I found at MC
Date: 7 FEB 1978 0042-EST
From: MRC at MIT-MC (Mark Crispin)
Sent-by: MRC0 at MIT-MC
Subject: Look what I found at MC
To: JMC at SU-AI, Les at SU-AI, MRC at SU-AI
Written for MINI-MICRO SYSTEMS
--------------------------------------------------------
---
Packet-Switching via Telenet--A Boon for Mini Users
by Richard Jay Solomon*
****NOTE TO EDITOR: TELENET MAY BE ONLINE WITH ITS
MAIL*** ****SYSTEM WHEN THIS ARTICLE GOES TO
PRESS--CHECK TENSE, AND ETC.***
When President Carter revealed after his campaign that
he had kept in touch with Vice-President Mondale's team
via computers, few in the mass media understood that
this system was no mere teletypewriter replacement.
Carter and Mondale sent each other messages, and their
staffs were able to transfer files and "teleconference"
via new form of electronic mail which is destined to
have far-reaching impacts in the next decade. And when
Representative Alec Rose, the maverick chairman of the
House Subcommittee on Congress' own use of computers,
pulls out a microprocessor- equipped portable terminal
at committee hearings, and then proceeds to confound
witnesses with up-to-date information from on-line data
banks, we can see that we have finally arrived at the
age where we will really be using computers for our
everyday informational needs.
Telenet Corporation, the operators of the world's
first public packet-switched communications network, is
scheduled to offer public mailbox service early this
year. In September, Telenet Corp., recently spun of
from Bolt, Beranek and Newman, prepared a stock
offering to go public. Most of the proceeds of the
stock issue will be to expand Telenet's capacities to
handle the new services.
↑↑
*Richard Jay Solomon, a packet user, wrote this article from
his farm-office via a time-sharing system connected to a
Telenet port within convenient telephone range of his rural
Massachusetts town. Mr. Solomon is a Research Fellow at
Harvard University's Program on Information Resources
Policy, Cambridge, Mass., specializing in communications
issues.
!TELENET. Page 2
Mini-Micro Systems/Richard J. Solomon/
Until recently, access to these futuristic data
networks had been restricted to a small community of
computer researchers or large corporations with
in-house data networks. Within the past two years, the
application of packet-switching techniques pioneered in
the Defense Department's ARPAnet has made it possible
for the general public to enjoy the revolution in data
communications.
Electronic mail, interactive teleconferencing,
access to dozens of on-line data banks from law to
medicine, and other forms of dispersed processing
eventually might have evolved without the minicomputer.
Minis have speeded up this evolution by permitting a
totally new form of communications switching--the
packet switch. Coupled with digital transmission
methods, using landlines and via satellites, this has
drastically lowered the costs of data transfers to
potentially a miniscule part of processing costs.
Larry Roberts, the Chairman of Telenet, and
formerly Director of Information Processing for the
Advanced Research Projects Agency (ARPA) of DOD,
recently pointed out that "experience with the ARPAnet
has demonstrated that computing service can be obtained
remotely through a computer network at one-third the
cost of a local dedicated system. To obtain this
factor-of-three cost reduction, however, one must have
available a highly responsive and reliable
communications system capable of handling the peak data
rate of burst traffic normally associated with computer
usage. In order to provide such an enhanced
communications service," he emphasized, "it is
necessary to embed significant computational and
logical capabilties within the network itself.
Packet switching has made the telecommunications
network an integral part of the data-processing system.
It is no longer clear where processing begins and
communications ends since the "switch"--a mini
programmed to perform as a front-end processor--does
not resemble either the old-fashioned telephone relay
(space-division switch) or the latest electronic
switching systems which use time-division and digital
transmission techniques. In packet switched networks,
the data is broken up into short blocks, which could
theoretically range from one bit plus addressing and
synchronization data to very long bit streams. Each
block or "packet" is interleaved with other blocks of
data from the interface processor or from other
processors down the net. At each minicomputer node,
all of the packets are examined; those which are
destined for host computers or terminals connected to
the local packet switch or node are removed from the
!TELENET. Page 3
Mini-Micro Systems/Richard J. Solomon/
network stream. All of the other packets are
retransmitted to the next node down the line, though in
a real life network, the mini will perform other
complex routing and traffic-flow monitoring tasks.
Due to this technique, transmission capacity need
be allocated only when data is actually being sent.
While there are parallels to this concept in
long-distance telephone transmission, the packet switch
is particularly attractive for computer-communications
since it avoids all of the conventional intermediate
switching nodes found on the telephone network, save
the local loop. The path of each packet is not
determined in advance, but is selected continuoulsy to
ensure rapid and reliable data transmission. Line or
equipment outages or traffic congestion can therefore
be readily bypassed. Packets are automatically checked
at each switch or node, and retransmitted if necessary.
The complex routings and re-formatting at the
destination nodes, as well as error control, all take a
fraction of a second and is invisible to the user. In
addition, the packet interface processors have the
ability to convert to different terminal codes, speeds
and other protocols, if necessary.
This is but a simplistic description of the packet
switching technique, often compared to a "hot potato"
network, where the minicomputer node barely gets to
hang onto the burst of data before "tossing" it to some
other node or terminal. The concept was originally
broached by Paul Baran, now of the University of
Southern California, about fifteen years ago as part of
a Rand Corporation project. The first sophisticated
applications of the packet network was the ARPAnet. It
went online in the early 1970's, linking together major
research computers across the United States, in Hawaii,
and in Norway and England.
The packet switching concept was developed for
ARPAnet by Bolt Beranek and Newman, consulting
engineers of Cambridge, Mass. BBN implemented and
still maintains the ARPAnet for DOD. In 1975, BBN
began operating the Telenet Corporation's public
version of a packet switched network, principally for
remote terminal to host computer connections. The more
advanced electronic message and computer-to-computer
capabilities, demonstrated on the ARPAnet were built
into the Telenet protocols.*
THE ARPANET
!TELENET. Page 4
Mini-Micro Systems/Richard J. Solomon/
In order to understand the potentials of
sophisticated packet switching for enhancing a network
of minis or adding to the utility of distributed
processing using "smart" terminals, let us look at
grandaddy ARPAnet:
The ARPAnet is often misunderstood as merely a
digital link between large computers for file transfers
or for direct linking for large number-crunching
parallel processing. As such, its capabilities would
be quite amazing, as any programmer who has had to
interface with dissimilar operating systems
simultaneously can testify. In actuality, the links
themselves are not normally digital. Most of the local
telephone loops between the minis interfacing with the
host computers and the AT&T Long Lines network are
4800-baud Dataphone private lines. Major nodes are
connected with 56 kilobit per second LongqLines trunks.
The mystique of the powerful packet switching on
the ARPAnet comes from the data processing done by the
minis inconjuction with complex protocols programmed on
each of the host computers connected to the nodes. The
nodes are termed Interface Message Processors (IMPS)
and are mostly Digital Equipment Corporation PDP-11s.
Some ARPAnet nodes have the added function of being
accessible from outside the network by dialup or
dedicated Dataphone lines. These are called Terminal
Interface Processors (TIPS), and each contain a litany
of protocols and programs permitting almost any form of
low-speed terminal (to 600 or in rare caes, 1200 baud)
to interface with the network. Remote job entry, and
direct computer interface is also standard practice
with most ARPAnet TIPS.
Using the ARPAnet for computer-to-computer
linkages ranges from simple tasks which could be
assigned one's secretary--such as mail and message
functions--to highly complex data transfers and
re-formatting between mini and very large computers
with quite dissimilar operating systems. The net's
protocols are constantly undergoing modifications in
order to standardize commands and formats.
In essence the ARPAnet will remain a large
research tool for communications networks. Attempts to
commercialize its design features appear to have been
ruled out in favor of starting from scratch in the
design of Telenet and its competitors--notably
Tymshare's Tymnet which went on-line as a packet
network last year, and the proposed packet networks of
ITT Domestic Transmission Systems (ITTDTS), a
↑↑ *Telenet is not to be confused with the "telnet"--the
name of the ARPAnet protocol.
!TELENET. Page 5
Mini-Micro Systems/Richard J. Solomon/
subsidiary of the International Telephone and Telegraph
Co., and Graphnet, a subsidiary of Graphic Scanning
Corp. ITTDTS and Graphnet will offer digital facsimile
as well as computer data transfer.
Telenet's network consists of Class 1 Central
Offices which are major switching centers, each
utilizing numerous minicomputers connected to multiple
high-speed digital links supplied by AT&T Long Lines.
The Class 1 centers are connected to Class 2 offices
which serve as concentrators for other Telent
Processors, multiplexers from customer access
equipment, and foreign exchange telephone lines from
cities with local network access at standard tariffs,
but without direct Telent facilties as yet. (see map)
The Network control Center is in Washington, D. C.
All transmission facilties as of Septemeber, 1977
are leased from AT&T or other telphone companies. It
is Telenet's policy to concentrate its capital and
expertise in the packet operation rather than to be coe
s Specialized Common Carrier. The Federal
Communications Commission terms Telenet and Tymshare
"Value Added Networks", a curioushybrid which gets the
priviledge of being regulated with little of the claims
to monopoly status often bestowed on regular carriers
based on their high capital investment in transmission
facilties.
Telenet currently purchases all of its equipment
from outside suppliers. Prime Computer, Inc., of
Framingham, Mass., supplies the minicomputers used on
the network. Digital Communications Corporation
supplies the proprietary Telenet processors developed
jointly between Telenet and DCC.
Telene's revenues in comparable periods for 1976
and 1977 have more than doubled, while losses, running
upwards of 3$3.7 million per year, are increasing at a
fairly slow rate. In its recent proposed porspectus,
no estimate is made as to when Telent will become
profitable, however it is noted thatit intends to plow
back into theoperations as much of its revenues as
possible. From the proceeds of the propoed stock
offering, $2 million is allocated for additional
switching minicomputers and Telenet processors, and an
additional $4 million for equipment to come from
revenues and danticipated debt financing. Transmission
and network operations in 1978 is expected to cost $6.9
million, systems development $1.2 million, and
marketing and administration, $5.3 million. With debt
service, Telenet's total budget for 1978 is expected to
be $21.4 million.
!TELENET. Page 6
Mini-Micro Systems/Richard J. Solomon/
While Telenet's system is designed for direct
connection between dissimilar computers, the network is
currently used almost exclusively for terminal to
computer communications. Dialup ports are limited to
1200 baud, but higher speeds have been proposed.
For computer-to-computer data transfers, Telenet
can currently support 56 kilobaud rates. Protocols,
like its prototype ARPAnet will facilitate transmission
between separate customers such as would be required if
electronic funds transfer or automated check and credit
clearinghouses are established. Telenet's application
before the FCC to offer electronic mail services
including store and forward and code translation was
temporarily approved for a six-month period from last
September. It had been opposed by Tymshare's Tymnet
and the U. S. Postal Service as well as the more
traditional electronic message carrier, Western Union,
and the international record carriers.
In a related move, Digital Broadcasting Corp., of
Vienna, Va., has proposed the establishment of an FM
one-way broadcast system connected to Telenet for
transmitting messages to their final destinations by
avoiding the local telephone loop. As envisioned by
Digital Broadcasting, low-cost CRT's or printing
terminals (ranging from $10 to $25 per month on a
lease) would receive packets transmitted over FM radio
station subcarriers (similar to Musak's restaurant
background music) or via power lines in high-density
building complexes. Terminals would be equipped with
bubble memories and microprocessors developed
exclusively for Digital Broadcasting by an undisclosed
firm in Japan. No modems, data sets or telephone
connections would be necessary to receive.
A more elaborate packet-radio experiment, using
bi-directional transmissions, is under way in the San
Francisco Bay area by Stanford Research Institute.
Funded by ARPA, and connected to one of SRI's ARPAnet
IMPS, this experimental data distribution system is
intended to test the emergency applications of a
non-wireline alternative to the standard telephone
system. It can support up to 1200 baud CRT terminals
and a range of mini and microprocessors.
In case you have been wondering where Ma Bell has
been during all of these innovations, we must point out
thatin the past AT&T has been severely restricted from
making explicit offerings of data communications
services. Since the 1956 Consent Decree by Western
Electric, AT&T's wholly-owned manufacturing subsidiary,
the Telephone Company must stay with regulated
common-carrier services.
!TELENET. Page 7
Mini-Micro Systems/Richard J. Solomon/
Data processing, until the advent of the VAC's,
was not considered a regulated business.
Bell has not been asleep, as some of its
detractors have foolishly surmised. Bell now offers a
Digital Dataphone Service in 23 metropolitan areas.
DDS is capable of being upgraded to Switched DDS,
should the FCC permit. This would put Bell in direct
competition with the Datadial offering of Southern
Pacific Communications, and the Execunet service of
MCI, both Specialized Common Carriers with their own
communications networks, and potential competitors of
Telenet in offering data communications and mailbox
services. (See MINI MICRO SYSTEMS March, 1976, for a
cost comparison between Telenet and Datadial, when the
latter was under the defunct Datran.)
One step beyond SDDS is the proposed Bell Packet
Network. This could place virtually every local
telephone exchange on a nationwide packet switch. One
would merely dial a local phone number, coded,
presumably, to indicate transmission speed, terminal
type, and other bells and whistles. (Pardon the pun.)
One must realize that a properly buffered local loop
(wire pair) can handle at least 4800 baud to the
exchange, and probably up to 9600 baud, with some error
correction. It would not be difficult to provide each
and every local telephone exchange with a minicomputer
to handle the necessary packet protocols.
Transmission from the mini beyond would be on the
digital linkages being rapidly installed between
exchanges and on intercity trunks. There are about 550
telephone exchanges in the United States. We leave it
to the reader to calculate how much would have to be
invested by Bell, and how long it would take to install
such a network.
The future of packet data switching is not to be
found in the hands of the system designers, nor the
carriers. The essential arena will be either at the
FCC, the Justice Department's Antitrust Division, or
the Congress where guidelines will have to be
established to define the market for free or
constricted entry into data communications. It is
worth noting that IBM--who will launch with its
partners Aetna and COMSAT, Satellite Business Systems's
satcom in 1979--has recently suggested that all parties
would best be served with unrestricted entry into the
datacomm business. They even include AT&T in that
proposal.
!TELENET. Page 8
Mini-Micro Systems/Richard J. Solomon/
Could IBM be conceding the race between
distributed mini-processing and low-cost communications
as well?
∂07-Feb-78 0812 JMB shuey address, here it is
Dr. Richard L. Shuey
GE Research and Development Center
Schenectady, NY 12305
(518) 385-8379
....Jeff
∂07-Feb-78 2113 LES Pub problem
The problem in NAMES[1,VEW] was that some of the "blank" lines contained
one or more spaces. I think I eliminated them all.
It needs genuine blankness to work.
∂07-Feb-78 2330 DCO STRING-MATCHING
To: JMC, DCL, RWW, ZM
I don't know if anyone is interested in the long open question of the
decidability of the string-matching equality problem (deciding if two strings,
containing only constant symbols and variables ranging over strings, can be
equivalent), but, if so, I have just received the roughly 100 page proof of its
decidability. Russian author, translated. I am going to make a copy for
Kreisel - anyone else want one?
∂08-Feb-78 1002 PAT meeting
Calfee would like to meet with Knoles(?), Freedman and you on Tuesday, 14th at
10:00 in his office. I've put it in your calendar.
I presume Knoles is really Ellen Nold.
∂08-Feb-78 1116 FTP:Levin at SUMEX-AIM LOTS facility
Date: 8 Feb 1978 1116-PST
From: Levin at SUMEX-AIM
Subject: LOTS facility
To: jmc at SAIL
John:
Ed is having a conference Feb. 23...group is the Health Care
Technology Center from U. of Missouri (Dr. Lindberg). They
would like a mini-tour of LOTS on the 23rd, late afternoon.
Is that possible? What arrangements are necessary?...thanks...laurie
-------
I suggest you call Queenette Baur, the LOTS secretary, to arrange it.
Mike Byron or J.Q. Johnson could probably do it.
∂08-Feb-78 1148 RPG System meeting
To: PRNOID.DIS[P,DOC]:
Don't forget to attend to system meeting wrt Protection/Privileges.
∂08-Feb-78 1548 FTP:Levin at SUMEX-AIM knuth letters for manna
Date: 8 Feb 1978 1540-PST
From: Levin at SUMEX-AIM
Subject: knuth letters for manna
To: jmc at SAIL
john: i don't know what letters you're talking about.
Sorpy, i need some more info...laurie
The A and P committee has a @→SYJA=LAYKQiKef↓iQCh↓QCmJ↓EKK\↓eKGK%mKH~)G←]G∃e]S]≤A≠C]9BOfAAe←a←MKHACAa←S]Q[K]h8@Aπ←ASKfA]KeJAMK]hAQ↑AgK9S←d~)MCGk1irAS8AeKg%IK]G∀\@Aπ=aSKf↓gQ←k1HACYM↑AO↑↓i↑A↔9kiP\4∀≡`p5
KDZ\p@@dDdp∪ ∃⊗@@∪ CgWKQh@@~))↑t@AMKS≥K]ECUZACh↓'+≠ `[β∪~0A∃≠ε@@@~)CMiKHAgKe%←kfA
←]gS⊃KeCi%←\AR↓CZA←_A←aS9S←\AQQChA→←eKgPAECg-Kih~)SfAi=↑Ak]%ckJA¬]HAm¬YkCE1JABAAKeg←8Ai↑A1←gJ\↓RAMCY←dAi!JASI∃BA←L4∃[CW%]NAQ%ZACgM←GSCQJAae=LAoSQPAiK9keJXj`Zj@AS\A
fAC]⊂AKJ\4∃oJA]SYXA!CmJAAe←EC YrAi!JA←]1rAGf↓aae←_AS\AQQJAo=eYHA]Q↑AQ¬fAEk%Yh~∃¬\A←a∃eCiS9NAgsMiKZA¬fAoK1XACf↓iCkO!hACE=khAi!KZXA]Q↑AQ¬fAiQ∀AEKgP~∃CE%YSir↓←LAC9sE←IdARAW9←nAi<AISe∃GhAgQkIK]QfAS\↓gsgi∃ZAEk%YIS]≤X~∃C9HAoQ<AQCf↓BAg←1SHAi!K←eKQSGCX↓ECGW≥e←k]⊂AEKg%IKf\↓QJAQ¬f~∃C1g↑AiUe]KH↓←khA¬\AS]
eKIS YJA]U[EKd↓←LAO=←HAa!HOfA%\ABAMQ←eh↓iS[J8~∃RAMChAS8A←\A!SfAG1CggKLAgKm∃eCXAQS[Kf↓C]HA]CfAS5aeKgMKHAEdAiQJ↓]Kn~)[CiKISCXA!JAoCLAIKm∃Y←aS9NAM←HAiQJ↓G←keMJ\Ak9M←eiUC]iK1rAoJ↓Y←CI∃HAQS4~∃k`↓oSiP↓g↑A[¬]rAgQkIK]QfAQJ↓ISI]PAOKh↓i↑AoISiJA∧AE←←,A←\AQQJAgUEUKGPX~∃EUhARA KYSKe this hasn't hurt his reputation too badly.
so the main thing to do now is start soliciting letters of recommendation
for him.
i guess the way to get his appointment joint with ee when they have no
slots is to make mccluskey 50-50 also. that is fine with me,
since i always thought he was 50-50. besides, last year i had occasion
to study his textbooks carefully and i was impressed at the masterful
exposition in them. i also feel that he has gotten into the
swing of contemporary circuit design now although for some years
he was perhaps too locked in to older style switching theory...
that probably isnt a fair criticism judging by the excellent students
he has turned out. in summary, i see a better future in having forest
50-50 for best future relationship between cs and ee,
and i think we should go all out to keep him at stanford.
incidentally, as you may guess, i have spoken with linvill shortly
before writing this note! but i found his comments convincing; in my
earlier conversation with jmc i doubted that ee
was billetless, because of two slots i didnt know they had filled.
the new situation in which retirements don't apparently
start at 65 means that the number of new slots opening up is happening
slower than it appeared last year it would.
forests other offer is from ut austin, according to linvill. lets not
lose him please! even during my sabbatical i will take time to write
a strong letter for his file.
Laurie:
Please print out this note from Don Knuth and see that Bob Floyd
gets a copy. He doesn't log in here often enough for direct
transmission to be reliable.
∂08-Feb-78 2234 DCL
To: GROUP.DIS[VCG,DCL]:
We are considering going back to our old time Tuesdays 2.30 for
the verification group seminar. This would allow CGN to attend, and it would
allow all of us to attend AI lab systems meetings. Please send objections to
DCL
It would also allow me to attend.
∂09-Feb-78 0900 JMC*
call sacerdoti, haugeland
∂09-Feb-78 1441 FTP:Levinthal at SUMEX-AIM Advnaced automation
Date: 9 Feb 1978 1441-PST
From: Levinthal at SUMEX-AIM
Subject: Advnaced automation
To: tob at SU-AI
cc: jmc at SU-AI
Attached is a new version of my previous memo which I intend to
put before W. Miller in the very near future - possibly today or
tomorrow. The changes aren't substantive except for the suggested
change in name. Any comment? I thought it answered some of the
comments about the use of the word Automation and still left it
more general than Productivity . CDRCDR(SLAIDE) sublimily suggests
a friendly and useful objective.
-------
∂09-Feb-78 1457 FTP:Levinthal at SUMEX-AIM advanced automation(con)
Date: 9 Feb 1978 1457-PST
From: Levinthal at SUMEX-AIM
Subject: advanced automation(con)
To: tob at SU-AI
cc: jmc at SU-AI
MEMO TO FILE:
SUBJECT: SLAIDE
FROM: ELLIOTT LEVINTHAL
DATE: FEBRUARY 9, 1978
The proposal is to organize a Stanford Laboratory of
Artificial Intelligence Developments in Engineering (SLAIDE). The
focus would be applications of scene analysis and robotics science
to problems of advanced automation,industrial productivity,and
physical rehabilitaion.
The initial impetus for this proposal was the convergence
of interests of the group at Prof. John Mccarthy's Artificial
Intelligence Laboratory working with Dr. Tom Binford on vision and
robotics and our work on the Viking mission to Mars. This has
involved us in joint consideration of the problems of automated and
semi-automated three demensional terrain analysis both with regard
to the present problems of Viking data analysis and future planetary
NASA missions.
Discussions of this proposal have been held with Prof. Edward
Feigenbaum and Prof. John McCarthy and it has their support. Prof.
Feigenbaum pointed to the successful collaboration already taking
place between his Department and the Department of Electrical Engin-
eering and which developed an academic program leading to a Master's
Degree. In addition discussions have been held with Prof. William
Reynolds, Department Chairman, Mechanical Engeering, and his
colleagues Professors DeBra, Piziali, and Leifer. Significant
interest was expressed in joint activities.
To make important contributions, such a laboratory would need
expertise in computer science, with an emphasis on vision research
in particular and artifical intelligence in general, mechanical
engineering with an emphasis on robotics, and electrical engineering
focusing on sensors and microcircuitry. To distinguish its function
from the basic activities of the individual departments, the
Laboratory would seek support in a few important problem areas that
require for their solution the synthesis of state-of-the-art skills
in relevant fields.
Among the problem areas that might be condidates for research
and support are the following:
1) NASA - Large Space Systems Technology, Teleoperators,
Manipulators, Fabrication in Space,
Planetary Rovers and Return Sample Missions
Image Analysis for Earth and Planetary Missions
2) DOD Defense Mapping Agency,
Automated photogrammetry and map making, DOD applications
of remote manipulators and scene analysis
3) ERDA - Telemanipulator repair, semi-automated mining (underground
or undersea)
4) HEW - Physical Rehabilitation, prosthetics, "seeing-eye dog"
automated 'servant' for disabled or bed-ridden convalescents.
(Note: The School of Medicine has recently voted to
establish a Department of Rehabilitation Medicine and it is
possible that an Enginering Center will be supported by the
Government at PAVAH.)
5) Industry -robotics for assembly line use and large construction
projcts.
-------
∂09-Feb-78 1722 DPB
February 9, 1978
To: Computer Science Department Faculty and Advisers
From: Denny Brown
Associate Chairman
Subject: Gray Tuesday Meeting
The purpose of this memo is to remind you of the Gray Tuesday meeting
which will be held on Tuesday, Feb. 14 at 1:15 in Polya 204. If all goes
well we should be able to complete this annual review of the progress of
our PhD students by 3 pm. Please be prompt.
Below is a list of your advisees according to our records. The list
includes MS and CSCE students for your information. (These will not be
reviewed Tuesday.) The list includes each PhD student's quarter of entry
and status according to current guidelines.
Also enclosed is my memo of March 1, 1977 which describes the current
policy regarding reasonable progress of a PhD student.
------------------------------------------------------------------------
McCarthy John Total = 11 PhDs = 8 MSs = 3 CSCEs = 0
PhD Advisees Entry Current Status
Bulnes Juan a71 ((Needs to do Orals.))
Farmwald Paul a74 ((Needs working title.) (Needs Reading
Committee.))
Filman Robert E a76 (Candidacy required by June. Completed.)
Goad Christopher a75 ((Needs working title.) (Needs Reading
Committee.))
Moravec Hans a71 ((Needs to do Orals.))
Pattis Richard E a76 (Candidacy required by June. Completed.)
Robinson Andrew a76 ((Hasn't passed Comprehensive.))
Gray Tuesday is Feb.14, and this is what the CSD records think of you.
What's your plan.
Wilkins David a73 (Working on dissertation. Working title is:
Complete Problem Solving Using a Large Knowledge Base
Readers are: GREEN NILSSON)
MS Advisees Entry
Pettit Teri Kim a77
Agoston Max a75
Olumi Mohammed a76
------------------------------------------------------------------------
March 1, 1977
TO: CSD Ph.D. Students
SUBJECT: Evaluations
The CSD faculty met on February 1, 1977 to evaluate the
progress of each Ph.D. student. The following guidelines were
used in this evaluation:
Within the first two years, a student should:
(1) Pass the comprehensive examination
(2) Apply for candidacy
Within three years a student should:
(3) Pass area qualifying exam
Within about a year of passing the area qual, a student should have:
(4) A thesis advisor (chairman of reading committee)
(5) A thesis topic
(6) A reading committee
Subsequently,
(7) The thesis advisor and other members of the reading
committee should be able to report on evidence
of progress towards completion of the
dissertation.
Evaluations were three valued:
[white] Evidence of satisfactory progress
[gray] Insufficient evidence
[black] Evidence of unsatisfactory progress.
Additionally, suggestions were often made concerning particular
items for the student's attention.
The faculty will meet again in May to reevaluate students. Students
should provide evidence to their advisor of their continuing
progress.
∂09-Feb-78 1755 HPM Cloudy tuesday
∂09-Feb-78 1742 JMC
Gray Tuesday is Feb.14, and this is what the CSD records think of you.
What's your plan.
Moravec Hans a71 ((Needs to do Orals.))
[Am trying very hard (believe it or not) to get the obstacle avoider
running successfully within the next few months, and have started to write
up some of the system in preparation for serious writing after it runs.
I don't think trying to do orals would be a good idea before a first
draft is finished, but I will listen to recommendations.]
∂10-Feb-78 0807 BS David Harel
David Harel was here a short time ago, I understand at our invitation ,
regarding a possible faculty appointment. Bob Floyd told me this week
that we should give him some expense money or an honorarium for his talks.
The honorarium would be easier now since we have no expense receipts, etc.
Bob suggested $100 - $125. Is this o.k. with you? Incidentally, I have
been able to get expense money through the Dean's Office for such visits,
but I can't get it after the fact. They want to know in advance and most
of the time I don't know in advance.
Betty
Yes, it's ok with me. I suggest you send a message to all faculty about
the advantages of advance notice.
∂10-Feb-78 0958 BS Faculty Appointee Visits
In reply to your suggestion, I have reminded faculty before about letting
me know in advance about the visits of possible appointees. The reminder
hasn't done much good, but I'll try again.
Betty
∂10-Feb-78 1132 DEW My progress
To: JMC
CC: BERLINER at CMU-10A
10 February 1978
John,
In regard to your message which follows:
Gray Tuesday is Feb.14, and this is what the CSD records think of you.
What's your plan.
Wilkins David a73 (Working on dissertation. Working title is:
Complete Problem Solving Using a Large Knowledge Base
Readers are: GREEN NILSSON)
I'm not sure how much of a plan I need. I'll tell you what's happening
and you tell me what more I need.
I've been working closely with Hans Berliner at CMU who's been acting
as my adviser. About 90% of the development has been done on my
program and he and I are both satisfied with the results and agree
that it's time to write my thesis. I have just recently started outlining
chapters and will soon be writing.
There remains some developmental work. I also must do lots of testing and
gathering of statistics. This will take a lot of time since the program
is slow and our machine is too crowded for me to run anytime except late
at night.
If I get supported this summer and everything goes perfectly, the
earliest possible completion date will be the end of the summer.
However, I think a more reasonable estimate is the either the end of
Autumn quater 78 or winter quarter 79. If things do not go well
or computing time becomes even less available, then it might take as long
as spring 79 to finish, but I hope not.
Since I have just started writing, I would rather not specify my title or
reading committee for a month or two. However there seems to be a lot
of pressure to do this now. Tentatively here's my title though it will
probably change: "Using Patterns to Solve Problems and Control Search".
The title listed in the CSD records will defintely not be my title.
As for a reading committee, Hans Berliner will be on it and I have been
assuming you (JMC) will be also although I haven't talked to you about
it. For the third person, I intend to get either Bruce Buchanan or
Cordell Green. Both Cordell and Nils Nilsson have expressed willingness,
though I have no formal commitments. Perhaps Nils will be a fourth reader.
Let me know if anyone wants to know more.
David
That is sufficient information, and I have, of course, been expecting
to read the thesis.
∂10-Feb-78 1307 JB Black Tuesday and Orals.
I would like to have a brief talk with you about the question of my orals.
∂10-Feb-78 1441 JB
THESIS PROGRESS REPORT AND ORALS.
The code for the Goal Command Language has run successfully
by the beginning of this week and produced fully automatic proofs
of several theorems (in particular of the existence and unicity of
the ordered pair, which RWW had set as one of the main goals for
automatic theorem prooving in our GCL; also all of the theorems from the
beginning of Kelly's Appendix that I have been able to try have been
demonstrated fully automatically).
The code is complete but stills needs some clean up and debugging.
Unfortunately I am beeing delayed in my experimentation by bugs in the
present version of FOL, while a new version from RWW is expected within
a few days. My experiments rely heavily on the simplify code, which
has been recently updated by AMR and is to go into the new version of FOL.
PLAN: I expect to have debugged most of the code and have carried out
a very substantial number of experiments by the end of the month,
so as to begin writing the the thesis properly in March.
I expect the write up to go very smoothly, since it is by now clear
what the GCL does.
ORALS: I have so far preferred to postpone my orals because the very
writing of the thesis seems to be the best preparation for them.
So I shall do my orals in the Spring quarter.
∂10-Feb-78 1541 BS New employees
Thought I should let you know that we have two new staff members (clerical).
Bethanne Loncasty begins Monday as secretary to Bruce Buchanan. This is a
new position. Vicki Fahrenholz, our publications secretary, worked her last
full day today; she works three half days next week. Her replacement is
Suzanne Wieting (pronounced Weeting) and she also starts Monday.
Betty
∂11-Feb-78 0253 AMR
∂09-Feb-78 1748 JMC
Gray Tuesday is Feb.14, and this is what the CSD records think of you.
What's your plan.
Robinson Andrew a76 ((Hasn't passed Comprehensive.))
John : i am planning to take comps next time around
∂11-Feb-78 1236 REF Counter example [?] of circumscription.
Working from the file EXAMPLE[W78,JMC] as of 2/10/778 15:57
Imagine instead that the axioms for our system were
1) on(x,p,move(x,p,s)) ∧ on(A,C,move(B,C,s))
That is, things are on after they move. Additionally, the block
A is "glued" to the block C, and when C moves, A moves too.
2) We select the same P, and consider the schema:
3) Q(x)∧ (on(x,p,move(x,p,s))∧ on(A,C,move(B,C,s)) ⊃
∀y.(P(y)⊃Q(y))
4) With the selection of Q(y)≡ y=x.
Here's where the "lack of quantifiers" will get us into trouble.
"x" is subtley quantified over all blocks, but I think it will miss the
special case of A.
Substituting, we get:
5) x=x∧ (on(x,p,move(x,p,s))∧ on(A,C,move(B,C,s)) ⊃
∀y.∃q.¬(on(y q move(x p s))≡on(y p s))⊃y=x)
Now, the premise is obviously true from the initial axioms.
Hence, we have
6) ∀y.(¬y=x⊃ ∀q.((on(y q move(x p s))≡on(y p s))
But ¬ on(A,C,move(B,C,s))≡on(A,C,s)
if A wasn't on C in s (a reasonable supposition.
Comments?
∂11-Feb-78 1337 FTP:Levinthal at SUMEX-AIM Advanced Automation
Date: 11 Feb 1978 1335-PST
From: Levinthal at SUMEX-AIM
Subject: Advanced Automation
To: feigenbaum
cc: jmc at SU-AI, tob at SU-AI
I saw Miller yesterday and gave him a copy of of the latest version
of my memo. He was quite interested and supportive. He felt the
appointment problems could be resolved. He suggested that the next
step would be a meeting of the interested parties and himself and
G Leiberman.Prior to that I feel we need further intereaction
between ourselves on detailed questions of implementation;timing,
location,space,appointments,remaining issues,if any, of philosophy
and goals. It would be nice, but not essential, to have accomplished
as much of above as possible before Tom and I go to Washington in
the beginning of March. It would enhance our abilities to seek
research opportunities. Ed perhaps your secretary could try to
find a suitable time and place for us to met?
-------
∂11-Feb-78 1457 DCL
To: GROUP.DIS[VCG,DCL]:
***********************************************************************
VERIFICATION GROUP SEMINAR TUESDAY 21ST. FEBRUARY
PLACE: ERL 237
TIME: 2.30 pm
TITLE: Verification in Languages with the Class Construct
SPEAKER: David Luckham and Wolf Polak
This is the second part of our seminar on the Class construct and Wolf will
present examples hopefully over and beyond what other people have been
able to do so far.
PLEASE NOTE: We have gone back to TUESDAYS AT 2.30.
There is no seminar next week.
***************************************************************************
∂11-Feb-78 1749 REF Notes On Situations and possible worlds.
There exists in our epistemological axiomatizations a fundamental
confusion between the "possible world" notion and the "situation" notion.
Each, it seems, is to express a notion about a possible reality. It is
often unclear which should be employed for a given problem domain.
In my understanding, the relevent difference lies in the
possession of the possible world of a "history", detailing how it was
reached. The situation, on the other hand (at least as I am defining it)
is merely an expression of the current status of the problem actors.
Primarily, the possible world is an appropriate vehicle for reasoning
about what must have happened to reach a given point; the situation, on
the other hand, is useful for "proceeding" from that point into the
"future" (most systems are independent of their past histories for future
actions). Thus, problem statements of the form "what must have happened
... " lend themselves to the possible worlds approach; questions such as
"how can x be achieved" are more easily handled by the situation
formalism.
Note that if we seek to prove that x can never occur, it is useful
to be able to assume a possible world containing x, reached from our given
situation, and to reason about the various steps involved in reaching that
presumed world, seeking a contradiction.
Now, situations and possible worlds, as I have defined them, are
kin but not identical. One important difference is that the situation of
a possible world is always derivable by examining that possible world,
whereas often many possible worlds could have happened to reach a given
situation. They are similar in many of the uses we seek of them; in the
typical problem axioms we will often have predicates which extract
information from one of these state vectors. In the blocks world, the
predicate ON serves this purpose; in the chess world, POS and VALUEON have
this relationship.
Now, it is useful if one's histories `begin' somewhere. In the
chess world, we have a precise origin, the initial chess position, for all
of our questions. Possession of this origin permits a more powerful form
of induction. The blocks world, however, does not have this fixed point;
rather, in any problem statement that wishes to reason about sequences of
actions using the historical approach must call some situation "basic",
and reason from there.
Possessing a possible-world/historical state has three primary
advantages. First, the historical state lends itself to a natural form of
induction schema (for example, chess induction and "reachable" induction).
Second, historical states are ideal for reasoning about what must have
happened to reach this current situation. Lastly, the historical state is
a good format for expressing a sequence of actions. Such states become a
sort of branch-free program (which do, however, admit a useful form of
variable binding - universal quantification). The branches for such
systems might, more easily, be included in the formal language.
Using only the current status situation provides a formalism which
is easier to reason forward from. Additionally, the important properties
of the situational vector can often be expressed neatly in the semantic
model; for example, the chess eye, CLT's blind robot axiomatization, or
the obvious (and soon to be done) attachments to the missionaries/
cannibals problem.
Carrying in the historical formalism into the future produces a
strange sort of creature, as the discussion of Berliner's problem in the
chess paper suggests. One obtains an historical state, about which one
knows certain properties (and which, fortunately, are the only access to
this state) but which lacks the clean history one expects of an historical
vector. We are able to state that "predicate P is true about history H",
and that "If X did A two events ago, then Y did B (and X, C, then Y, D)"
but not which "happened". We are expressing a form of programming with
branches in this form. It works; it is, however, somewhat unsettling.
Events themselves are an interesting side point. Our problems
have dealt with simple, well defined events (moving a block, or a
chesspiece, or a robot). It is clear that these are inadequate for a more
general world, for they effect only things in discrete time. The
historical state vector has an advantage here; we may have need to speak
of "how this state vector was at time t".
When speaking of the real world, or perhaps of some problem world,
we may have occasion to label one particular possible world as "reality".
Thus, there is an historical state where Napoleon won Waterloo (in fact,
there are an infinite number of such states). However, it is not true in
"the reality" state that this happened.
Existence. When we speak of the existence of an object in a
possible worlds formalism, perhaps we should also be required to speak of
the worlds we expect that existence to occupy. Much as the fluent
notation takes a typical predicate, say "ON", and permits its expression
as ON(x,y)(S), perhaps the notion of existence should allow ∃x
y.ON(x,y)(S), the ∃x y.ON(x,y) being a kind of fluent. Statements which
purport an existence without mentioning a state presumably purport that
existence for all states.
∂11-Feb-78 2122 REF Glueing
However about the glueing axiom stating:
∀z x y s.((GLUED(z,x)∧ON(x,y,s))⊃ON(z,y,s))
it would seem to be a somewhat tougher nut.
Thanks, I'll try that example, but perhaps you'll get it first.
∂12-Feb-78 1604 FTP:HANS BERLINER(C410HB03) at CMU-10A Dave Wilkins's progress
Date: 12 Feb 1978 1900-EST
Sender: HANS.BERLINER at CMU-10A
Subject: Dave Wilkins's progress
From: HANS BERLINER(C410HB03) at CMU-10A
To: jmc at SU-AI
- - - -
John,
I have been working with Dave rather closely over the last six weeks. He
is 85 to 90% complete with his programming and about 40% complete with his
testing of his program. He has just begun to do his writing. I think that
he should be finished in about 8 months; most of the effort over that time
going to writing. I don't really know much about how he writes so that
will be a factor in the exact amount of time. His other work is excellent.
I feel that he ha found ome fine general method of handling problems and
that his thesis will be the best example yet of how to use knowledge to
control and minimize search.
One other matter: I feel that I will have to get a personal look at his
program's performance one of these days (as of now I only see output of
runs that he has conducted). This is necessary in my mind in order to be
sure it responds as well as I think it does, and possibly to try some
things that would be difficult to formulate in conversations over the
ARPANET about a protocol that the program did several days earlier.
Therefore, I would like to come out to Stanford and work with him for
aobut a week some time, probably aout 2 months from now being best. CMU
will not contribute any funds toward such a trip as they feel that they
are contributing my time as it is. Therefore, I would like to know how you
feel about this, and whether funding can be secured from Stanford. I think
you can see that it is very important if I am to be the technical director
of his work, that I have some close-up work with him. Ideally, it would be
nice to come out for his oral too, but that would involve another trip,
and I will wait to see what you think about all this before going any
further.
Hans
-------
Yes, we will pay for a trip of a week's duration. It would help
spread the cost if you would give a talk about some AI topic. I agree
that it would be good if you came for the oral too, and we'll see what
we can do.
∂12-Feb-78 1811 JBR
Vera caled, said she'd be home before 8pm.
∂12-Feb-78 1835 FTP:Geoff at SRI-KA (Geoffrey S. Goodfellow) 'Computer Reveltion'
Date: 12 Feb 1978 1833-PST
Sender: GEOFF at SRI-KA
Subject: 'Computer Reveltion'
From: Geoff at SRI-KA (Geoffrey S. Goodfellow)
To: JMC at SAIL
Message-ID: <[SRI-KA]12-Feb-78 18:33:37.GEOFF>
Ch. 7 will have something each night this upcoming week all
about computers in our lives. Should be interesting to see how
the news handles this report and how they will make THEM vs. US
look.
-------
∂13-Feb-78 1236 FTP:HANS BERLINER(C410HB03) at CMU-10A trip and talk
Date: 13 Feb 1978 1532-EST
Sender: HANS.BERLINER at CMU-10A
Subject: trip and talk
From: HANS BERLINER(C410HB03) at CMU-10A
To: jmc at SU-AI
- - - -
John,
I think one of the first 3 weeks of May would be a good time to come out.
I have a talk I can give on a NEW (I really mean new) tree search algorithm
I have discovered for 2 person games. If you want to schedule me, I will then
plan the trip around that date, else I will leave it open and have the time
dictated by Wilkins' progress.
Hans
-------
∂13-Feb-78 1714 JBR
You have exceeded your disk quota.
The files listed below have been purged to reduce your disk
area to your quota of 2000
Before purging, your files occupied 2993
You have too many files. The purger may not select the
optimum set.
PPSAV.TMP[ESS,JMC]
CRYPT.DMP[ 2,JMC]
CODE.DMP[ 2,JMC]
SOURC2.LAP[ 1,JMC]
HONIG.XGP[LET,JMC]
MINKER.XGP[LET,JMC]
AIZERM.XGP[LET,JMC]
LISP.XGP[F77,JMC]
NAMES.XGP[ESS,JMC]
SHUEY.XGP[ESS,JMC]
AKEMAN.XGP[LET,JMC]
CARTWR.XGP[LET,JMC]
ACKEMA.XGP[LET,JMC]
KEPLIN.XGP[LET,JMC]
MASSY.XGP[LET,JMC]
SUZMAN.XGP[LET,JMC]
JAPANT.XGP[LET,JMC]
SUNDES.XGP[SEN,JMC]
REPRES.PRO[ 1,JMC]
EPIS[ 1,JMC]
FUNS[ 1,JMC]
P1[ 1,JMC]
PART[ 1,JMC]
PERM[ 1,JMC]
PATH[ 1,JMC]
PATH2[ 1,JMC]
ANTIN[ 1,JMC]
SYLL[ 1,JMC]
MEET[ 1,JMC]
COMPIL[ 1,JMC]
TIMES[ 1,JMC]
TESTA.SAI[ 1,JMC]
TESTB.SAI[ 1,JMC]
TESTC.SAI[ 1,JMC]
TESTD.SAI[ 1,JMC]
DADDA[ 1,JMC]
PROB1[ 1,JMC]
ORDIN[ 1,JMC]
ROTA.SAI[ 1,JMC]
ROTAT.SAI[ 1,JMC]
ROTB.SAI[ 1,JMC]
ROTC.SAI[ 1,JMC]
SEMAA[ 1,JMC]
INTER[ 1,JMC]
SEMAB[ 1,JMC]
CONVER.SAI[ 1,JMC]
RELREP[ 1,JMC]
MC[245,JMC]
PATH[245,JMC]
PATH2[245,JMC]
DEMO[ 1,JMC]
LISPAD[ 1,JMC]
TESTE.SAI[ 1,JMC]
ICOM[ 1,JMC]
TRANS1[ 1,JMC]
SIMUL[ 1,JMC]
COMPU2[ 1,JMC]
SOURCE[ 1,JMC]
SOURC2[ 1,JMC]
PUZZ.SAI[ 1,JMC]
TCLUB[ 1,JMC]
TCLUBA[ 1,JMC]
NEWDOC[ 1,JMC]
BOARDS.SAI[ 1,JMC]
REFER.ENC[225,JMC]
MCCRAC.LET[ 1,JMC]
PATHS.RLS[225,JMC]
PARKER[ 1,JMC]
GRPALG.RLS[225,JMC]
GRPDAT.RLS[225,JMC]
GROPER.RLS[225,JMC]
DIMEN.RLS[225,JMC]
CHAR.RLS[225,JMC]
LARRY2[ 1,JMC]
COMPU2.LET[ 1,JMC]
DEG5.IN[225,JMC]
S4.REP[225,JMC]
REPRES.RLS[225,JMC]
DEG6.IN[225,JMC]
R42.IN[225,JMC]
ANNOUN[225,JMC]
PUZZ.SAI[225,JMC]
PUZZA.SAI[225,JMC]
PANIC.SOS[225,JMC]
PUZZB.SAI[225,JMC]
ENCYC1.PRO[ESS,JMC]
PUZZE.SAI[225,JMC]
MINE.DIR[ AI,JMC]
PTCH2.DIR[ AI,JMC]
DRIVE.DIR[ AI,JMC]
KNOW2.AI[ESS,JMC]
ARTNA1.ART[ESS,JMC]
RECOG.LET[ 1,JMC]
MEMMTC.QUA[ESS,JMC]
LANG1.AI[ESS,JMC]
S5.QUA[ESS,JMC]
MTC71.QUA[ESS,JMC]
R1303.ART[ESS,JMC]
R1301.ART[ESS,JMC]
N30[ 1,JMC]
QA.PRO[ESS,JMC]
ARTNAT.ART[ESS,JMC]
SEMAN.AI[ESS,JMC]
FRANK.STA[ESS,JMC]
MTCSYL.QUA[ESS,JMC]
COMPUT.STA[ESS,JMC]
AISY69.QUA[ESS,JMC]
FILEC.FAI[ESS,JMC]
LETTVI.REV[ESS,JMC]
R1300.ART[ESS,JMC]
LCFMEM.MTC[ESS,JMC]
PEARL.NOT[ESS,JMC]
BACKSL[ 1,JMC]
NETJO2.PRO[ESS,JMC]
IPT73.REP[ESS,JMC]
NSFX.2[ESS,JMC]
NSFX.6[ESS,JMC]
NSFX.5[ESS,JMC]
NSFX.7[ESS,JMC]
NSFX.9[ESS,JMC]
NSFX.8[ESS,JMC]
NSFX.4[ESS,JMC]
NSFX.3[ESS,JMC]
HAYES.FRM[ESS,JMC]
ADTEL2.NOT[ESS,JMC]
PLATT.NOT[ESS,JMC]
TASK2.MAR[ESS,JMC]
TASK1.MAR[ESS,JMC]
LETTER.HE2[ESS,JMC]
XLET[ESS,JMC]
ZF.AX[ESS,JMC]
FOO73A[ESS,JMC]
FOON.PR1[ESS,JMC]
ROTE.SAI[ 1,JMC]
ROTD.SAI[ 1,JMC]
ROTF.SAI[ 1,JMC]
SWOPSI.RE[ESS,JMC]
HART.ME1[ESS,JMC]
GGGG[ESS,JMC]
PRO[ESS,JMC]
M29AUG.MES[ESS,JMC]
SYMPOS[ESS,JMC]
CH1.BH2[ESS,JMC]
CLOCK.FAI[ESS,JMC]
SLIDES[ESS,JMC]
FACMIN.MEM[ 1,JMC]
FOO1[ESS,JMC]
WUTHER.QUE[ESS,JMC]
ROY[LET,JMC]
TOLLES.LET[LET,JMC]
FORDH.LET[LET,JMC]
AD.NET[ESS,JMC]
COMPUT[ 1,JMC]
KNUTH.SAI[225,JMC]
TUCKER.LET[LET,JMC]
TUCKER.LE2[LET,JMC]
TOLLES.LE1[LET,JMC]
TCLUB.MEM[ESS,JMC]
PUTCH[ AI,JMC]
PUTCH.DIR[ AI,JMC]
DRIVE[ AI,JMC]
AIBIB[ AI,JMC]
MINE[ AI,JMC]
NOTES2[ AI,JMC]
PUTCH3[ AI,JMC]
RUSS[ AI,JMC]
CHYLIN.LE1[ AI,JMC]
GEN1[ AI,JMC]
DRIVE3[ AI,JMC]
MPRO[ AI,JMC]
SLAMEC.LE2[LET,JMC]
BOWDEN.LE1[LET,JMC]
CHERN.LE2[LET,JMC]
MAY.ME[LET,JMC]
NSFX.10[ESS,JMC]
HENRY[ESS,JMC]
MC.AI[ESS,JMC]
BIRKHO.LE1[LET,JMC]
FLANAG.LE1[LET,JMC]
ERIK.LE6[LET,JMC]
TAMBOV[ 1,JMC]
BARAB[ 1,JMC]
HARPER.LE1[LET,JMC]
PUZZE.F4[225,JMC]
ERIK.LE7[LET,JMC]
BAJCSY.LE1[LET,JMC]
STARKM.LE1[LET,JMC]
MT[ 1,JMC]
BRIT.LET[LET,JMC]
IRANI.LE1[LET,JMC]
BOWEN.LE1[LET,JMC]
JAF.STA[ESS,JMC]
RECUR[ESS,JMC]
ANDERS.RE1[ 1,JMC]
OLSEN.LE1[LET,JMC]
ERIK.LE0[LET,JMC]
SUBCOM.LE1[LET,JMC]
MICHIE.LE2[LET,JMC]
NEWSCI.LE1[LET,JMC]
AHLHAU.LE1[LET,JMC]
MEMOS.CUR[LET,JMC]
PUBINT.LE1[LET,JMC]
HORNIN.LE1[LET,JMC]
FELDST.LE1[LET,JMC]
CHECK.DEP[LET,JMC]
TUCKER.LE1[LET,JMC]
SCANDU.LE1[LET,JMC]
MELTZ.LE1[LET,JMC]
PIASET.LE1[LET,JMC]
MILNER.REC[LET,JMC]
FLESS.LE1[LET,JMC]
TRESSL.LE1[LET,JMC]
PIASET.LE2[LET,JMC]
LICKL.LE1[LET,JMC]
COOP.LE1[LET,JMC]
POE.LE1[LET,JMC]
JMILLE.LE2[LET,JMC]
IGARAS.LE1[LET,JMC]
ITO.LE1[LET,JMC]
CLARK.LE1[LET,JMC]
FLESS.LE2[LET,JMC]
MICHIE.LE3[LET,JMC]
TAVARE.LE1[LET,JMC]
SCIENC.LE1[LET,JMC]
SCRIB.LE1[LET,JMC]
ATKIN.LE1[LET,JMC]
KOCHEN.LE1[LET,JMC]
GREER.FAC[LET,JMC]
CU.LE1[LET,JMC]
RALSTO.LE2[LET,JMC]
MESSAG.SCI[LET,JMC]
MCCRAC.LE2[LET,JMC]
RALSTO.LE3[LET,JMC]
CHRON.LE1[LET,JMC]
LOGIC.QUA[LET,JMC]
SHIGER.LE1[LET,JMC]
PAMELA.LE1[LET,JMC]
TAUGNE.LE1[LET,JMC]
MORGAN.LE2[LET,JMC]
FROOK.LE1[LET,JMC]
LOMONA.LE1[LET,JMC]
APR.ME[LET,JMC]
LOGIC.PUB[LET,JMC]
CONSUL.LE1[LET,JMC]
VERE.LE1[LET,JMC]
PAWLAK.LE1[LET,JMC]
TSUDA.LE1[LET,JMC]
NOLL.LE1[LET,JMC]
ROBERT.LE1[LET,JMC]
RILI.LE1[LET,JMC]
RALSTO.LE1[LET,JMC]
MORGAN.LE1[LET,JMC]
MIYA.LE1[LET,JMC]
HEALTH.LE1[LET,JMC]
WILKS.LE1[LET,JMC]
LAST.LE1[LET,JMC]
SEDEL.LE1[LET,JMC]
CAMPB.LE1[LET,JMC]
ARDEN.LE1[LET,JMC]
BURNS.LE1[LET,JMC]
SHOCK.LE1[LET,JMC]
CU.LE2[LET,JMC]
DAVID.LE1[LET,JMC]
SCHOEL.LE1[LET,JMC]
CALHOU.LE1[LET,JMC]
FINCH.LE1[LET,JMC]
LYKOS.LE1[LET,JMC]
JMILLE.LE1[LET,JMC]
MITCH.LE1[LET,JMC]
JERRY.LE1[LET,JMC]
MCQUIL.LE1[LET,JMC]
PAM.LE1[LET,JMC]
DRLUK.LE1[LET,JMC]
OCT.ME[LET,JMC]
FORSHA.LE1[LET,JMC]
COMPT.LE1[LET,JMC]
STORK.LE1[LET,JMC]
ASTRO2.ART[ESS,JMC]
R1302.ART[ESS,JMC]
FRIEDM.LE1[LET,JMC]
FF[ESS,JMC]
AIRLIN.MTC[ESS,JMC]
JFOL.USE[ESS,JMC]
MEARS.LET[LET,JMC]
GRIMM.BKP[ESS,JMC]
SCIAM.LE1[LET,JMC]
NYT.LE1[LET,JMC]
NAVROZ.LE1[LET,JMC]
NOTES1[ AI,JMC]
TECH[ 1,JMC]
LISPN[ 1,JMC]
TXP1.LSP[ 1,JMC]
STARK.LE1[LET,JMC]
MDAVIS.LE1[LET,JMC]
IGARAS.LE2[LET,JMC]
REPRES[LET,JMC]
FLESS.LE0[LET,JMC]
KNOW.NOT[ AI,JMC]
GLUGG[LIT,JMC]
GLUGG2[LIT,JMC]
KNOW.ART[ AI,JMC]
MEIR.LE1[LET,JMC]
JAF.LE2[LET,JMC]
PIERCE[ 1,JMC]
MCCRAC.LE1[LET,JMC]
LARRY[ 1,JMC]
TELLER.LE1[LET,JMC]
PERSON[ 1,JMC]
YOUNG.LE1[LET,JMC]
KILPAT.LE1[LET,JMC]
JAF.STA[LET,JMC]
NAS.NOT[ESS,JMC]
CRYPT.FAI[ 2,JMC]
WEISSK.LE1[LET,JMC]
MICHIE.LE1[LET,JMC]
ONEILL.NOT[ESS,JMC]
AAAS.PRO[ESS,JMC]
3D.NSF[ESS,JMC]
ISRAEL.LE1[LET,JMC]
ISRAEL.ART[LET,JMC]
CHENEY.LE1[LET,JMC]
OCTOBE.OUT[LET,JMC]
CORBAT.LE1[LET,JMC]
EFOO[ESS,JMC]
CALHOU.LE2[LET,JMC]
COMNEE[ 1,JMC]
COMTOP.LIS[ESS,JMC]
MICHIE.LE4[LET,JMC]
NET1[ 1,JMC]
ALUMNI[ 1,JMC]
EVANS.LE1[LET,JMC]
EVANS.LE2[LET,JMC]
INFO[ 1,JMC]
INFO.NEE[ESS,JMC]
NETDOC.PLN[ESS,JMC]
LIB.PRO[ 1,JMC]
SUFFIC.FRE[LET,JMC]
OCT.USE[ESS,JMC]
OLIVER.LE1[LET,JMC]
INTRO.AI[ 2,JMC]
ADMIS.LET[LET,JMC]
COMMON.NS[LET,JMC]
DRELL.LE1[LET,JMC]
SIGART.BUL[LET,JMC]
SIGART.LE1[LET,JMC]
SIGART.LE2[LET,JMC]
BOWEN.LE2[LET,JMC]
RUSSOF.NOT[ESS,JMC]
INVOIC[LET,JMC]
ACM.LE2[LET,JMC]
ACM.LE1[LET,JMC]
JOSHI.LE1[LET,JMC]
FERRIS.LE1[LET,JMC]
NYT.LE2[LET,JMC]
SOCIAL[ 1,JMC]
BDAY.LIT[LIT,JMC]
NONTER.PRB[258,JMC]
CORREC.ABS[258,JMC]
BACKUP.MCP[258,JMC]
HOMOMO.NOT[258,JMC]
MCP.PR1[258,JMC]
COND.PRF[258,JMC]
PQR.PRF[258,JMC]
SCHEM.WRU[258,JMC]
BRACEW.LE1[LET,JMC]
GREEK.JMC[258,JMC]
MCCUS.PRO[258,JMC]
FOLSCO.MEM[258,JMC]
EXTFOR.MEM[258,JMC]
ACTOR0[258,JMC]
NAUR.LE1[LET,JMC]
FEMANO.LE1[LET,JMC]
MTC.GRP[ESS,JMC]
PETERS.LE1[LET,JMC]
JAN.ME[LET,JMC]
LEGUIN.CRI[LIT,JMC]
JAN14.DBX[258,JMC]
SUM.CAR[ 1,JMC]
GELLMA.LE1[LET,JMC]
ESSEX.LE1[LET,JMC]
HOBBS.LE1[LET,JMC]
BREMER.LE1[LET,JMC]
SHELDO.NS[LET,JMC]
FOLMAN.MOD[258,JMC]
APPEND.CON[258,JMC]
LOOKUP.SPE[ 1,JMC]
CONSUL.POX[ESS,JMC]
AILET[LET,JMC]
REYNOL.LE1[LET,JMC]
ACM.LE3[LET,JMC]
LEAST.CM1[258,JMC]
NORBYE.LE1[LET,JMC]
JAN.OUT[LET,JMC]
DEC74.IN[LET,JMC]
JO.NOT[258,JMC]
MTC.BLA[258,JMC]
LOGIC.QUA[258,JMC]
OUTLIN.256[258,JMC]
HAND.2[258,JMC]
MANNA.PR2[258,JMC]
MANNA.CM2[258,JMC]
HARDPR.OOF[258,JMC]
MANNA.WRU[258,JMC]
DEMO.PR1[258,JMC]
OKAMOT.LE1[LET,JMC]
HAND3.W75[258,JMC]
HAND.W75[258,JMC]
CHRI.LE1[ 2,JMC]
VIS[ 2,JMC]
VIS1[ 2,JMC]
VIS.CAP[ 2,JMC]
NNN[ 2,JMC]
VITASK[ 2,JMC]
FRAGA.LE1[LET,JMC]
ENERGY.LE1[LET,JMC]
GORDON.LE2[LET,JMC]
GORDON.LE1[LET,JMC]
FIELDS.ME2[LET,JMC]
DEC.ME[LET,JMC]
TASK[LIT,JMC]
FEB.ME[LET,JMC]
ADHOC.THE[258,JMC]
SAKAI.LE1[LET,JMC]
LICK.LE1[LET,JMC]
LICK.ME1[LET,JMC]
LICK[ESS,JMC]
PARTIT.LSP[ 1,JMC]
HAND.1[258,JMC]
BIBLE[245,JMC]
USER[245,JMC]
IDEAS[225,JMC]
SUZUKI.REC[LET,JMC]
MUFTIC.LE1[LET,JMC]
SAKAI.LE2[LET,JMC]
VOR1[ 1,JMC]
MCP3.AX[258,JMC]
EXTFOR.AX[258,JMC]
COMPIL.AX[258,JMC]
COND.AX[258,JMC]
INTEGE.AX[258,JMC]
INDUC.AX[258,JMC]
ZF.AX[258,JMC]
INTEG2.AX[258,JMC]
EQUAL.AX[258,JMC]
MCP.AX[258,JMC]
MCP2.AX[258,JMC]
ASHLEY.LE1[LET,JMC]
74DEC1.AJT[LET,JMC]
SMITH.LE1[LET,JMC]
GAREY.LE1[LET,JMC]
BERG.LE2[LET,JMC]
KOREA.POL[S75,JMC]
KOREA.2[S75,JMC]
RUSSIA.LE1[LET,JMC]
GELFAN.LE1[LET,JMC]
MT.PRO[ 1,JMC]
JUN75.OUT[MSG,JMC]
JUL75.OUT[MSG,JMC]
AUG75.OUT[MSG,JMC]
SEP75.OUT[MSG,JMC]
JAN75.OUT[MSG,JMC]
DEC74.OUT[MSG,JMC]
MAR75.OUT[MSG,JMC]
APR75.OUT[MSG,JMC]
MAY75.OUT[MSG,JMC]
INGERM.LE1[LET,JMC]
APR75.IN[MSG,JMC]
MAR75.IN[MSG,JMC]
FRA[ 1,JMC]
FR2[ 1,JMC]
SQUIBS.PRO[ 1,JMC]
MODEST[S75,JMC]
CHESS1[CUR,JMC]
SOCIAL.DEF[CUR,JMC]
RAMSEY[CUR,JMC]
HOW.ESS[CUR,JMC]
LEADER.ART[CUR,JMC]
LICK.PRE[CUR,JMC]
REPRES.LIC[CUR,JMC]
REPRES.LI2[CUR,JMC]
GOALS.LIC[CUR,JMC]
TWOMOD.ESS[CUR,JMC]
FORMAL.PRO[CUR,JMC]
SALARY.MEM[CUR,JMC]
SCIP[CUR,JMC]
IDEOLO[CUR,JMC]
ECON.NOT[CUR,JMC]
AICIRC.ABS[CUR,JMC]
CPDUST[CUR,JMC]
IDEOLO.ART[CUR,JMC]
APHOR.AI[CUR,JMC]
ECO.ESS[CUR,JMC]
RACKET.ESS[CUR,JMC]
PROPOS.BLA[CUR,JMC]
PHIL[CUR,JMC]
COMMON[CUR,JMC]
MENTAL[CUR,JMC]
NEWS75.MOD[CUR,JMC]
EXEC.SUM[CUR,JMC]
PRODUC.ESS[CUR,JMC]
STANDA.ESS[CUR,JMC]
ARPMTC[CUR,JMC]
IDEOLO.ESS[CUR,JMC]
JAN75.IN[MSG,JMC]
ARPA75.OUT[MSG,JMC]
1974.IN[MSG,JMC]
AUG75.IN[MSG,JMC]
JUL75.IN[MSG,JMC]
FEB75.IN[MSG,JMC]
XGPPUB.1[ESS,JMC]
ADMIT.PUB[CUR,JMC]
KHABAR.AI[S75,JMC]
NEWS75.SUP[ESS,JMC]
NAKATA.LE1[LET,JMC]
LEM.LE1[LET,JMC]
TWOEAS.MEM[258,JMC]
HOMER.LE1[LET,JMC]
SCHULT.LE1[LET,JMC]
REPRES.DOC[ESS,JMC]
REPRES.NOT[ESS,JMC]
HENDRI.REV[ESS,JMC]
REPRES.REP[ESS,JMC]
OPTIM[F75,JMC]
DYSON.LE1[LET,JMC]
LUMELS.LE1[LET,JMC]
NEWBOR.REV[F75,JMC]
TASK.TA[F75,JMC]
NICK.FIL[F75,JMC]
SEXP[F75,JMC]
AI[F75,JMC]
LATIME.LE1[LET,JMC]
GENERA[F75,JMC]
LISP.MAN[S75,JMC]
SANDEW.LE2[LET,JMC]
PHIL.ART[F75,JMC]
DEFINE[F75,JMC]
BULLET[F75,JMC]
DATA1.204[F75,JMC]
DATA4.204[F75,JMC]
DATA3.204[F75,JMC]
ISHII.LE1[LET,JMC]
TEXTOR.LE1[LET,JMC]
ED1.DO[ESS,JMC]
EDH.DO[ESS,JMC]
OFFICE.PLN[F75,JMC]
REPRES.P75[ 1,JMC]
AIORG.PLN[ 1,JMC]
ASHENH.LE1[LET,JMC]
ASHENH.LE2[LET,JMC]
IJCAI.DOC[F75,JMC]
IJCAI.ENQ[F75,JMC]
AAAS.76[ESS,JMC]
DCL2.75[LET,JMC]
LUK.LE1[ 2,JMC]
AAAS[ESS,JMC]
SCIENC.LE2[LET,JMC]
HOARE.LE1[LET,JMC]
SCIENC.LE3[LET,JMC]
ABELSO.LE1[LET,JMC]
LUCKHA.LE1[LET,JMC]
NILSSO.ME1[LET,JMC]
HJERPP.LE1[LET,JMC]
MOSES.LE1[LET,JMC]
RELATI.POX[CUR,JMC]
BROWN.LE1[LET,JMC]
GILFIL.REV[F75,JMC]
SCOTT.LE1[LET,JMC]
SENSE.AD[LET,JMC]
PR1.DEC[F75,JMC]
AJT.MEM[ESS,JMC]
RAWLS.REV[ESS,JMC]
SENSE.POS[F75,JMC]
CYCLOP[F75,JMC]
GINI.LE1[LET,JMC]
MARTEL.LE1[LET,JMC]
MARTEL.DOC[LET,JMC]
FRIED.LE2[LET,JMC]
FRIED.LE1[LET,JMC]
ROTH.LE1[LET,JMC]
BRINKE.LE1[LET,JMC]
FIX[F75,JMC]
FIXUP.LBK[F75,JMC]
TABLE[F75,JMC]
TERMPA.206[F75,JMC]
FIXUP2.LBK[F75,JMC]
RYLE.REV[F75,JMC]
FIXUP[F75,JMC]
TUNNEY.LE1[LET,JMC]
NEWS75.PUB[CUR,JMC]
FIXUP2[F75,JMC]
BRINKL.LE1[LET,JMC]
JAF.LE1[LET,JMC]
GARDNE.LE2[LET,JMC]
SENSE.LE1[LET,JMC]
MOTIV.ART[ AI,JMC]
DAILY.ART[F75,JMC]
SENSE.LE2[LET,JMC]
REVAL2.LBK[F75,JMC]
REVAL3.LBK[F75,JMC]
AUTOMA.2[F75,JMC]
AUTOMA[F75,JMC]
COUNTE[F75,JMC]
PR1.AX[F75,JMC]
HOTER.1[CUR,JMC]
HOTER[CUR,JMC]
MELTZE.LE[LET,JMC]
NOV.ME[LET,JMC]
MACHIN[F75,JMC]
HEAD.LET[LET,JMC]
BETHE.LE1[LET,JMC]
SENSE.DAT[F75,JMC]
BETHE.LE2[LET,JMC]
GOLDMA.LE1[ 2,JMC]
FINITI[F75,JMC]
SOULE.LE1[LET,JMC]
THESIS[F75,JMC]
MINIMA[F75,JMC]
SCIENT.LE1[LET,JMC]
FIB[F75,JMC]
DEC75.IN[LET,JMC]
NEWYOR.LE1[LET,JMC]
CHERN.LE1[LET,JMC]
KAHN.LE1[LET,JMC]
FIELDS.ME1[LET,JMC]
BLEDSO.LE1[LET,JMC]
BLEDSO.DOC[LET,JMC]
YOST.LE1[LET,JMC]
FIELDS.SUP[LET,JMC]
RAPHAE.LE1[LET,JMC]
GARDNE.LE1[LET,JMC]
GABOR.LE1[LET,JMC]
RHODES.LE1[LET,JMC]
LISP.INT[F75,JMC]
BLOCKS.226[F75,JMC]
NOTES.226[F75,JMC]
TAUT.PRF[F75,JMC]
UTILI.AIL[F75,JMC]
MOSSIG.LE1[LET,JMC]
PET.LIT[LIT,JMC]
TOLEDO.LE1[LET,JMC]
CBCL[F75,JMC]
SET[F75,JMC]
SET.AX[F75,JMC]
SENSE.LE3[LET,JMC]
FILMAN.RE2[LET,JMC]
GIL1[ 2,JMC]
KNOPOF.LE1[LET,JMC]
YAMADA.LE1[LET,JMC]
CULIK.LE1[LET,JMC]
ANDREI.ADR[ESS,JMC]
JAN76.IN[LET,JMC]
MILLER.LE0[LET,JMC]
LUCKHA.BLA[F75,JMC]
JAN76.OUT[LET,JMC]
ARPA75.IN[MSG,JMC]
ADMIT.TEL[LET,JMC]
LICK.1[LET,JMC]
MENTAL.NOT[F75,JMC]
PHIL.ART[ESS,JMC]
WILSON.LE1[LET,JMC]
WILKIN.LE1[LET,JMC]
OSIS.LE1[LET,JMC]
GERROL.LE1[LET,JMC]
SFWA[S76,JMC]
COMPUT.CSD[ESS,JMC]
CSDCOM[ 1,JMC]
CSDDIS.N25[ESS,JMC]
QUOTE[SEN,JMC]
QUOTE.2[SEN,JMC]
BAT0.LOG[F75,JMC]
CELLUL.MEN[F75,JMC]
RELDEF.MEN[F75,JMC]
MOTIV.MEN[F75,JMC]
CONTEN.MEN[F75,JMC]
NOTE.MEN[F75,JMC]
POLEMI.MEN[F75,JMC]
NOTE1.MEN[F75,JMC]
AI.MEN[F75,JMC]
MOTIVA.MEN[F75,JMC]
SELFCO.MEN[F75,JMC]
CHANGE.MEN[F75,JMC]
TECHNI.MEN[F75,JMC]
AUTOMA.MEN[F75,JMC]
MODEL.MEN[F75,JMC]
EPISTE.MEN[F75,JMC]
CRYPT.MEM[ 2,JMC]
CRYPT.CHK[ 2,JMC]
COMMUN[S76,JMC]
PSYCHO.AP[ESS,JMC]
TASK.FOO[ESS,JMC]
ANDREI.DAT[ESS,JMC]
SUSAN.TAX[ESS,JMC]
ALL[ESS,JMC]
IGC.PRO[ESS,JMC]
JO.NO2[ESS,JMC]
IGSCS.PLN[ESS,JMC]
DRAW.PRO[ESS,JMC]
VIEW1[ESS,JMC]
RACRUL.MEM[ESS,JMC]
SPUT.PRO[ESS,JMC]
LORE[ESS,JMC]
TECH2.BGL[ESS,JMC]
SUBTES[ESS,JMC]
CSDCOM.REQ[ESS,JMC]
LIST[ESS,JMC]
FREE[ESS,JMC]
GUBBB[ESS,JMC]
BANK[ESS,JMC]
COLE.LET[ESS,JMC]
LIGHT.6[ESS,JMC]
SERA.ZOO[ESS,JMC]
NEWMEX.SPE[ESS,JMC]
PUBLIC.INT[ESS,JMC]
PRO2[ESS,JMC]
JAN75.DBX[ESS,JMC]
LIVER.SAI[ESS,JMC]
IBM.SNG[ESS,JMC]
REGIST[ESS,JMC]
PRIZE.PRO[ESS,JMC]
IBM.ABS[ESS,JMC]
GORIN[ESS,JMC]
NEWREP.LE1[ESS,JMC]
LUCK.REC[ESS,JMC]
CH1.BH[ESS,JMC]
ENERGY.DEM[ESS,JMC]
BROOKS.PLN[ESS,JMC]
NSFX.1[ESS,JMC]
ENERGY.STA[ESS,JMC]
LI[ESS,JMC]
TES[ESS,JMC]
ASSIGN[ESS,JMC]
PDP370.PLN[ESS,JMC]
CONTEN.DOC[ESS,JMC]
FP[ESS,JMC]
TIME.HIS[ESS,JMC]
CODE.PRO[ESS,JMC]
ENERGY.PRO[ESS,JMC]
RATIO.NOT[ESS,JMC]
CHESSP.DIS[ESS,JMC]
TASK[ESS,JMC]
REV[ESS,JMC]
RAY.AP[ESS,JMC]
REFER[ESS,JMC]
REPRES.BIB[ESS,JMC]
ENERGY.QUE[ESS,JMC]
SLAGLE.REC[ESS,JMC]
GENER.AI[ESS,JMC]
ENERGY.FAC[ESS,JMC]
CHARAC.PRO[ESS,JMC]
VAL.JM2[ESS,JMC]
FELDST.REV[ESS,JMC]
MAGS.LIS[ESS,JMC]
ED.DO[ESS,JMC]
SCITOP.LIS[ESS,JMC]
GREET[ESS,JMC]
MUSIC.GRP[ESS,JMC]
EVENT.NOT[ESS,JMC]
GRK.TES[ESS,JMC]
POEM[ESS,JMC]
MEMO.PUB[SEN,JMC]
HOOK.LE1[LET,JMC]
WEISKO.LE1[LET,JMC]
MONOP.STA[ESS,JMC]
BULLET.AP[ESS,JMC]
AI2.QUA[ESS,JMC]
SPACE.OP[ESS,JMC]
AISY2.QUA[ESS,JMC]
DATAMA.LE1[LET,JMC]
CHAP1.ART[ESS,JMC]
CHAP2.ART[ESS,JMC]
R1500.2[ESS,JMC]
SUFFIC[ESS,JMC]
INDIA.PRO[ESS,JMC]
PROP.ECO[ESS,JMC]
TECTOP.LIS[ESS,JMC]
R1500.3[ESS,JMC]
R1500.1[ESS,JMC]
TURIN2.LEC[ESS,JMC]
HOTER.PUB[ESS,JMC]
HOTER.DOC[ESS,JMC]
STEVE.SPI[LET,JMC]
GALLAG.SPI[LET,JMC]
STEVE.ME1[LET,JMC]
STEVE.ME2[LET,JMC]
POSTER[SEN,JMC]
SENSE.LAB[W76,JMC]
SENSE.PL1[W76,JMC]
AAAS[W76,JMC]
FREGE[W76,JMC]
ENERGY[W76,JMC]
FOL.COM[W76,JMC]
MEET2.SEN[W76,JMC]
IDEOLO[W76,JMC]
HEAVY.SET[W76,JMC]
SENSE.NOT[W76,JMC]
LISP2[W76,JMC]
TASKS.SET[W76,JMC]
CS.226[W76,JMC]
PDP10[W76,JMC]
RUSSEL.PRF[W76,JMC]
LISP.OLD[W76,JMC]
UNIT.PRF[W76,JMC]
Z2.AX[W76,JMC]
HEAVY.AX[W76,JMC]
BUREAU[W76,JMC]
SENSE.PST[W76,JMC]
BLOCKS.AX[W76,JMC]
OBVIOU[W76,JMC]
NUMBER.AX[W76,JMC]
COMMUN[W76,JMC]
HEAVY.THE[W76,JMC]
KOREA[W76,JMC]
SENSE.BLA[W76,JMC]
MOYNIH.1[W76,JMC]
REM.WRU[W76,JMC]
RESOUR[W76,JMC]
AAAS.2[W76,JMC]
SNEKOT.DAT[W76,JMC]
SCI[LET,JMC]
ASHLEY.COV[SEN,JMC]
LICKLI.SPI[LET,JMC]
SENSE.HOM[SEN,JMC]
COMMEN[S76,JMC]
PROBLE[S76,JMC]
BOARD.ART[ AI,JMC]
PTCH2[ AI,JMC]
SIGN[SEN,JMC]
PAUSD.CMN[ 1,JMC]
NEEDS.CSD[ 1,JMC]
PACKAR.LE1[LET,JMC]
SHUTDO[S76,JMC]
PUBLIC.DOC[LET,JMC]
PUBLIC.LE2[LET,JMC]
KNOW.PRF[F75,JMC]
KNOW.AX2[F75,JMC]
KNOW.CM1[F75,JMC]
MORAVE[S76,JMC]
PNUELI.LE1[LET,JMC]
PNUELI.LE2[LET,JMC]
FILE[S76,JMC]
CALIF[SEN,JMC]
STATEM.LET[SEN,JMC]
SAMPLE.LET[LET,JMC]
SEMAN[ 1,JMC]
EXTENS[S76,JMC]
MAY76.IN[LET,JMC]
JUN76.IN[LET,JMC]
PACIFI.LE2[LET,JMC]
CARNAP[S76,JMC]
PLF.LE1[LET,JMC]
HOTER.POX[PUB,JMC]
ADVISO[S76,JMC]
HAYASH.LE1[LET,JMC]
LENG.LE1[LOT,JMC]
JUL10[S76,JMC]
LETTER.PUB[LET,JMC]
MENTAL.SUP[F75,JMC]
LUCKHA.LE2[LET,JMC]
COMPUT.REQ[LET,JMC]
DWORKI.LE1[LET,JMC]
KNUTH.LE1[LET,JMC]
LEM.LE2[LET,JMC]
∂13-Feb-78 2114 DEK message from south america
To: JMC, feigenbaum at SUMEX-AIM
after reading the manna letters, i am pleased to support his appointment
to a tenure position. burstall and wegbreit were a bit cool but taken as
a whole the case looks excellent to me and should make it successfully
through the h+s committee with no more than the usual long hassle.
Please give Bob Floyd a copy of the following:
∂13-Feb-78 2207 TOB
John
I am not sure that I communicated this to you
before, but Chern suggested that we would have
trouble if I did not get more faculty support.
He was concerned that you did not show up for
the workshop (I overlooked letting you know
about it and asking you to put in an appearance).
and thought that the proposal would encounter
questions along the lines of: McCarthy is not
very involved and Binford is a research associate.
So he asked that I get more faculty support. So
I asked Bernie Roth and Feigenbaum to act in an
advisory capacity. Feigenbaum does not have time
for more. Roth may have the time and sounds interested
but says that he thinks of this as my thing. He
has a reputation of staying with theoretical things
and avoiding large projects, but that may not be
valid information. Dan DeBra seems very interested.
He too, is busy and I don't know how much he would
be involved. But I have a very good impression of
him and am surprised at his interest now.
If either put energy into it now, I would enjoy it.
Chern would like more than just lending a name I think.
He wants Feldman and talked with you about that. Jerry
has all he wants to do and would have a limited role.
He wants to support the proposal and he has an investment
in the design review of AL. We can take advantage of some
work in distributed computing. We will probably support
Amy Lansky for this coming summer to bring up SAIL-PLITS,
a version of SAIL with communication facilities. She brought
it up at SRI last summer, and wants to come out again.
He will do some advising on AL planning systems. That is
about the level of collaboration that I see possible.
There is a possibility for collaboration with Voelcker about
Computer Aided Design. He has a system, PADL, which does CAD
in a batch mode for restricted geometric primitives (circular cylinders
and blocks aligned with coordinate axes). He suggests a problem
which I suggested in NSF proposals before, automated planning
of dimensional inspection. If done as a collaboration, PADL
would be coupled with AL. That has one advantage. That class
of primitives avoids the tough problems, and simplifies the task
greatly. Somehow, I prefer tough problems.
Chern supports strongly our working in CAD and I think it is
a good idea. We are back in it again for vision, and I think that
we have a good start now. We will be showing some new programs
soon. I welcome suggestions. I remember that you were enthusiastic
about mechanical design programs. I am looking at it for
model-based vision, for sheet metal fabrication like aircraft, where
each part may be slightly different, but with only slight variations
(wing cross section elements), and for using with sensing.
I have looked for a community of users for a regional center
but I have not found one yet. Perhaps I can get support from
some industry people. I was pessimistic but have not really
tried anyone in industry.
Tom
∂13-Feb-78 2252 RAK SENSE announcement
I changed the grammar in a few places. I will not be around very much
tomorrow and am leaving Wed for Houston (although I think I will be here
on Feb 28). So most of what I can do has to happen tonight. I propose,
therefore, that I check out and complete the addresses by looking at the
book and get the labels to pub out ok.
If I am around tomorrow, I will do more, but I have an almost ∞ number
of errands to run.
Fair enough for a start?
Dick
Thanks much for that. If Les is around, he can help.
∂13-Feb-78 2318 RAK Who is Charles Vest?
He is listed as a Prof. Aero and Astro but is not in the faculty directory.
Should I drop him?
Also, is your manifesto ending specifically directed at Panofsky? It looks
like it.....
I don't know him. Maybe he was a visitor that year or maybe Taylor
misread a name.
Not only Panofsky but also George Pake who is the present
president of APS, but I think now that it is
insufficiently tactful and should be deleted.
Dick
For your information:
∂14-Feb-78 1156 CCG leave
I have decided to accept your offer of support and to take the title
of assistant professor on leave. I will stay at the lab, receive salary
from the Arpa contract and my NSF grant. I would like the leave to begin
the end of march. I won't be teaching CS224, so a replacement should be found as
soon as possible. I believe Nils did it last year, in cooperation with Ed.
I will be supervising my remaining 9 graduate students, completing the PSI
book, getting a JACM paper out, and doing some interviewing/traveling.
My main obligation is to the nearly-finished graduate students and to the
technical success of the project.
Thanks
Cordell
∂14-Feb-78 1615 FTP:Levinthal at SUMEX-AIM elemental abundance
Date: 14 Feb 1978 1615-PST
From: Levinthal at SUMEX-AIM
Subject: elemental abundance
To: jmc at SU-AI
RELATIVE COSMIC ABUNDANCES OF ELEMENTS
HYDROGEN 10,000,000
HELIUM 1,400,000
OXYGEN 6,800
CARBON 3,000
-------
∂15-Feb-78 1347 CCG les
I briefed les on my plans and the financial aspects. I believe he has the
information for you.
Thanks
Cordell
∂15-Feb-78 1535 FTP:Feigenbaum at SUMEX-AIM something to think about re history of our science
Date: 15 Feb 1978 1535-PST
From: Feigenbaum at SUMEX-AIM
Subject: something to think about re history of our science
To: buchanan, jmc at SAIL, les at SAIL
Date: 15 Feb 1978 1504-PST
From: Levinthal
Subject: cybernetic artifacts
To: lederberg, feigenbaum
Mail from MIT-AI rcvd at 15-Feb-78 1356-PST
Date: 15 FEB 1978 1655-EST
From: MINSKY at MIT-AI (Marvin Minsky)
Sent-by: MINSK0 at MIT-AI
To: NASA at MIT-AI
The Smithsonian is thinking about collecting important Cybernetic Artifacts.
Can you think of anything in your laboratory that
you think has historical interest? Or suggest other
sources?
--minsky
I am not sure what a cybernetic artifact is. Would the first
MS to be correctly interpeted by Dendral qualify? In any case I thought
you'd be interested in the message.
-------
-------
∂15-Feb-78 1553 FTP:Feigenbaum at SUMEX-AIM Cordell's leave
Date: 15 Feb 1978 1552-PST
From: Feigenbaum at SUMEX-AIM
Subject: Cordell's leave
To: jmc at SAIL
John, I got Cordell's note to you. The leave as discussed is not the same as
the leave that you and I had talked about,i.e. Cordell wants the leave to start
THIS year so that he won't have to teach CS224 in the Spring. I had mentioned
to you that he could be on leave for next year without bothering anything.
If he is to be on leave for this year, Spring Quarter, a replcement will have
to be arranged for CS224 pronto or The Dean will not approve the application for
leave (as is fair to all). That means that you as Acting Chairman will have to
find such a replacement. I would do it but I am teaching CS101 and would have
to find a replacement there. Or you could do it. Or you could try to recruit
Nils Nilsson to do it. Nils would do an excellent job, but I'm skeptical
of relying on any other outsider (other than Nils) to run this very importantintroductory
course in our sequence.
Easiest thing to do is to deny Cordell's request for a leave THIS year on the
grounds that it is just too late to make alternate arrangements.
Ed
-------
∂15-Feb-78 2217 DEK baskett papers
It just occurred to me that the job of getting letters for Baskett
might be a little simpler if it were done by a joint EE-CS committee.
For example, you could suggest that Mike Flynn initiate such action.
If you think this is a good idea, be sure to show him the kind of
letters CS has been using for tenure matters, as they were intended to
be in the style Royden wants.
Please give a copy of this to Bob Floyd.
∂16-Feb-78 1526 FTP:Wiederhold at SUMEX-AIM Cuuriculum
Date: 16 Feb 1978 1506-PST
From: Wiederhold at SUMEX-AIM
Subject: Cuuriculum
To: rwf at SAIL, jmc at SAIL, 2 at SAIL, feigenbaum at SAIL
cc: dpb at SAIL, taynai, jlh at SAIL, jmb at SAIL, sso at SAIL
Since we have had no feedback on the system curriculum
I assume that the curriculum as described will go into
the bulletin now, and that any faculty meeting subsequent
has the option of making changes in prrof, if deemede neccessary.
Gio
-------
Say again what the curriculum does about program verification.
Is that part of Systems?
∂16-Feb-78 1702 FTP:Wiederhold at SUMEX-AIM (Response to message)
Date: 16 Feb 1978 1658-PST
From: Wiederhold at SUMEX-AIM
Subject: (Response to message)
To: JMC at SU-AI, feigenbaum
cc: DPB at SU-AI, JLH at SU-AI, JMB at SU-AI, RWF at SU-AI,
cc: SSO at SU-AI
In response to the message sent 16 Feb 1978 1541-PST from JMC at SU-AI (John McCarthy)
At bob Floyds suggestion program verification was removed as a core course
from the system curriculum, since it is a subject covered in the Theory
area. We do expect that Susan Owicki et al will be having graduate
seminars on their specific interests in verification.
I personally would like to see some of this work in systems, it
was a topic mentioned in the old CS140A&B sequence, but covered infrequently
and inconsistently. Bob seemed to feel that maturization is neccessary
prior to fixing this topic in the basic teaching sequence.
For further discussion Susan is really the proper person to talk/type to.
Gio
-------
Leaving verification in the theory area suits me too. Ultimately, verification
will be part of every area of computer science, e.g. the hardware people
will show how to prove facts about hardware and the numerical analysts
will show how to prove facts about numerical algorithms. Some of this
will use techniques peculiar to the particular branch of computer science,
but it will mostly be based on techniques developed in what I have
called mathematical theory of computation. I am curious whether others
agree with this opinion, even though I don't think it suggests any
additional immediate decisions.
∂16-Feb-78 1757 FTP:Wiederhold at SUMEX-AIM (Response to message)
Date: 16 Feb 1978 1748-PST
From: Wiederhold at SUMEX-AIM
Subject: (Response to message)
To: JMC at SU-AI, feigenbaum, DPB at SU-AI, JLH at SU-AI,
To: JMB at SU-AI, RWF at SU-AI, SSO at SU-AI
In response to the message sent 16 Feb 1978 1740-PST from JMC at SU-AI (John McCarthy)
I agree, at that time there may be a place in the basic courses that for
instance relate the theory to system practice. gio
-------
∂17-Feb-78 0221 LES S-1 communications
To: LCW
CC: LLW, JMC, JBR
Your 15 Feb. message suggests to me that we need to discuss communication
mechanisms between LLL and SAIL. There appear to be at least three kinds
of guidance that will be emitted by the LLL group, as follows.
1. Formal agreements, such as contractual agreements. Clearly we are
obligated to fulfill our part of such agreements.
2. Coordinated guidance, representing strongly held positions on technical
or procedural issues that are important to the successful integration of
the various system elements. The LLL group should develop mechanisms
for coordinating positions of this sort and they should be clearly
identified as such. The SAIL group will give careful attention to such
guidance. We recognize that a refusal on our part to follow such
guidance might jeopardize our working relationship.
3. We expect and welcome suggestions of all sorts. Of course, we will
reject your bad ideas as readily as you will (or should) reject ours.
The suggestions in your message strike me as not very helpful. We expect
Jeff Rubin to do the basic planning for the operating system and it is
up to him to develop procedures that suit his working style. I will
not try to tell him how to do that unless he either asks me or I become
convinced that he is getting into trouble.
∂17-Feb-78 0242 LES Cybernetic Artifacts
To: Feigenbaum at SUMEX-AIM
CC: JMC
Assuming that people do not qualify as cybernetic artifacts, we still might
have some things of interest, e.g. part of a computer-controlled arm that
was either the second or third one in the world (I'm not sure which).
We also have all of the fourth and fifth such arms, i.e. the Rancho
electric and Stanford hydraulic arms. Later on, we might give them the
Cart, which began life as a lunar rover simulator.
∂17-Feb-78 0254 LCW SAIL Contract
To: LES
CC: LCW, LLW, JBR, JMC
The contract was mailed Wednesday to Stanford. It has to be signed by
Stanford, then returned to LLL for immediate signature. I have observed
this cycle, under pressure, to require approximately 4 working days. The
purpose of this note is to inform LES that the contract is now in
Stanford's hands.
Curt
∂17-Feb-78 0806 FTP:Feigenbaum at SUMEX-AIM copy of msg to marvin
Date: 17 Feb 1978 0802-PST
From: Feigenbaum at SUMEX-AIM
Subject: copy of msg to marvin
To: les at SAIL, jmc at SAIL
Marvin, to whom does one write (or call) at the Smithsonian if one wants
to donate cybernetic artifacts, such as early arms and carts?
Best,
Ed
-------
∂17-Feb-78 0901 SSO Herrero
Herrero never showed up at my office on Thursday morning.
Hope there wasn't a confusion about time.
Re the verification discussion (I couldn't read my mail yesterday)
I think that the idea of proving correctness, in an informal way,
should be introduced early to programmers. I expect to do it a
bit in the new 107. I will offer an advanced course next spring
with a title like "problems in concurrent programming".
I would expect to spend a fair amount of time on verification
as well as language and design issues.
There was no confusion about time. I fear this means that Herrero's
trip was personal, to see his student son, and that talking with
the CS Department was incidental. Therefore, I doubt we will hear
more from ITT Europe. Hope I'm wrong. On the other hand, he looked
a little peaky, so maybe he came down with flu. Anyway thanks.
∂17-Feb-78 1009 BS Letter for signature
Re the matter we discussed yesterday, I have prepared a letter and a form
for your signature.
Because of a great many personal things I have to do, I would like to take
next week as vacation. Plus I have a PG&E emergency that I have to take care
of this afternoon. Is this o.k.? And, if so, do you want me to send the letter
and the form to the lab for your signature?
Betty
Please give them to Laurie Levin, who is serving as my campus secretary,
and ask her to have me sign them on Tuesday.
∂17-Feb-78 1203 TOB proposal
If you are going to make any comments, do it
now. We are at an absolute deadline.
Tom
∂17-Feb-78 1247 BPM
"Chess"
by MALCOLM W. BROWNE
(c) 1978 New York Times News Service
11:48 am, 17 February 1978
NEW YORK - In 1968, David Levy, a British international chess
master, bet a group of computer experts 1,250 pounds that in the
decade that followed no chess-playing computer could beat him. This
year, Levy expects to collect his wager, but it is turning out to be
a much closer race than many had imagined possible.
Chess-playing skill is widely regarded as a legitimate gauge of at
least one aspect of human intelligence. Most human chess masters had
long believed that although computers could rapidly sift through an
enormous number of possibilities, machines lacked certain qualities
vital to real mastery of the game.
But new computer programs are proving to be such formidable
opponents in tournament play, experts say, that old ideas about the
limitations of computer intelligence must be revised.
The man-versus-machine controversy this year will center on two
conspicuous antagonists.
One is Levy, a 32-year-old Londoner who writes chess books and is
classified as a ''weak'' international master - one of the few
hundred best chess players in the world.
Opposing him will be a computer program called ''Chess 5.0'' divised
by David J. Slate, a 33-year-old computer expert of Northwestern
University, in Evanston, Ill. Slate's program, as used on a powerful
Control Data Corporation Cyber 176 computer, is also among the
several hundred best chess players in the world.
In separate interviews, Slate and Levy agreed that Levy would
probably beat the computer in a final match, which may be played in
August at the Canadian National Exposition in Toronto.
If so, Levy will win 1,250 pounds (about $2,400) from four
professors in computer technology - Donald Michie of Edinburgh
University, Seymour Papert of Massachusetts Institute of Technology,
John McCarthy of Stanford University, and Edward Kozdrowiski of
Princeton University. He also hopes to receive large royalties from
television coverage.
Referring to the Northwestern computer group Levy said: ''Unless
they make some major improvement in their Chess 4.6 program, my
chances are good. They don't have to get much better to beat me, and
they've already beaten me in speed matches, five seconds a move. But
they're running out of time.''
Slate said: ''Using our present Chess 4.6 program, I'll be very
happy if we take one and a half or two games out of eight from Levy.
If our new Chess 5.0 program proves to work and be as adaptable as we
hope, we could really put a dent in Levy.''
The Chess 4.6 program has already shaken the chess community to some
extent.
In 1976, the Paul Masson Chess Tournament in California attracted
some 700 contestants, including Slate's program, which was entered in
the Class B level against 128 good amateur players. The computer
easily won all its games to take a $750 prize, which Northwestern had
agreed in advance to relinquish.
In February 1977, Chess 4.6 entered the Minnesota Open Tournament,
playing against some highly ranked opponents.
Chess 4.6 won the tournament 5 to 1, emerging with an official
rating of 2,271 - a chess master.
The Chess 4.6 program is probably the best in the world. Last year
it easily defeated the Soviet ''Kaissa'' program in Toronto, 4 to 0.
Levy and Slate agree that eventually computer programs will
successfully challenge even grand masters.
Protecting friendly pieces and attacking opposing pieces are the
main objects of the computer program, he said, and the program is
weak in planning longrange strategy and subtle positional advantages.
Devising champion computer chess programs, difficult as it is, is
vastly simpler than other problems facing experts in artificial
intelligence, Slate and other experts said.
The single biggest problem now is in giving computers natural
language ability - the ability to converse with untrained humans in
human language.
Grouping computer memory data in terms of ''frames'' - associative
scenarios such as ''ordering a meal at a restaurant'' or
''participating in a fertility rite'' - is one of the new techniques
computer experts are particularly enthusiastic about.
Framed information will be vastly easier to retrive, they say, and
will be available in forms much closer to human associative recall.
While computer scientists are deeply divided over the question of
whether there can be real artificial intelligence [as opposed to artificial
artificial intelligence, I suppose], most of those
interviewed agreed that the so-called Turing test could eventually
settle the argument.
The test is named for the late English mathematician Alan Mathison
Turing, who proposed it in the late 1940s. In the test, a human is
seated at a telecommunications keyboard, which he uses to converse
for a half hour with a computer. He may ask any question, except such
purely anthropomorphic questions as ''What color are your eyes?''
If after the session the human is unable to decide whether his
interlocutor is human or machine, the machine may be said to have
human intelligence, according to the test.
One leading computer expert, Dr. Hans Berliner of Carnegie-Mellon
University, asserts that such a machine already partly exists in the
form of the ''Chess 4.6'' program.
''Most chess players are defeated by the machine and very, very few
players would risk money betting whether they had been beaten by a
human or a machine in this case,'' he said.
∂17-Feb-78 1259 LES S1 software at SAIL
To: Wiederhold at SUMEX-AIM
CC: JMC
Moving your work onto the SAIL machine sounds plausible, but as you
probably know, we are already in a disk space bind. In order to do this,
we will likely have to seek funding for another disk drive. This will be
needed in part to support our own S1 work.
For further negotiations on this idea, please get in touch with JMC.
∂17-Feb-78 1437 FTP:Wiederhold at SUMEX-AIM S1 work.
Date: 17 Feb 1978 1436-PST
From: Wiederhold at SUMEX-AIM
Subject: S1 work.
To: jmc at SAIL
cc: llw at SAIL, lcw at SAIL
In order to further our S1 work I had hoped to be able to make use of
the facilities at SAIL to a greater extent than we do now. I believe that
Lowell Woods assumption is that such cooperation will be part of increased
Stanford S1 activity. For that I reason I would like to direct to you
the request I made to LES earlier today, and his response.
--------------
17-Feb-78 12:00:21-PST,865;000000000001
Mail from SU-AI rcvd at 17-Feb-78 1200-PST
Date: 17 Feb 1978 1156-PST
From: Wiederhold at SUMEX-AIM
Subject: S1 software work.
To: les at SAIL
cc: S1 at SAIL
With this note I request that all current projects of the S1 software
group here ( SOPA compiler, LDI loader, SASS assembler,
runtime , SIMulator, and FORTRAN ) be permitted to move to SAIL.
The situation at SLAC has long impeded progress, mainly due to
limitations of diskspace. For most programs we have not been
able to keep both Source and Object files avaoilable, and had
to resort to inventive JCL manipulations to make things run.
I hope that the sail environment will prove more hospitable.
The move will of neccessity be somewhat gradual, and I expect that
the impact will not be significant, if at all, prior to the
contract being blessed by all. Gio
-------
-------
17-Feb-78 12:59:06-PST,529;000000000001
Mail from SU-AI rcvd at 17-Feb-78 1258-PST
Date: 17 Feb 1978 1259-PST
From: LES at SU-AI (Les Earnest)
Subject: S1 software at SAIL
To: Wiederhold at SUMEX-AIM
CC: JMC
Moving your work onto the SAIL machine sounds plausible, but as you
probably know, we are already in a disk space bind. In order to do this,
we will likely have to seek funding for another disk drive. This will be
needed in part to support our own S1 work.
For further negotiations on this idea, please get in touch with JMC.
-------
--------------
I hope that arrangements can be worked out, I do not believe that we
will need unusually extensive resources. Gio
-------
∂17-Feb-78 1540 CCG via PARC-MAXC cs224
Steve Tappel wants to be the TA for CS224. He's excellent and
will relieve the burden somewhat. Let me know what I can do to help
find a teacher.
Cordell
∂18-Feb-78 1305 BPM Reasoning about knowledge problem
To: JMC, RWW
I don't know who put it on the bulletin board, but I heard a similar problem
a week or so ago from Lyle Ramshaw (RAMSHAW@PARC-MAXC). Someone else had
given it to him, perhaps Al Aho or Bob Sedgewick.
∂18-Feb-78 1416 DEK reasoningaboutknowledge
That problem seems to be an oldie, but I don't remember where I first read it.
A very interesting solution to the problem appears in Vaughan Pratt's CGOL
manual (online at MIT), where in the appendices he gives a lot of example
LISP programs, including one which solves this! In fact the solution given
there was composed on-line one night after a party when Vaughan was quite
inebriated, so it is a unique document! Well worth reading, as I recall it
was approximately after his prime-sieve algorithm.
∂18-Feb-78 1532 DON reasoning about knowledge
To: JMC, RWW
Further bug in phraseology: Not only don't the men know that the numbers are
between 1 and 100 exclusive, but neither knows what information has been given
to either himself or the other! That is, Mr. P needs to be told that the value
he has been given is the product of the two numbers, and that Mr. S knows their
sum, and similarly for Mr. S.
∂18-Feb-78 2106 DEA Reasoning about Knowledge
I put the reasoning about knowledge problem on the board. I admit that
after looking at it, the original version said "S says: I knew you didn't
know. Neither do I", and the version I posted says "S says: I knew you
don't know. Neither do I." Perhaps my dialect of English treats tense
and aspect differently than yours, but I don't see how that difference could
have any effect on the problem. If there is any other problem, please
let me know.
- Doug
The problem is nicely worded except for that point - much better than
the version in the CGOL manual - for example. However, "I knew you
don't know" suggests "I knew you didn't know and wouldn't know even after
hearing this sentence". This excludes all odd numbers of the form
(prime + 2↑k), and ends up excluding all numbers. Before coming to
this conclusion, I preferred that interpretation because it made the
problem more interesting.
∂19-Feb-78 0303 FTP:REM at MIT-MC (Robert Elton Maas) My new data-compression algorithm
Date: 19 FEB 1978 0607-EST
From: REM at MIT-MC (Robert Elton Maas)
Subject: My new data-compression algorithm
To: RWG at SU-AI, JMC at SU-AI, REM at SU-AI, WD at SU-AI
It works. I have a crude Macsyma program that crunches and uncrunches
using either original radix-conversion method or new finite-precision
radix-conversion method. Uncrunch is indeed inverse of crunch.
∂19-Feb-78 0430 REM via SU-TIP#1
To: CRUNCH.DIS[1,REM]:
My new datacompression algorithm works, have Macsyma program that does
it both ways (crunch and uncrunch) in test loop, will be writing up for
publication soon, if you didn't see my abstract yet then contact me.
Macsyma program is self-adapting almost-optimal code for single-token
partial-bit encoding.
∂19-Feb-78 1044 FTP:PRATT at MIT-AI (Vaughan Pratt)
Date: 19 FEB 1978 1345-EST
From: PRATT at MIT-AI (Vaughan Pratt)
To: jmc at SU-AI
I knew you DIDN'T know (as I think my reference makes clear). Interesting
that changing "didn't" to "don't" leads to a different problem.
I trust the manual gives 13 and 4. When I first worked it out, I was so
punchy that I could solve the problem but not keep track of what I was
typing, and for some time the manual claimed it was 14 and 3 or something
like that.
You might enjoy exploring how the problem is affected
1. by not telling S,P about the 100 limit
2. by not having the 100 limit.
I won a $5 bet from Sam Winograd because he didn't spot the possibility
that 1 and 2 might be different. (I bet him that if the limit was imposed
then there was only one solution, which he very happily took me up on.)
By the way, do you have a published source for the problem.?
∂19-Feb-78 1059 RWW
∂18-Feb-78 0011 DEA "Reasoning about Knowledge"
The problem on the board was an interesting little problem I picked up
from Rich Fikes who got it from Danny Bobrow, who got it from someone
at Xerox who got it from someone at ISI. I don't know what the solution is
although I have been assured that there is one. I haven't had any time to
think about it, but I thought it would be fun to devote a few spare cycles to it.
I thought others might find it interesting too, so I put it up on the board.
- Doug
I have solved the problem informally, but have not yet been able to state
the problem in a formal system for reasoning about knowledge. It requires
much more formalism than the problem of the wise men with spots on their
foreheads or the problem of the unfaithful wives.
The problem is discussed and solved in Vaughan Pratt's CGOL manual,
CGOL.PRA[UP,DOC] page 26.
∂19-Feb-78 1100 RWW
∂18-Feb-78 1305 BPM Reasoning about knowledge problem
To: JMC, RWW
I don't know who put it on the bulletin board, but I heard a similar problem
a week or so ago from Lyle Ramshaw (RAMSHAW@PARC-MAXC). Someone else had
given it to him, perhaps Al Aho or Bob Sedgewick.
∂19-Feb-78 1232 FTP:PRATT at MIT-AI (Vaughan Pratt) Reasoning about knowledge
Date: 19 FEB 1978 1533-EST
From: PRATT at MIT-AI (Vaughan Pratt)
Subject: Reasoning about knowledge
To: JMC at SU-AI
CC: RIVEST at MIT-AI
It just occured to me that you (JMC) might like to think about Rivest's
problem of proving properties of encryption schemes (such as,
A tells B a cipher, and provided he doesn't tell C too, C has little
chance of cracking messages between A and B, where "little" is left
uninterpreted in the logic, but in practice is a function of the
quality of the cipher). The properties to be proved are whether
various ways of passing ciphers around are "safe," and similar
things. This is very much a practical application for reasoning
about knowledge.
Incidentally, "modal logic," though nowadays used pretty much exclusively
for Lewis' system, in a broader sense embraces all logics for reasoning
about knowledge.
By the way, I wrote out a dynamic logic version of your monkey-and-bananas
problem - it's on AI:PRATT;MONBAN > and should have been solved in a
fraction of a second by our proof checker except that the proof-checker
hasn't yet been told about parameterized programs.
Sorry, I don't have a published reference to the S-P problem, you'll
have to go to Winograd or McDermott (at Yale) to trace it back, or to
the author of your notice.
∂19-Feb-78 1527 LES Tape RFQ
Remarks?
∂AIL 18 February 1978$addre$$9-track magnetic tapes, request for quote∞
Stanford wishes to purchase a controller and two magnetic tape drives compatible
with IBM 9-track standards.
1. This equipment is to connect to a DECsystem 10 computer (KL10 CPU) at the
Stanford Artificial Intelligence Laboratory.
The controller may connect to either an IBM-equivalent channel (specifically,
a Foonly C-1 channel) or to the KL10 Memory Bus.
A memory bus connection would be about $5,000 less expensive for us than the
IBM channel.
2. Both tape drives must be capable of operating at 1600 bytes/inch. In addition,
at least one must operate at 800 bytes/inch. We are willing to consider
used equipment.
3. The operating line voltage for the system shall be taken from
commercially available power at Stanford. This power is 117V/208V single
phase or 208V/440V 4-wire Wye-connected three phase at a frequency of 60
Hertz plus or minus 2 Hertz. The system shall operate over an input
voltage fluctuation of plus or minus 10α% of normal. Protection against
damage from power transients is very desirable.
4. The system shall operate as specified within a room temperature range
of +55 to +100 degrees F. Any internal cooling requirements shall be
supplied by the Seller.
5. The Seller shall supply four complete sets of instruction manuals
adequate for operation and maintenance of the equipment. The Seller shall
also supply four complete sets of circuit schematics, logic drawings, and
mechanical drawings as are necessary for maintenance, operation, and
repair, and wiring diagrams for the entire system.
Delivery will not be considered complete until all documentation is delivered.
6. Proposals should give information on tape speed, physical size of the units,
power consumption, price, delivery, and warrantees.
There should be a list of recommended spare parts and their
costs and statement should be given about whether transportation
costs are included.
7. Written quotes meeting these specifications should be sent to the
undersigned and should arrive no later than 8 March 1978 at 5:00 PM
Pacific Standard Time.
Cost and delivery will be the primary considerations in vendor selection.
8. The firm that submits the accepted proposal will be expected to enter
into a mutually satisfactory contract that will contain clauses required
by the Armed Forces Procurement Regulations in effect on the date of
contract.
9. Questions concerning technical or business aspects of this request
should be addressed to the undersigned, who can also be reached by
telephone at 415 497-4202.
.sgn
∂21-Feb-78 0647 FTP:Feigenbaum at SUMEX-AIM cs224
Date: 21 Feb 1978 0646-PST
From: Feigenbaum at SUMEX-AIM
Subject: cs224
To: jmc at SAIL, ccg at SAIL
cc: dpb at SAIL
Please call off search for CS224 teacher to replace Cordell. I will do it.
I found a CS101 replacement for myself (high probability anyway).
Ed
-------
∂21-Feb-78 1149 RCM your comments on chapter 4 of my thesis
I believe that my thesis will be sef contained, in the sense that all of the
techniques necessary to do my examples are explained. However, I do not intend
to review the entire field of theorem proving as I did in my masters thesis, and
it is for this reason that I am refering the reader to that source. My masters
thesis is published as MIT-AI-TR-347, and I consider this a legitimate reference.
Our field certainly depends to a very great extent on laboratory technical reports
as a mainstay of the literature.
∂21-Feb-78 1243 FTP:Levin at SUMEX-AIM signature needed
Date: 21 Feb 1978 1241-PST
From: Levin at SUMEX-AIM
Subject: signature needed
To: jmc at SAIL
john, can you stop by my office when you're in POLYA...i tried
to catch you after you class, but missed...need you to
sign papers re: cordell...laurie
-------
∂21-Feb-78 1258 FTP:Levinthal at SUMEX-AIM Jordan Baruch,DOC
Date: 21 Feb 1978 1256-PST
From: Levinthal at SUMEX-AIM
Subject: Jordan Baruch,DOC
To: jmc at SU-AI
John, Would you like to call Baruch and see if you can set up an
appointment for me to see him March 6 or7 or,less desirably MArch
10. Do you have a copy of the automated mining proposal to which
you referred, or does Reynolds?
-------
∂21-Feb-78 2207 FTP:CARL at MIT-AI (Carl Hewitt)
Date: 22 FEB 1978 0108-EST
From: CARL at MIT-AI (Carl Hewitt)
To: CARL at MIT-AI, jmc at SU-AI
John,
Jean Sammet asked Barbara Liskov to please ask to you separate
your paper into two papers by making the appendix on the definition
of LISP into a separate paper with a title and your name.
Sincerely,
Carl
By "appendix on the definition of LISP, does she mean the micro-manual,
and should I include and eval function.
∂22-Feb-78 1012 FTP:Wiederhold at SUMEX-AIM Re: move to AI Lab machine
Date: 22 Feb 1978 1006-PST
From: Wiederhold at SUMEX-AIM
Subject: Re: move to AI Lab machine
To: JMC at SU-AI
cc: les at SAIL, llw at SAIL, lcw at SAIL
In response to your message sent 17 Feb 1978 1635-PST
First answers:
There are 7 people involved, most of them quite part-time , and all
of them students.
I will have to do some work to estimate disk space, there should be
space for about 4-5 substantial programs, with sources and object files.
Some are actually now at sail in some version or other.
Desk space is not essential, especially if reasonable terminal access
exists from campus. It would of course be very helpful
if E could be supported from more terminal types.
SLAC was not ( could not) charge money, but the service was also
weak, unless very heavy compute jobs were needed.
--
I would like to go over things with you, a more general
issue is to what extent to plan joint efforts on the S1 project.
If you have time tomorrow (thursday ) morning, I could come to the
Lab anytime. Gio
-------
∂22-Feb-78 1057 PAT message from Teller
He is sorry, but he won't be able to keep the luncheon appointment
for tomorrow.
∂22-Feb-78 1240 AJT
HI,
I GAVE THE LATEST DRAFT TO BUCHANAN OVER THREE WEEKS AGO, AND
AM STILL WAITING FOR COMMENTS (I'VE GIVEN UP PESTERING HIM FOR
THE MOMENT). CAN I ASSUME THAT, IF AND WHEN
BUCHANAN HAS GIVENB HIS COMMENTS, AND I'VE FIXED UP WHATEVER HE
WANTS CHANGED, THAT YOU WILL BE READY TO GIVE YOUR BLESSING.
NATURALLKY, I'LLL WANT TO GIVE A COPY TO SHEPARD TOO BEFORE
THE LAST WORD IS SAID. IF YOU CAN POKE BUCHANAN WITHOUT
EFFORT OR EMBARRASSMENT, BY ALL MEANS DO SO....
ARTHUR
∂22-Feb-78 1345 FTP:Feigenbaum at SUMEX-AIM is anyone handling this for the department?
Date: 22 Feb 1978 1341-PST
From: Feigenbaum at SUMEX-AIM
Subject: is anyone handling this for the department?
To: dpb at SAIL
cc: jmc at SAIL
Mail from LL-11 rcvd at 22-Feb-78 1232-PST
Date: 22 Feb 1978 at 1527-EST
From: mrl at LL-11
Subject: Visit to Stanford
To: feigenbaum@sumex-aim
CC: mrl
-----
mrl (Michael R. Lowry)
ll11 (Lincoln Lab pdp11)
2/17/78
Dear Proffessor Feigenbaum,
I was admitted to Stanford's PhD program in computer science
last April. I have postponed matriculation for one year in order to
complete a combined masters/bachelors program at MIT. Before making
a final decision as to whether to matriculate at Stanford or continue
at MIT I want to visit Stanford for a week. I would like to:
1)Talk to graduate students and faculty members.
2)Get an overview of Stanford's current research programs in AI.
3)Attend a few classes and seminars.
4)Get some idea of the non-academic aspects of life
at Stanford and in the Bay area.
The best times for me to visitwould be from March 16-23
or April 11-18. If either of these dates meshes with
Stanford's schedule, I would appreciate your putting me in
contact with someone who could help me make arrangements for a
visit (such as living quarters).
Sincerely yours,
Michael R. Lowry
-----
--------------------------------------------------------------
if no one is handling this, Denny would you do something reasonable about
this?
Ed
----------------------------------------------------------------
-------
∂22-Feb-78 1628 JJK Proposal needs your signature
Professor McCarthy -- A proposal to NSF needs your signature as Acting
Chairman. I will leave the proposal and the yellow "Routing Sheet" with
space for your signature in your Polya mailbox, so that you can sign it
and return it to me when you are in Polya tomorrow morning. Thank you!
-- Jonathan King
∂22-Feb-78 1658 FTP:Creary at SUMEX-AIM Rescheduling of Appointment
Date: 22 Feb 1978 1656-PST
From: Creary at SUMEX-AIM
Subject: Rescheduling of Appointment
To: JMC at SU-AI
Would lunchtime tomorrow (12:30 Thurs.) be ok for our discussion? If not, then
what time would you suggest? -- Lew Creary
-------
Thursday lunch would be fine.
∂22-Feb-78 1717 FTP:Hart at SRI-KL (Peter Hart) Ted (?) Hieronymous (sp?)
Date: 22 Feb 1978 1716-PST
From: Hart at SRI-KL (Peter Hart)
Subject: Ted (?) Hieronymous (sp?)
To: JMC at SAIL
John,
Earl mentioned that you spoke of Hiero..etc as a candidate
for the ARPA office. Could you give me any details?
Thanks,
Peter
Incidentally, any comments on my "AI Opportunities" piece?
-------
Jim Hieronymus
735-4832(work), 321-6733 is a PhD in physics from Cornell and worked
on speech recognitiona at the AI Lab for two or three years until
we were definitely not included in ARPA speech. He also did some
vision work. He now works for Amdahl Corporation. I think he knows
quite a bit about AI and is level-headed. I believe he would make
a good impression at an interview.
∂23-Feb-78 0012 MRC
∂22-Feb-78 2321 JMC
I am unhappy about rate of work on Dialnet.
mrc - I've been waiting for those modems to come in. I poked Les
to check on the order. I hope the stuff will come in shortly. I
am just as anxious to have something up as you are. However, I
feel things are at the point now where lack of software is the
stumbling factor; and I cannot write any software until I know how
the hardware is going to work. Consider that it is going to have
an entirely different interface than the current dialer (I started
writing code for the current dialer, only to find out that it would
be totally useless).
Remember that I started asking for the hardware to be ordered quite
a while ago. If when I first said "let's go get the hardware" it had
been ordered it would have been here by now (? maybe I am overoptimistic
about how fast "they" deliver things).
Finally, I have been very ill the past two weeks, although I have not
taken time off from work. Pending the lab results, it could be either
a strep throat or mono(!) or both. It has not prevented me for
continuing to learn about the way things work here, from working on
software in a closely-related field (which will be retrofitted to
Dialnet when that comes about); but it has prevented me from having
any sort of social life or anything resembling rest.
When my modems come in and are hardwarily debugged then you'll see
visible results!
-- Mark
∂23-Feb-78 0110 MRC byte size for Dialnet
To: JMC, LES, GFF, Ryland at RUTGERS-10, RMS, GLS
To: Moon at MIT-MC, JBR, ME, BH
One thing that's constantly popped up in discussions about the Dialnet low
level protocols is my choice of 8-bits as the byte size. There seems to
be a rather large number of people who feel that 8-bits is unacceptable
since many hardware and software configurations cannot implement it.
I am giving serious consideration to abandoning 8-bits and going to 7.
Perhaps I can use the 8th bit as a parity bit or something. I solicit
your opinions on this matter.
The reasons why I selected 8-bits are:
. it is a power of 2. This makes it nice for efficient data
transfer for many machines (8/16/32 bit machines, obvious;
for 36-bit pdp10's or s1's you use 72-bit double words). 7
is only good for ASCII text transfers; admittedly probably
the most common type of transfer. 8 also works with EBCDIC,
although since ASCII is an industry standard and IBM has to
talk ASCII I don't particularly care about EBCDIC.
. it uses all the data bits the modem will send.
. it provides for command codes being 200 and above to 377,
ala the ARPAnet.
Of course, the first reason is the most compelling. The question is
whether or not that gain is worth the problem of not allowing some
people to join the community.
In other news (in case some of you haven't heard the latest), SAIL has
finally ordered a pair of modems and hairy hardware for SAIL and LOTS.
Also, this winner at Columbia University apparently is going to be
setting up an East Coast Dialnet link between a 20 and a Unix pdp11,
AND apparently is going to be talking to us as well. All this equipment
is 1200/1200 baud full-duplex.
Cheers,
-- Mark
∂23-Feb-78 0111 MRC
Thanks for reminding me on something I meant to ask a lot of people (byte size)
∂23-Feb-78 0748 FTP:RYLAND at RUTGERS-10 byte size for Dialnet
Date: 23 Feb 1978 (Thursday) 1049-EST
From: RYLAND at RUTGERS-10
Subject: byte size for Dialnet
To: MRC at SU-AI
cc: les at SU-AI, jmc at SU-AI
Mark:
I strongly feel that Dialnet should stick with 8-bit bytes, for the reasons
you mentioned. It's just not worth it to be incompatible with the rest of
the world, for the sake of a few PDP-10's! Since most Dialnet traffic will
be ascii anyway, it doesn't really matter; if binary data needs to be shipped,
let the FTP programs worry about packing and unpacking data. 'Nuff said.
Cheers,
Chris
∂23-Feb-78 1100 JMC*
Accept Kays' dinner invitation.
∂23-Feb-78 1357 FTP:Levinthal at SUMEX-AIM automated mining
Date: 23 Feb 1978 1348-PST
From: Levinthal at SUMEX-AIM
Subject: automated mining
To: jmc at SU-AI
cc: tob at SU-AI
I spoke to Reynolds about this proposal.He does feel it ought to be
pursued. He has a copy of proposal. He will be glad to give it to
me but wanted to be sure it was ok with you. This message is to
ask if it is ok, or do you have a copy I could look at?
-------
You are welcome to the proposal, and if Reynolds' copy is
accessible, that is more convenient than digging up another.
∂23-Feb-78 1619 REF
Consider the set of all possible worlds, W.
Each is characterized by the following:
The function M(W) which returns the value of m in that world
The function N(W) which returns the value of n in that world
thus:
∀W1 W2.((M(W1)=M(W2)∧N(W1)=N(W2))≡W1=W2)
Here we naturally obtain m as the larger. We also have:
∀W.(1<M(W)<100∧1<N(W)<100))
There are two constants, sx (sum's x... the number given S) and
px (product's x... the number given P). Of course, 4≤sx≤198, 4≤px≤99*99.
There is a predicate A(Person, World, Given number, Situation) that states
that World is possible (reasonable, accessible) for person with given
number in situation.
Thus, in any situation, s, we have:
∀W s.(A(P,W,px,s)⊃M(W)*N(W)=px) ! If W's to be true, it must match px;
∀W s.(A(S,W,sx,s)⊃M(W)+N(W)=sx)
∀W s.(A(S,W,sx,s)⊃∃x.A(P W x s)) ! If W's to be true, there must be some px;
∀W s.(A(P,W,sx,s)⊃∃x.A(S W x s))
There is, of course, a right answer, indicated by the predicate REALITY
∀W.(REALITY(W)≡(M(W)*N(W)=px∧M(W)+N(W)=sx))
∃W.REALITY(W)
When P says "I don't know what n and m are, he is stating:
∃W1 W2.(A(P W1 px s0)∧A(P W2 px s0)∧W2≠W1).
When S says " I know (knew) you didn't know", he is stating:
∀W1.(A(S,W1,sx,s0)⊃(A(P W1 px s0)⊃∃W2.(W1≠W2∧A(P,W2,px,s0)))
When S say "I don't know, either", he is stating:
∃W1 W2.(A(S W1 sx s0)∧A(S W2 sx s0)∧W2≠W1).
Using the general principle, "If a world is possible in situation
si+1, then it must both satisfy everything learned since situation si,
and have been possible in situation si", P can now conclude
∀W s1.(A(P W px s1)≡
1] (A(P W px s0)∧
2] (∃W1 W2.(A(P W1 px s0)∧A(P W2 px s0)∧W2≠W1))∧
3] (∀W1.(A(S,W1,sx,s0)⊃(A(P W1 px s0)∧∃W2.(W1≠W2∧A(P,W2,px,s0))))∧
4] (∃W1 W2.(A(S W1 sx s0)∧A(S W2 sx s0)∧W2≠W1)).
(I've fudged my situations here a little; P stating "I don't know"
should produce a new situation, and S's statement should build on that situation).
When P states "Now I know", he is, of course, implying that there is a unique
W satisfying the above formula.
Now, there are four clauses in the above formula.
The first states that the solution world must satisfy M(W)*N(W)=px, and that
∃x. M(W)+N(W)=x. The second states that there are two different factor
partitions of px. The third states that for all addition partitions of sx,
there are two different factor partitions of px. The fourth states that
there are two different addition partitions of sx.
The usual reasoning can now be applied to the problem.(!?)
∂23-Feb-78 1715 FTP:Feigenbaum at SUMEX-AIM Gio is now an ARPA-IPTO Principal Investigator!
Date: 23 Feb 1978 1711-PST
From: Feigenbaum at SUMEX-AIM
Subject: Gio is now an ARPA-IPTO Principal Investigator!
To: jmc at SAIL, les at SAIL, bs at SAIL, buchanan, lederberg
Mail from USC-ISI rcvd at 23-Feb-78 1556-PST
Date: 23 FEB 1978 1542-PST
From: CARLSON at USC-ISI
Subject: CONGRATULATIONS
To: WIEDERHOLD at SUMEX-AIM, FEIGENBAUM at SUMEX-AIM,
To: SAGALOWICZ at SRI-KL, SACERDOTI at SRI-KL
cc: CARLSON
DR. FOSSUM HAS SIGNED THE ARPA ORDER FOR YOUR KNOWLEDGE BASE
MANAGEMENT SYSTEMS RESEARCH.
BILL
-------
-------
∂23-Feb-78 1725 FTP:Levinthal at SUMEX-AIM SLAIDE
Date: 23 Feb 1978 1724-PST
From: Levinthal at SUMEX-AIM
Subject: SLAIDE
To: feigenbaum, jmc at SU-AI
Date: 21 Feb 1978 1401-PST
From: Levinthal
Subject: missing appendages
To: tob at SU-AI
MEMO: TO POSSIBLE PARTICIPANTS
FROM: ELLIOTT LEVINTHAL
DATE: FEBRUARY 21, 1978
SUBJECT: SLAIDE - Stanford Laboratory of Artifical Intelligence
Developments in Engineering ( see memo below for
other suggested names )
The purpose of this brief memo is bring you up-to-date on
efforts to establish a laboratory such as SLAIDE and to suggest the
next steps that might be taken.
I met with Provost William Miller to get his reaction to the
plans being formulated, as set forth in the attached memo. Miller
expressed interest in the proposal and was quite supportive of its
goals. He felt that the questions I raised about requirements for
Adjunct Professorial appointments could be resolved. He suggested
that the interested principle participants should next meet with
Prof. Gerald Leiberman, vice-Provost for Research, and himself
to discuss implementation.
Prior to that meeting a specific proposal should be
developed. By means of this memo I am asking you to let me know
whether or not you are interested in participating in the generation
of the proposal. If you are, also let me know when during the next
four weeks you will be available to met for at least one hour.
I will be in Washington the week of March 6th in connection with
a NASA workshop of Machine Intelligence and Robotics and hope at that
time to get some idea of the interest and support various
Governemt agencies would have in such a Laboratory.
MEMO TO FILE:
SUBJECT: SLAIDE
FROM: ELLIOTT LEVINTHAL
DATE: FEBRUARY 9, 1978
The proposal is to organize a Stanford Laboratory of
Artificial Intelligence Developments in Engineering (SLAIDE)*.The
focus would be applications of scene analysis and robotics science
to problems of advanced automation,industrial productivity,and
physical rehabilitaion.
The initial impetus for this proposal was the convergence
of interests of the group at Prof. John Mccarthy's Artificial
intelligence Laboratory working with Dr. Tom Binford on vision and
robotics and our work on the Viking mission to Mars. This has
involved us in joint consideration of the problems of automated and
semi-automated three demensional terrain analysis both with regard
to the present problems of Viking data analysis and future planetary
NASA missions.
Discussions of this proposal have been held with Prof. Edward
Feigenbaum and Prof. John McCarthy and it has their support. Prof.
Feigenbaum pointed to the successful collaboration already taking
place between his Department and the Department of Electrical Engin-
eering and which developed an academic program leading to a Master's
Degree. In addition discussions have been held with Prof. William
Reynolds, Department Chairman, Mechanical Engeering, and his
colleagues Professors DeBra, Piziali, and Leifer. Significant
interest was expressed in joint activities.
To make important contributions, such a laboratory would need
expertise in computer science, with an emphasis on vision research
in particular and artifical intelligence in general, mechanical
engineering with an emphasis on robotics, and electrical engineering
focusing on sensors and microcircuitry. To distinguish its function
from the basic activities of the individual departments, the
Laboratory would seek support in a few important problem areas that
require for their solution the synthesis of state-of-the-art skills
in relevant fields.
Among the problem areas that might be condidates for research
and support are the following:
1) NASA - Large Space Systems Technology, Teleoperators,
Manipulators, Fabrication in Space,
Planetary Rovers and Return Sample Missions
Image Analysis for Earth and Planetary Missions
2) DOD Defense Mapping Agency,
Automated photogrammetry and map making, DOD applications
of remote manipulators and scene analysis
3) ERDA - Telemanipulator repair, semi-automated mining (underground
or undersea)
4) HEW - Physical Rehabilitation, prosthetics, "seeing-eye dog"
automated 'servant' for disabled or bed-ridden convalescents.
(Note: The School of Medicine has recently voted to
establish a Department of Rehabilitation Medicine and it is
possible that an Enginering Center will be supported by the
Government at PAVAH.)
5) Industry -robotics for assembly line use and large construction
projcts.
* This is only one of several suggested names. Others are listed
below.
IAR Institute for Automation Research
CAR Center for Automation Research
CARA Center for Advanced Research in Automation
IARA Institute for Advanced Research in Automation
IAR Institute for Automation Research
CAR Center for Automation Research
CRAVE Center for Robotics, Automation and Vision Experiments
CAAR Center for Advanced Automation Research
SCAR Stanford Center for Automation Research
SCARE Stanford Center for Automation Research and Engineering
CAA Center for Advanced Automation
IAA Institute for Advanced Automation
ARC Automation Research Center
SAC Stanford Automation Center
SARC Stanford Automation Research Center
SAAC Stanford Advanced Automation Center
SAARC Stanford Advanced Automation Research Center
ARI Automation Research Institute
SARI Stanford Automation Research Institute
ATI Automation Technology Institute
SATI Stanford Automation Technology Institute
SIAM Stanford Institute for Advanced Manufacturing
SAI Stanford Automation Institute
SPI Stanford Productivity Institute
ART Automation Research Thinktank (tower)
SEAL Stanford Engineering and Automtion Laboratory
CAT Center for Automation Technology
SCAT Stanford Center for Automation Technology
CART Center for Automation Research and Technology
SLAT Stanford Laboratory for Automation Technology
FEAR Facility for Engineering Automation Research
-------
-------
I meant to foward these messages to you.
-------
∂24-Feb-78 1420 LES Vadic modem delivery
To: MRC
CC: JMC
Scheduled for 2 days after April Fool's Day.
∂24-Feb-78 1553 JBR
You're IMLAC seems to be sending ∞ junk to the system. I have
turned of its receiver card in the TTY scanner. Panofsky left about an
hour ago to go skiing. He may be at home still.
∂24-Feb-78 1700 FTP:Levin at SUMEX-AIM contribution to the department
Date: 24 Feb 1978 1639-PST
From: Levin at SUMEX-AIM
Subject: contribution to the department
To: jmc at SAIL
john:
Prof. Herriot gave 50.00 to the Forsythe Memorial Fund...
do you want to say anything special in the thank you letter.
...laurie
-------
nothing special
∂25-Feb-78 1419 LES Arpanet demonstrations
To: Russell at USC-ISI
CC: JMC, TED, JBR, ME, BH
Dave,
Yes, we expect our system to be up during the periods of your planned
demonstrations, namely,
EST PST
Thursday 2 March 78 0800-1100 1100-1400
Monday 6 March 78 1130-1400 1430-1700
Thursday 9 March 78 0800-1100 1100-1400
System maintenance normally begins at 1700 PST. In order to insure that
we do not step on your 6 March demonstration, we will plan to begin any
maintenance on that day at 1730.
Les
∂26-Feb-78 0138 FTP:PRATT at MIT-AI (Vaughan Pratt) Popular Electronic's story on QUASAR
Date: 26 FEB 1978 0438-EST
From: PRATT at MIT-AI (Vaughan Pratt)
Subject: Popular Electronic's story on QUASAR
To: MGW at MIT-AI
CC: MINSKY at MIT-AI, JMC at SU-AI
I would think that the appropriate thing in this case would be a
carefully worded letter to the editor that briefly documented two
or three instances of suspicious activity on QUASAR's part, and
then stated flatly that the thing is a fraud and promises features
beyond the wildest dreams of those presently at the cutting edge of
research in this field.
Make sure the whole thing is letter-to-the-editor size, or you may
find you best paragraph cut.
∂26-Feb-78 1514 JJK New NSF proposal
Proposal is located on PRO.ZIL[CSD,JJK]
∂26-Feb-78 1535 JJK
To: JMC, JMB, SSO, JLH, wiederhold at SUMEX-AIM
To: feigenbaum at SUMEX-AIM
∂26-Feb-78 1522 JMC statement from faculty about proposal
I will approve the proposal if I get a statement from the Systems
Faculty that each of them has seen it and agrees that it is a good
bet for the use of our limited ability to extract instruction
aiding money from NSF. Net messages or phone calls will suffice,
but they should be based on having seen the proposal. By the way,
exactly who is included in "systems faculty"?
-------------
I am forwarding this message to Jeff Barth and Gio Wiederhold who can
reply directly to you. They are the only ones cited in the proposal in
order to keep it strictly a Computer Science Department proposal.
However, I have also consulted with John Hennessy, Sue Owicki, and Bill
van Cleemput, who will read the proposal too. I will ask them to address
comments to both of us. (Jeff, Gio, John, Sue -- proposal is on the file
PRO.ZIL[CSD,JJK]).
∂26-Feb-78 1730 FTP:HEARN at DEC-MARLBORO lisp history paper
Date: 26 Feb 1978 2023-EST
From: HEARN at DEC-MARLBORO
Subject: lisp history paper
To: jmc at SAIL
John, do you have an update of your LISP history paper online at SAIL
which I could look at, and would you mind if I sent it to a couple of
people who are interested in the subject?
thanks,
tony
-------
The current version is and will be LISP[F77,JMC]. I am about to make
some remarks claiming that one of the reasons for the longevity of
LISP is that it is peculiarly suitable for making other systems like
REDUCE and FOL and QA4, because it provides a command language for
testing things but mainly because LISP is an easy target language
for program generators from yet higher level languages. Do you
have any comments. Jean Sammett is pressing me for a final version
of the paper, so only quick comments will be influential.
∂26-Feb-78 1731 JLH
To: JMC, JJK, WVC, JMB, JLH
In response to proposal PRO.ZIL[1,JJK].
I approve of the proposal and have made several comments to
Jonathan. I feel the idea of introducing nonCS types
to microcomputers is advantageous, both to NSF and
to the students.
Jonathan asked me to clarify the use of the EE Microcomputer
Laboratory facility. This facility supports
two regular courses and several independent projects.
The courses are oriented towards people who want to understand
and design microprocessor systems. They include both hardware and
software. The courses are EE281 Microcomputer Lab and
EE315 Adv. Microcomputer Lab. The former we anticipate
will register 100 students/year while the latter
will take about 75 (including a summer offering for the later).
Therefore, these facilities (currently 2 Z80 development
systems - shortly 4) can hardly support additional use.
John Hennessy
∂26-Feb-78 1902 REM
To: CRUNCH.DIS[1,REM]:
Preliminary draft of my paper describing my new data-compressing algorithm
is now available for your personal reading, filename INTCOM.WRU[1,REM].
(No login or password is needed to TYPE or FTP it from here to whereever
you are located.) Please don't distribute it to outsiders, I want maximum
chance to patent the algorithms and/or applications of them, and to get
the paper published quickly. Any suggestions??
∂26-Feb-78 1941 DCL
To: GROUP.DIS[VCG,DCL]:
***********************************************************************
VERIFICATION GROUP SEMINAR WEDNESDAY 28TH. FEBRUARY
PLACE: ERL 237
TIME: 2:30 pm.
TITLE: A Model of Access and Synchronization in Systems of Concurrent Processes
SPEAKER: Dick Karp
*******************************************************************
∂26-Feb-78 1952 DCL
To: GROUP.DIS[VCG,DCL]:
***********************************************************************
************************************ CORRECTION **********************
VERIFICATION GROUP SEMINAR TUESDAY 28TH. FEBRUARY
PLACE: ERL 237
TIME: 2:30 pm.
TITLE: A Model of Access and Synchronization in Systems of Concurrent Processes
SPEAKER: Dick Karp
*******************************************************************
∂27-Feb-78 0148 MRC via SU-TIP#1 IMSSS line
To: JMC, LES
Since it won't be until April when the modems arrive (and who knows how long
until the necessary hardware kludgery is done), I'd like to install Dialnet
on the IMSSS line. It seems a perfect thing to fool around with and get a
lot of the really hairy stuff done; and Dialnettifying the IMSSS line doesn't
seem to be much of a time waste since I guess 99% of an IMSSS implementation
would be needed for the SAIL/LOTS implementation.
I don't intend to worry about IMSSS/SUMEX; they can always copy my IMSSS
sources over there and bring up Dialnet protocol in parallel.
What do you think? Even if the IMSSS line gets flushed eventually, it will
be fun to do and I would get a lot of progress on Dialnet. Question: who is
the bureaucrat over there to talk to about setting up a directory for me
and getting me whatever access is needed to the linker? The <SUAI> directory
there has this epsilon disk allocation and apparently a lot of random losers
use it.
I'd still be at the lab most of the time since IMSSS still does not have Tenex
EMACS and unless they do I won't edit there.
∂27-Feb-78 0200 MRC via SU-TIP#1 IMSSS continued
To: JMC, LES
of course, this means parallel programs have to be written for here as well,
so this idea is half IMSSS programming and half SAIL programming.
∂27-Feb-78 0959 FTP:Levinthal at SUMEX-AIM mtg with teller
Date: 27 Feb 1978 0959-PST
From: Levinthal at SUMEX-AIM
Subject: mtg with teller
To: jmc at SU-AI
could make it tonight. how about before sense mtg tuesday ( is that
mtg still on). Other alternate dates are after March 13th.
-------
Let's make it tonight at 8 then. Teller's address is
943 Casa Nueva (RONO on letter box) at Stanford.
∂27-Feb-78 1149 JMB Zilog teachine proposal
To: JMC
CC: JJK, WVC, JLH, SSO
I have read the NSF proposal for undergraduate teaching funds and
think it is well founded and should be submitted.
....Jeff
∂27-Feb-78 1207 FTP:Levinthal at SUMEX-AIM mtg
Date: 27 Feb 1978 1207-PST
From: Levinthal at SUMEX-AIM
Subject: mtg
To: jmc at SU-AI
ok at Tellers. Can you help. I am trying to chase down Reynolds
copy of proposal. He is out of town. Who was PI? What agency
was it sent to ? About when was this done? Thanks.
-------
I was PI. My secretary, Patte Wood, will try to find a copy and
call you or send you a message.
∂27-Feb-78 1628 FTP:Levin at SUMEX-AIM faculty who login regularly
Date: 27 Feb 1978 1628-PST
From: Levin at SUMEX-AIM
Subject: faculty who login regularly
To: jmc at SAIL
jeff barth@sail
bruce buchanan@sumex
ed
cordell@sail
don knuth@sail
josh lederberg@sumex
susan owicki@sail
gio wiederhold@sumex and sail
terry winograd@parc
john hennessy@sail
...ll
-------
∂27-Feb-78 2020 DPB
To: feigenbaum at SUMEX-AIM
CC: JMC
∂22-Feb-78 1345 FTP:Feigenbaum at SUMEX-AIM is anyone handling this for the department?
Date: 22 Feb 1978 1341-PST
From: Feigenbaum at SUMEX-AIM
Subject: is anyone handling this for the department?
To: dpb at SAIL
cc: jmc at SAIL
Mail from LL-11 rcvd at 22-Feb-78 1232-PST
Date: 22 Feb 1978 at 1527-EST
From: mrl at LL-11
Subject: Visit to Stanford
∂27-Feb-78 2021 DPB
To: feigenbaum at SUMEX-AIM
CC: JMC
[Sorry about the previous non-message. I will handle the Lowry visit. -Denny]
∂27-Feb-78 2029 DPB
To: REP, AZS
CC: FFY, JMB, GHG, JMC
∂23-Feb-78 1308 CSD Winter Student Meeting
Alfred and I are working up an agenda for the Winter Student meeting.
Is there anything you particularly need students for. I know we need members
for the comp. committee. Do you have the faculty members yet?
Rich
Spring 5/78: F.Yao, Barth, Golub, McCarthy
Above are the faculty members. There has been sentiment for having
more than two student members. I agree with the sentiment in principle.
-Denny
∂27-Feb-78 2132 DPB New DataMedia package available
To: BPM, TED, LES, JMC, JMB, SSO, JLH, EJM, BS, GIO
To: kanerva at SUMEX-AIM, rindfleisch at SUMEX-AIM
To: buchanan at SUMEX-AIM, wilcox at SUMEX-AIM
To: kahler at SUMEX-AIM
Martin Hardy of Product Associates in Redwood City has been working
on software for the new micro-processor based DataMedia terminal.
He is prepared to deliver the following package immediately (modulo
the normal 60-day DM lead time) :
DM 3025 CRT $1700
(less educational discount) -340
Keyboard (similar but not identical
to the SUMEX keyboard)
plus his program $ 200
Prentice 1200/150 modem $ 325
Installation of modem, switch, cable $ 75
------
Total price (approximate) $1960
The software will make the 3025 look like our modified 2500A DMs both
to the user and to the host computer software. (I.e. Bold face, no
protection, no parity instead an EDIT key, etc.) These terminals will
run TVEDIT and E without change to host software.
The keyboard will be a modified version of the standard keyboard.
That is, in addition to the main key pad, there will be a top row of
special keys. We can specify what codes we want under those. Therefore,
what is now the special row to the left of the main pad on the SUMEX
kbd could be included in the upper row. The normal numeric pad will
be replaced with the same pad as is normally on the SUMEX keyboard, except
we will have a few more keys to "play with."
Some (Ted Panofsky at SAIL, Pentti Kanerva at IMSSS) have suggested that
in the long run we will want to design a new "smart" keyboard. Hardy is
interested in building what we want. In the short run, however, we will
order keyboards close to identical to the SUMEX keyboard.
This message is being sent to as many people as I know who are or might
be interested is this package. Please forward to any other interested
parties. -Denny Brown (DPB@SAIL)
∂28-Feb-78 1300 HVA Noisy Phone
I reported phone problem to repair service this morning; they believe the
trouble is on the line, and not in the instrument, and they are attempting
to fix it.
∂28-Feb-78 1516 FTP:Buchanan at SUMEX-AIM Arthur Thomas
Date: 28 Feb 1978 1516-PST
From: Buchanan at SUMEX-AIM
Subject: Arthur Thomas
To: jmc at SU-AI
I told Arthur that his thesis was acceptable as it stands. It is
not a tidy package that is easy to evaluate, but he's put enough time
in on it and more time won't help much. There are still missing
dates on references (and other missing items) that he says you will
insist on having filled in. But I'll sign off on it when asked.
bgb
-------
∂01-Mar-78 0000 JMC*
Ask Patte for Dijkstra's anti-Perlis blast.
∂01-Mar-78 0836 CET
To: MKA, JMC
February 28, 1978
Max Agoston
638 La Grande, Apt. 2
Sunnyvale, CA 94087
Dear Max:
Your magic number: 4833
You have passed the Computer Science Comprehensive Examination
with a grade meeting the requirements of the department's M.S.
program.
You have met all of the department requirements for the Master's
Degree, and your name has been added to the list of Degree Candidates
for Winter Quarter.
Congratulations on a fine performance!
Sincerely,
Denny Brown
Associate Chairman
/cet
∂01-Mar-78 1000 JMC*
Call Panofsky.
∂01-Mar-78 1300 RDR* via AMET
LOTS should be putting terminals in student residences
where usage warrants the terminal(s).
∂01-Mar-78 1330 FTP:Feigenbaum at SUMEX-AIM Re: Henrici and Moler appointments
Date: 1 Mar 1978 1323-PST
From: Feigenbaum at SUMEX-AIM
Subject: Re: Henrici and Moler appointments
To: BS at SU-AI
cc: jmc at SAIL
In response to your message sent 1 Mar 1978 1024-PST
50% appointments for Henrici and Moler are OK because they will be
taking on the full range of academic responsibilities (like Ph.D. advising,
exam preparation, etc.) and not JUST teaching a course.
Ed
-------
Glad you answered this inquiry. I felt I couldn't say yes to Golub on
my own because it involves next year's budget.
∂01-Mar-78 1626 WLS PRINLEVEL vs UWRITE
To: MACLSP.DIS[AID,RPG]:
----→ W A R N I N G ←-----
If PRINLEVEL or PRINLENGTH are non-NIL while you are doing output
to a file (via UWRITE), you may get garbage in the file. Specifically, if you
are printing an expression that contains a CONS with numeric CDR, a random
number may be printed into the file at that point.
The MACLISP wizards know about this, and it should be fixed soon.
[WLS]
∂02-Mar-78 1100 JMC*
Sundesert statement to George Pake.
∂02-Mar-78 1448 FTP:CARL at MIT-AI (Carl Hewitt)
Date: 2 MAR 1978 1748-EST
From: CARL at MIT-AI (Carl Hewitt)
To: CARL at MIT-AI, jmc at SU-AI
John,
I will be in Palo Alto March 27 to March 31. Could we have dinner some
evening that week.
Thanks,
Carl
I presently expect to be here from the 27th to 31 and will be glad to have
dinner or lunch.
∂02-Mar-78 1600 FTP:Feigenbaum at SUMEX-AIM library space in Jacks Hall
Date: 2 Mar 1978 1559-PST
From: Feigenbaum at SUMEX-AIM
Subject: library space in Jacks Hall
To: jmc at SAIL
John:
Ed would like you to critique the following:
March 2, 1978
To: Edward Shaw, Associate Provost
Amy Blue, Assistant Provost
From: Edward A. Feigenbaum, Chairman, Computer Science Department
Subject: The CSD Library location decision
Dear Ed and Amy,
The prospect of a decision to revoke the previous plan of housing the
Math and CS libraries together in Math Corner is causing serious
dismay among the CS faculty. This is seen as the biggest departmental
issue to come our way in a very long time. We oppose, in the
strongest possible terms, the change of plans being considered. I
would like to write out some of our reasons. Bob Floyd, the former
Chairman, has some financial arguments he will make to show that the
savings from the plan change are either small or negative; and in any
event the change is short-sighted and will severely damage our
academic program.
The Department, with pain but with a sense of the feasible, agreed to
meet its needs with much less space than originally planned so as to
realize relatively quickly its overriding goal: bringing the Computer
Science Department together so that it functions as an integral
intellectual unit. It was further painful to have to relinquish one
and a half floors of the new building to the Boys Town Center for the
obvious financial reasons. What we have left is marginally adequate
for our needs. Even the growth since the original "body count" specs
that were used in the design has put severe pressures on the space,
and we are not even in the building yet! Our research support has
grown steadily as the eminence of the department has grown and as the
faculty has shown a small but steady growth.
How does the removal of 2200 sq. ft. of space impact our academic
program Here are some examples:
1. Cutback in space for our Ph.D. students (fellowship holders and
Research Assistants). The quality of their currently superb education
will suffer, as will the quality of their contribution to our many
research projects (their contribution is very great).
!
2. Cutback in space for our professional Research Associates. These
talented and often highly-paid doctoral level professionals not only
provide full-time brainpower that helps make our research so
successful; but also provide consultation, training, and research
supervision for our graduate students to supplement faculty efforts.
Each such professional is "budgeted" an (inadequate) 80 sq. ft. in
the current plan, and the number of such professionals will have to
be cut back now and in the future if we suffer another space loss.
3. Our ability to house distinguished visitors will be severely cut
back. Because of the Department's faculty, projects, and general
reputation, we attract some of the most eminent scholars in the
discipline. The come here on leave and significantly enrich our
scientific and teaching environment (Prof. Jack Schwartz from NYU is
a current example; Prof. Henrici, the great numerical analyst from
Switzerland, coming next year, is another example).
4. Our continuing efforts to integrate more closely with our EE
colleagues of our joint Digital Systems Laboratory will be hampered
by our inability to house enough of them and their students to make a
difference, i.e. to get above the level of "tokenism". This goal
already has had to be compromised when a study we did last year
showed that it was impossible to house all of DSL with us. But we
were able to do something worthwhile. Even this would be compromised
by a further cutback.
Let me spare you the complete list of examples and close on another
note. For the first time in its history, the Computer Science
Department feels that it is being treated unfairly. As a consequence,
department morale is at its lowest point. The process of arriving at
a space plan is, as we all know, a very complicated one in which the
faculty and the university administration arrive at a negotiated
common view of what needs the space will satisfy and how the space
shall be allocated. There are many compromises and trade-offs made
during the many months of this informal negotiation. In the end, a
"package" emerges to which all lend their support, though each may
have his misgivings about individual items. We regard it as unfair
unilaterally to revise one element of the package, at the same time
keeping all the other constraints bound. We have hesitated to bring
up any but the most trivial changes to the package because we did not
want to destabilize the administrative and financial situation you
face. But you have not hesitated to propose destabilizing the
academic situation we face. The fact is that almost none of decisions
previously made are fluid; hence there is almost no chance for us to
recoup from the space crisis (disaster, as we see it) that will
attend a revision of the library housing plan.
!
In summary, I want to reiterate the axiom of the Computer Science
Department space planning. All of us--Miller, the President's
Advisory Committee, Forsythe, Floyd, Feigenbaum, Knuth, and the rest
of the faculty--view bringing the department together as the most
important space planning goal. The goal is hard to achieve because
the department is large. Its academic plus research annual budget is
in the 3-3.5 million dollar range. Removing 2200 sq. ft. from the
department's new home will make it very difficult, perhaps
impossible, to achieve the department's primary space goal and
maintain its present degree of excellence and eminence.
cc: W.F. Miller, Provost
H. Royden, Dean of H&S
G. Lieberman,Vice Provost
...thanks...laurie
-------
Your space letter looks good to me.
∂03-Mar-78 0116 LLW
To: LES, JMC, JBR, ME, TED
CC: LLW, TM, PMF
SAIL-LLL COORDINATION
It seems manifestly desirable that the monthly coordination
meetings between SAIL and LLL on the S-1 contract be as pleasant
and productive as possible. It is also the case that Curt Widdoes
will apparently be at SAIL daily and on a more-or-less full-time
basis, through the end of this coming Summer.
I have therefore asked him to represent LLL interests and
points of view (and me personally) with respect to all SAIL-LLL
contract coordination matters, and he has agreed to do so. He
has agreed to consult with me on all matters on which he has any
doubt as to my views or the LLL position before taking such matters
up with SAIL staff members, and I have agreed to stand behind any
and all representations he may make on behalf of LLL (or me personally).
I will of course review the motivating circumstances of all such
representations, as well as the representations themselves, with
Curt frequently, and will expect to have the major ones recapitulated
for whatever level of discussion seems appropriate at our monthly
coordination meetings (which I will invariably attend).
I believe that having someone of Curt's capabilities available at
SAIL essentially continuously for coordination purposes will completely
eliminate the possibility of any major unpleasant surprises for
either SAIL or LLL at any of our monthly coordination meetings, as well
as upper-bounding possible wasted motion to very small values;
I presume that all of you will welcome this step with enthusiasm
comparable to mine in taking it.
Lowell
∂03-Mar-78 0204 MRC host-host protocol
To: JMC, LES, Ryland at RUTGERS-10
I have added two new op-codes; EOF and INT. EOF means end of file, and
INT means interrupt. INT's meaning is up to the process using it. The
processes for which these codes were implemented, file transfer primarily,
would use EOF to mean "end of file, begin of protocol commands" and INT to
mean "abnormal end of data, begin of protocol commands, abort transfer".
The reason for adding these codes should be obvious and corrects a serious
deficiency in the protocol.
I am still recovering from a long illness (this is the third-fourth week);
it turned out to be mono. I'm not supposed to be working now, but I think
I can take it easy and get things going on this without a relapse.
Chris, what is the state of your hardware over there? Have you been able
to order your modems yet? I hope the cost isn't too prohibitive for you;
but with this configuration it should be able to run on Twenex pretty much
as an ordinary user process with ordinary terminal service without having
to use IOT luser mode or making monitor changes. Of course in the end we
will want to have it do magic wheelish things just for efficiency, but a
non-wheel implementation should be possible (I don't know how accessible
wheel is to you).
Cheers,
Mark
∂03-Mar-78 0315 100 : REM My super-winning new data-compression partial-bit algorithm
To: JMC, HPM, REM, RWG
Just a couple minutes ago I realized another super winning thing about
my interval-subdivision (with slightly-massaged histogram) algorithm.
I had drawn a data flow chart, showing the main loop of feedback during
crunch and during uncrunch, and noted an explicit delay of one token in
crunch loop not present in uncrunch loop, thus crunch can be done in
parallel whereas uncrunch must be done serially -- !! realized that if
artificial additional delay is interposed in a trivial way, both crunch
and uncrunch can use parallel processing to any desired degree (although
crunch will always have exactly one more unit of parallelism available than
uncrunch will) !! Thus, at slight cost in effecttiveness of compression,
a parallel-processor can zip through the data super-rapidly (speed is
linearly proportional to number of parallel units online). !! -- More later
when I get my flowcharts and modularized software documented online...
∂03-Mar-78 0639 DON Math Monthly
I'm trying to check a reference in an old Math Monthly, and PMF said he
thinks you might have some old issues stashed away down in the basement.
I checked the area near the entrance to the "wind tunnel", but there's
too much junk down there for me to look through it all. Do you indeed
have some old Monthlies down there, and if so can you describe roughly
where they are? (And do you mind if I look through them?)
I used to subscribe to the Monthly, stopped some years ago, and am just
starting again. You're welcome to look, but the Library files them
nicely.
∂03-Mar-78 0642 DON Feigenbaum(e)
I noticed that you had some mail refused due to your having misspelled
Feigenbaum's name. You might find it easier to mail the stuff to EAF
here at SAIL--it's automatically forwarded to FEIGENBAUM%SUMEX.
∂03-Mar-78 0952 FTP:Levin at SUMEX-AIM National Research Council site visit
Date: 3 Mar 1978 0952-PST
From: Levin at SUMEX-AIM
Subject: National Research Council site visit
To: jmc at SAIL
john:
The National Research Council has appointed a committee to study
the policy implications of the changing role of postdoctorals and
doctoral research staff in higher education and research in the U.S.
Dr. Frederick Balderston and perhaps one other committee member will
be participating in a site visit to Stanford, March 13-15.
They would like to have a one-hour discussion with faculty
("perhaps the chairman and one or two other members").
Bob Tarjan has reluctantly agreed to participate...
Shall I count you in or out?
....laurie
-------
If the meeting were early afternoon on the 13th, I could certainly
make it. Otherwise, schedule it, and I'll come if I can. I could
also come if it were at the AI Lab.
∂03-Mar-78 0946 RWW NEWFOL
To: FOL.DIS[W78,RWW]:
There is a new version of FOL up. It is entirely compiled (and thus
faster) and contains all the new features: proof switching, l/s pair
switching, eval, and multiple sort definitions for functions. The
documentation for these is forthcoming. It has not been tested very
extensively so there may be bugs. As soon as it has a shake out period
we will make this the official FOL. We also hope to work at making it
a bit smaller. Please report bugs to me or use the new gripe
feature. I am thinking of giving a seminar on how to use all this next
quarter is anyone interested???
rww
∂03-Mar-78 0953 RWW newfol
To: FOL.DIS[W78,RWW]:
Sorry I forgot to say how to run the new FOL.
R NEWFOL
does it.
rww
∂03-Mar-78 1337 FTP:Feigenbaum at SUMEX-AIM We won!
Date: 3 Mar 1978 1327-PST
From: Feigenbaum at SUMEX-AIM
Subject: We won!
To: jmc at SAIL, dek at SAIL
During the morning meeting, Ed Shaw made the decision NOT to house the
library in MJ Hall on the basis that our programmatic requirements
for space would not allow it.
I presented my memo and a verbal analysis of our space crunch. Bob Floyd
presented a financial analysis of the planned move. (Shaw called my memo
"a smoking gun").
Royden came but did not say much.
Don, I will send you an electronic copy of the memo.
Ed
-------
∂03-Mar-78 1430 FTP:Levin at SUMEX-AIM a favor to ask Vera
Date: 3 Mar 1978 1430-PST
From: Levin at SUMEX-AIM
Subject: a favor to ask Vera
To: jmc at SAIL
John:
Ed has to deliver a paper to IBM, San Jose by Tuesday morning,early.
Would you mind asking Vera if she could take it for us?
Thanks VERY much...laurie
-------
Sure, I'll ask, and I see no obstacle.
∂03-Mar-78 1438 FTP:Feigenbaum at SUMEX-AIM AI qual for end of Spring
Date: 3 Mar 1978 1438-PST
From: Feigenbaum at SUMEX-AIM
Subject: AI qual for end of Spring
To: jmc at SAIL, winograd at PARC, buchanan
I would like to offer the AI qual this Spring along the lines indicated
in the enclosed announcement. The experiment tried two years ago, offering
the short articles as an option, was popular and highly successful (i.e.
the students said that they found it an excellent learning experience).
The new plan is the result of my consulting with a group of students, some
of whom are planning to take this year's qual, some of whom had taken the
optional exam two years ago. The resulting written material may provide
an important boost to the AI-Handbook project.
I would be willing to serve as chairman of the AI Qual
committee if my plan is acceptable (or some reasonble variant thereof).
Please let me know how you feel about this, soon if possible so the announce-
ment can go out.
Ed
-----------------------------------------------------------------
The 1978 AI Qual
The AI Qual this Spring will involve a writing requirement as well
as a one hour oral exam. Students will be required to prepare four
encylopedia-like summaries of AI research topics, two in each of two
major areas of the AI field. Drafts of these papers, each 3-5 pages in
length, will be reviewed prior to the oral exam by members of the faculty.
Final drafts must be approved by the AI Qual faculty committee for an
unconditional PASS on the Qual. Papers will then be reviewed for
suitability for the AI Handbook, which is being prepared by students
and faculty in the department.
We feel that the preparation of clear, written expositions of
AI concepts, techniques and systems in a style that is suitable for
reading by a non-AI oriented computer scientist or student is an
excellent demonstration of understanding of the material. Since the
papers must discuss the relation of the material covered to other
techniques and systems in the same area of AI research, writing such
a paper requires a deep understanding of issues and history over a
broad range of AI research. An experiment offering such writing projects
as an option for the AI Qual was done in 1976, and was an educational
success. This year we are incorporating such work as a regular part
of the exam.
All students interested in taking the AI Qual this Spring should
sign up with Carolyn Taynai in the department office. There will be a
meeting during the week of March 13-19 at which students will be given
their writing assignments, including summaries of what is to be covered
in each of their four papers. Drafts will be due at the end of May, a
week or so before the oral exam in early June.
-------
Congratulations on the space victory. I agree with your proposal for
conducting the qual and hereby appoint you chairman of the AI qual
committee subject to confirmation by the Department Chairman when
he resumes office on April Fool's Day.
∂03-Mar-78 1449 FTP:Feigenbaum at SUMEX-AIM LISP timing experiments
Date: 3 Mar 1978 1443-PST
From: Feigenbaum at SUMEX-AIM
Subject: LISP timing experiments
To: jmc at SAIL, les at SAIL
Date: 3 Mar 1978 1422-PST
From: Rindfleisch
Subject: FYI - LISP TIMINGS
To: SWEER, LEDERBERG, FEIGENBAUM, BUCHANAN, SMITH
Mail from SRI-KL rcvd at 3-Mar-78 1025-PST
Date: 3 Mar 1978 1025-PST
From: Lynch at SRI-KL (Dan Lynch)
Subject: Lisp Benchmark Timing across many Machines
To: Hart, Sacerdoti, Raphael, Heathman, MMcM, Victor,
To: Ellis at ISIB, McKinley at ISIB, Rindfleisch at SUMEX-AIM,
To: Lynch
Recently I got timing information from DEC on the 2020. I had them run
the Lisp "benchmark" job that I have been running for years on various
machines. That prompted me to run the same test on as many machines
as possible (by type) to get comparison information. The table below
summarizes the simple throughput data that were collected. All tests
were run on unloaded machines. Notice the difference between 256K
and 512K of memory on the KA machine. There may appear to be some
anomalies in the peformance data on the 2020. I have no explanation.
It looks like the 2020 is at least as good as a KA.
Here is the data:
Lisp Benchmark Timing Data
KA KA 2020 KI KL-1090T
256K 512K 512K 512K 1024K
Function
Build (ms) 31.8 29.9 25.0 16.7 5.1
(ratio) (0.94) (1.00) (1.20) (1.79) (5.86)
Sort (ms) 178.7 140.8 158.3 85.2 30.9
(ratio) (0.79) (1.00) (0.89) (1.65) (4.56)
Slosh (ms) 123.2 112.9 99.7 69.0 21.7
(ratio) (0.92) (1.00) (1.13) (1.64) (5.20)
-------
-------
-------
Thanks for the info on LISP timing. Did you get the message congratulating
you on space victory? I mailed it to EAF here, and it is supposed
to be forwarded automatically.
∂03-Mar-78 1510 FTP:Feigenbaum at SUMEX-AIM (Response to message)
Date: 3 Mar 1978 1509-PST
From: Feigenbaum at SUMEX-AIM
Subject: (Response to message)
To: JMC at SU-AI
In response to your message sent 3 Mar 1978 1453-PST
I got it John. Thanks.
The April fool.
---------------------------------------------------------------------------
FOOLS RUSH IN WHERE NOBEL LAUREATES HAVE TREAD
Have you heard? I'm taking over the PIship of SUMEX on July 1 when Josh
leaves. Department+Heuristic Prog. Project+SUMEX= chaos in EAF's life.
-------
∂03-Mar-78 1512 FTP:Feigenbaum at SUMEX-AIM familial affairs
Date: 3 Mar 1978 1512-PST
From: Feigenbaum at SUMEX-AIM
Subject: familial affairs
To: jmc at SAIL
Congratulations! I heard via a very indirect grapevine that Suzie is getting
married. Please send my best wishes to her.
Time sure does march on.
Ed
-------
∂03-Mar-78 1621 FTP:RYLAND at RUTGERS-10 things
Date: 3 Mar 1978 (Friday) 1921-EST
From: RYLAND at RUTGERS-10
Subject: things
To: MRC at SU-AI
cc: les at SU-AI, jmc at SU-AI
mark,
sorry to hear you've been ill; i've only had a short bout with the russian
flu.
ok on the h/h protocol mods.
actually, i was going to ask if i could bite the ftp chunk as a more or less
independent project for the next coupla months. i've had some experience
with it, but still don't have the protocol handbook yet (would it be possible
for you to get me even an old one?).
as for implementation, i'd certainly to be willing to go the effciency route,
even if it's 'wheely'; i'm more or less the system hacker here (we've hacked
up our monitor/exec a bit). but i'd be disappointed if we have to go the
monitor-hacking route. as you say, we probably don't have to. more thots
on this a bit later. right now, I'm trying to dig out from underneath a big
pile of work, both here and school. should get some useful work done on
protocol thinking next week, tho. get better! cheers
--chris
∂03-Mar-78 1635 FTP:RYLAND at RUTGERS-10 things
Date: 3 Mar 1978 (Friday) 1921-EST
From: RYLAND at RUTGERS-10
Subject: things
To: MRC at SU-AI
cc: les at SU-AI, jmc at SU-AI
mark,
sorry to hear you've been ill; i've only had a short bout with the russian
flu.
ok on the h/h protocol mods.
actually, i was going to ask if i could bite the ftp chunk as a more or less
independent project for the next coupla months. i've had some experience
with it, but still don't have the protocol handbook yet (would it be possible
for you to get me even an old one?).
as for implementation, i'd certainly to be willing to go the effciency route,
even if it's 'wheely'; i'm more or less the system hacker here (we've hacked
up our monitor/exec a bit). but i'd be disappointed if we have to go the
monitor-hacking route. as you say, we probably don't have to. more thots
on this a bit later. right now, I'm trying to dig out from underneath a big
pile of work, both here and school. should get some useful work done on
protocol thinking next week, tho. get better! cheers
--chris
∂04-Mar-78 1304 DCL
To: GROUP.DIS[VCG,DCL]:
***********************************************************************
VERIFICATION GROUP SEMINAR TUESDAY 7TH. MARCH
PLACE: ERL 237
TIME: 2:30 pm.
TITLE: CRITIQUE OF IRONMAN SPECIFICATIONS
DISCUSSION Please prepare whatever comments we all feel like making
about the four DOD1 preliminary language designs. We will try
to summarize these in a short report for ARPA. This has to
be ready by Friday March 10th.
NEXT WEEK, 14th: Jim Honing will talk.
*******************************************************************
∂04-Mar-78 1629 FTP:Creary at SUMEX-AIM Talk on Circumscription
Date: 4 Mar 1978 1629-PST
From: Creary at SUMEX-AIM
Subject: Talk on Circumscription
To: JMC at SU-AI
I just discovered yesterday in my briefcase, much to my regret, the note
that was intended to remind me of your talk last Wednesday to the statistics
department. The talk completely slipped my mind -- I even permitted a
conflicting meeting to be scheduled for Wednesday afternoon on short notice.
I hope that I'll be able to get at least a brief verbal report on how
the talk was received and what major points were made. -- Lew
-------
The talk turned out to be to the IMSSS seminar, but some statistics
people were there.
∂05-Mar-78 1735 FTP:HEARN at DEC-MARLBORO Your LISP paper
Date: 5 Mar 1978 2032-EST
From: HEARN at DEC-MARLBORO
Subject: Your LISP paper
To: jmc at SAIL
I agree with you that the reason lisp is suitable for
systems like REDUCE etc is that it is an easy and highly convenient
target language from yet higher level languages. One might criticize
its dependence on binary trees or the form of its evaluator, but
these are minor in the face of all the available lisp software.
If you wanted to be very kind to me, you might reference
the latest Standard LISP Report I sent you recently at the point
where you refer to "standard lisp". The full reference is:
MARTI, J. B., HEARN, A. C., GRISS, M. L and GRISS, C; Standard LISP
Report, University of Utah Symbolic Computation Group Report No 60,
January 1978.
I enjoyed your paper, and feel that it gives a fair survey
of the "history".
tony
-------
∂07-Mar-78 1055 BS NIH Conference
CC: JMC, Feigenbaum at SUMEX-AIM
On Monday and Tuesday, March 13 and 14, there is to be an NIH presented
conference on Administration of NIH Grant Programs. The conference will
be on campus, will take the better part of both days, and will cost $15.
I would like permission to attend.
Betty
Go ahead and attend the conference.
∂07-Mar-78 1101 FTP:Levin at SUMEX-AIM National Research Council meeting
Date: 7 Mar 1978 0944-PST
From: Levin at SUMEX-AIM
Subject: National Research Council meeting
To: jmc at SAIL
john:
can you see them on the 15th of March? How
about 10:30? Serra House? All the other people
involved are down here, that's the reason Serra was chosen.
...laurie
-------
∂07-Mar-78 2317 MRC Teleconferencing
To: LES, JMC
I recently agreed to be part of a group that Geoff Goodfellow is organizing
regarding teleconferencing; ie, extending the concept of linking and party
linking one step further to having the computer be both communications
medium and secretary, recording minutes of the conference, etc. Any
subsystems developed along this line would obviously be display-oriented,
so that several users could talk at the same time, etc.; much like a linking
system Les once mentioned to me as what "someday should be". There would be
other things, like archiving on the Datacomputer (for the ARPAnet version),
and internet communication, and all sorts of things like that.
What I was wondering was should this sort of thing be considered for
Dialnet? Certainly this kind of teleconferencing is the wave of the
future and ARPA is going to be quite excited about an ARPAnet implementation
(including being willing to put real money into the project, since this
sort of thing is what the military types have been asking the hackers for
for a while). The problem is that it would be quite expensive as far as
phone costs, especially since to be effective the teleconferencing must
involve a "ring" of connections and at least two different remote hosts
will have to be connected to ensure the integrety of the node's membership
in the conference should a node experience system interruption. Of course,
connecting everybody to everybody else is a possibility but even more
expensive.
I am inclined to put the issue on the shelf for the time being, pending
the resolution of more important issues. However, I think it would be a
good idea for us to consider it and at least read up on what is being done
on teleconferencing on the ARPAnet.
Mark
∂07-Mar-78 2326 MRC IMSSS people
To: JMC, LES
I haven't heard any reply from the IMSSS people about my setting up the
Dialnet protocol at their end. So I am going to go over there in person
tommorrow and poke around and see what happens.
∂08-Mar-78 0440 DCL
To: GROUP.DIS[VCG,DCL]:
***********************************************************************
VERIFICATION GROUP SEMINAR TUESDAY 14th. MARCH
PLACE: ERL 237
TIME: 3.00 pm
TITLE: Verification and the Design of Euclid
SPEAKER: JIM HORNING
Xerox Research Labs.
Euclid is a language for writing system programs that are to be verified.
It evolved from Pascal by a series of changes intended to make it more
suitable for system programming and/or more verifiable. Although Euclid
is intended for a rather restricted class of applications, most of the
verification concerns are relevant to languages designed for more general
purposes. I will discuss some of the changes to Pascal that were
specifically motivated by concern for verification, and attempt to assess
their effectiveness.
From this, I will attempt to extrapolate some conclusions about the
helpfulness of verification in providing guidelines for language design.
∂08-Mar-78 0939 FTP:Levin at SUMEX-AIM national research council
Date: 8 Mar 1978 0939-PST
From: Levin at SUMEX-AIM
Subject: national research council
To: jmc at SAIL
john:
i rescheduled the meeting for mon. 13th. which is better
for you 1:00 or 2:00?...laurie
-------
I slightly prefer 1pm.
∂08-Mar-78 1444 FTP:Tajnai at SUMEX-AIM Faculty Meeting, 4 April 2:15 p.m.
Date: 8 Mar 1978 1443-PST
From: Tajnai at SUMEX-AIM
Subject: Faculty Meeting, 4 April 2:15 p.m.
To: faculty.list:
cc: BS at SAIL, Levin
FACULTY MEETING
TIME: Tuesday, April 4, 2:15 p.m.
PLACE: 204 Polya Hall
PRELIMINARY AGENDA:
1. Degree Candidates
2. Admission Results
3. Curriculum Results
4. Non Academic Council Members as Principal Dissertation Advisors
5. Comprehensive Examination
Please send additional agenda items to Carolyn Tajnai
(CET@SAIL, Tajnai@Sumex)
-------
∂09-Mar-78 0948 MRC
To: Pratt at MIT-AI, JMC
09-Mar-78 0944 Pratt at MIT-AI (Vaughan Pratt)
Is Stanford thinking about developing or acquiring personal computers
a la ALTO, LISP machine or the LCS machine?
mrc - To the best of my knowledge, no. We are, as you know, building
another large timesharing computer, the S-1. Additionally, sort of a
"LOTS II" is being set up with the business school's Twenex machine
(not really, but sort of--it's only for business school types, but it
will run all of LOTS' software and remove some of LOTS' load). I am
sending a copy of this message to McCarthy who probably can answer
this much better than I can. -- Mark
∂09-Mar-78 0958 FTP:PRATT at MIT-AI (Vaughan Pratt)
Date: 9 MAR 1978 1251-EST
From: PRATT at MIT-AI (Vaughan Pratt)
To: jmc at SU-AI
Is Stanford planning to acquire personal computers a la ALTO or
LISP machine?
There is some sentiment for it, but I have been claiming that the
goal of a terminal (cheap) for every graduate student and faculty
member should be achieved before personal computers become the
main goal. The department is trying to raise the money for its
own 2020, and with the S-1, we shall probably have enough compute
capacity and file capacity to support enough terminals.
∂09-Mar-78 1512 SSO Faculty meeting curriculum
To: JMC
CC: RWF, JLH, JMB, GIO
How about a report on the state of picking a standard programming
language for intro. courses?
Sue, are you in a position to give said report? If not, who?
∂09-Mar-78 1849 FTP:RYLAND at RUTGERS-10 Dialnet
Date: 9 Mar 1978 (Thursday) 2150-EST
From: RYLAND at RUTGERS-10
Subject: Dialnet
To: mrc at SU-AI
cc: jmc at SU-AI, les at SU-AI
Mark: Just a reminder about my pending question: can you knock off a chunk
of the protocols for me to work on, so that I can be concrete about wht
I'm doing? Thanks. Several other tidbits: my boss says we can most likely
get the modems, and even better, we probably already have 'em (they're
being used by another group, but since the phone rates started to go by
the hour, it's cheaper for them to lease lines). Um, just thinking ahead,
it'll be sort of expensive to try and debug our protocols cross-country, so
it's important that we at least two machines talking Dialnet locally; we
could possibly even just loop back on ourselves...any ideas? More REAL
thoughts coming a bit later. I'm still held up until I can get my hands
on that ARPAnet protocols handbook. Sigh. CHeers... --chris
∂10-Mar-78 0219 CLT MICRO
There are three sample solutions and explanation in your mailbox
∂10-Mar-78 0831 SSO Language choice
To: JMC
CC: RWF, JLH, GIO, SSO
John, The last faculty meeting left the matter under consideration
by the curriculum committee. I don't know whether anything has been
done there. I would be glad to announce that Pascal is the newly
adopted official language--informal discussions among the systems faculty
have strongly favored Pascal.
Sue
1. I have one long term substantive objection to Pascal. It
doesn't have conditional expressions. Therefore, it doesn't allow
conditional expression recursion as a way of defining functions,
and this turns out to be one of the two ways, and in my opinion the
more flexible, of writing programs that can be proved to meet their
specifications. Since we aren't ready to introduce correctness
proofs to CS105 and CS106 students, this is a limited objection
at present.
2. Putting on my LOTS hat, I object to the adoption of a
language without at least a one quarter trial with a smaller group.
Is anyone preparing to make such a trial this Spring, or has
there been one?
∂10-Mar-78 0924 SSO Pascal
To: JMC
CC: RWF, JLH, JMB, GIO
This quarter both Jeff (CS240) and John Hennesy (CS140A) are
using Pascal. I think they are both quite happy with it. The compiler
is fast and produces fast code.
∂10-Mar-78 0942 JLH Pascal
To: JMC, SSO
CC: JMB, RWF, Wiederhold at SUMEX-AIM
John, I have been using the Pascal compiler heavily this
quarter (CS140A - Systems Programming). I am happy
with the compiler. It is very fast, and has a nice
debuggiing package with reasonable error facilities.
As far as it being a programming language there are several
minor points, which are irritating:
1) no conditional expressions, or optimal expression evaluation
2) static arrays can occasionally be a problem, particularly
with strong type checking.
3) Althougth strong type checking is sometimes a pain -
its benfits are great - especially for students
4) Possibly the biggest objection is that string processing
is very difficult. We have formed a plan for augmenting the
(with some predefined procedures) which will
remove this difficulty. It will not be SAIL, but
it will combine most of
the standard string processing mechanisms - and PASCAL's
strong type checking.
I do not intend to define Pascal here. (I do not believe it needs defense).
haviHowever, I would like to stante that teaching with a language having
a rich type structure like Pascal is a real joy.
John
∂10-Mar-78 1011 FTP:Wiederhold at SUMEX-AIM Re: Language choice
Date: 10 Mar 1978 1008-PST
From: Wiederhold at SUMEX-AIM
Subject: Re: Language choice
To: SSO at SU-AI, JMC at SU-AI
cc: RWF at SU-AI, JLH at SU-AI, GIO at SU-AI, SSO at SU-AI
In response to the message sent 10 Mar 1978 0831-PST from SSO at SU-AI (Susan Owicki)
To the best of knowledge ( I believe I'm a member ) the curriculum
committee has not met since the last faculty meeting. It is hence best
for faculty members who have used, or plan to use LOTS for introductory
courses to bring the topic to the floor. Gio
-------
∂10-Mar-78 1355 JBR
To: JMC
CC: LES
The account 2,JMC now has the ACT privilege. You can now
access protected files in areas with the ACT group access bit from 2,JMC
∂10-Mar-78 1424 JSW Admissions
To: EAF, RWF, JGH, DEK, JMC, EJM, buchanan at SUMEX-AIM, JMB, FXB, CCG
To: TAR, TW, ACY, FFY, earnest, DCL, DPB
CC: JSW
Please respond to Gene H. Golub's memo of March 8, l978, by March l5, l978.
∂10-Mar-78 1709 TOB
Site visit for NSF proposal
Chern plans to arrange the NSF site visit during the first 10 days
of April. Please let me know when in that period is difficult for
you. I will let Chern know about possible conflicts.
I plan to be in the east from the 2nd to 14th. Can you and he
manage without me?
∂10-Mar-78 2018 REM via SU-TIP#3 APPENDIX TO MY INTCOM.WRU
To: CRUNCH.DIS[1,REM]:
(1) I now have a fully-modular pure-lambda-function (no side-effects,
all continuations explicitly passed and returned) version of my
Interval-Subdivision data-compression algorithm, coded in MACSYMA.
(2) The code has been GRINDed and edited with explanations and
comments to form the appendix to my paper. The annotated appendix
is INTCOM.APX[1,REM]. Feel free to look at it, but don't distribute.
(3) Next to code in machine language (FAIL).....
∂11-Mar-78 0556 REM
"probability" is my worst most-frequent spelling error,
"preceding" is my second. "existence" (sp??)
must be another I'd better watch for...
Re non-markovian -- Yes, whenever there is a procedure that
accurately (on the average, of course) predicts the
probability distribution for the next character, it can be
plugged in as the "markovian-model" black box, doesn't have
to be Markovian, as you point out. I suspect, however,
that left-context-dependent predictors will be able
to switch context as fast as any chi-sq detector could, although
only actual testing will show for sure.
∂11-Mar-78 0556 REM via SU-TIP#3 More on great wonderment of my algorithm
To: CRUNCH.DIS[1,REM]:
A few hours ago I realized that my method has the same advantage that
COBOL does with respect to I/O formats, you specify it once and it
works both for input and output inversely without any bugs. In my
case, you give me any black box that predicts probability of next
character, interfacing correctly via the calling sequences
MODEL <- INITMODEL(N), MODEL <- UPMODEL(MODEL,TOKEN,N,J),
and HIST <- SELHIST(MODEL), and all we have to do is load it into
Macsyma (or whatever the system uses) to replace the old modules,
and both crunch and uncrunch will be updated to inverses of each
other using your particular model. I plan to make these three
functions formal arguments to the CRU2 and UNC2 functions, just
like INBIT, BITOUT, INTOK, TOKOUT, so that switching models is
more flexible, including simultaneously running several different
(totally different) models and comparing their output.
Note, JMC pointed out that any predictor, not necessarily a Markovian
one, will suffice for INITMODEL/UPMODEL/SELHIST. The writeup will
be updated in this regard sometime...
∂11-Mar-78 0614 REM via SU-TIP#3
To: CRUNCH.DIS[1,REM]:
When I put the tokenizer in the flow chart (not yet transcribed
into the writeup, but implemented as INTOK and TOKOUT), what I had
in mind was either some constant tokenizer such as my (actually
Mike Wilber's) pseudo-Baudot finite-state machine, or some type of
dictionary-gatherer. Let me explain the lattter more fully. The
tokenizer would parse full words, check to see if new or old,
normalize all to upper-case (breaking words like "McCarthy" into
separate words each of which conforms to one of 3 cases: lowercase,
uppercase, first-letter-capitalized), and emit special tokens
like this: <newwordtoken> <letter> <letter> ... <letter> <eow>
to officially enter a new word, and then <oldword> <word#> <case>.
The main program would have a Markovian model for predicting the
histogram for the next token given the preceeding context.
JMC's suggestion of non-markovian models could be implemented in
two ways. The way he seemed to suggest was to have a trivial
tokenizer and then let the INITMODEL/UPMODEL/SELHIST software do
a chi-sq test etc. to switch contexts formally (as contrasted with
my idea of letting left-context sort-of switch contexts all by
itself). An alternative is to do the same sort of thing as my
original idea for dictionary-gatherer. The tokenizer would have
all the heuristics, and emit a <change context> token whenever
it wanted the INITMODEL/UPMODEL/SELHIST software to erase its
memory to begin training for a new context. The tokenizer could
have arbitrary look-ahead without messing up UNC=CRU↑(-1),
providing that the detokenizer actually inverted what the tokenizer did.
I'm leaning toward thinking that is the right way to do it,
i.e. three parts of program, tokenizer for doing look-ahead and
reducing the file to a Markovian stream plus occasional magic commands,
the Markovian model-software to generate histograms and obey the
magic commands, and the bitizer to convert those histograms and tokens
into nested intervals and thence to bits.
∂12-Mar-78 0336 REM via SU-TIP#3 First complete draft of interval-subdivision data-compression-algorithm
To: CRUNCH.DIS[1,REM]:
I have now installed the overall data-flowchart, from which I designed the
software and from which I came up with all the amazing realizations about
parallelism and cobol-like automatic inverses. I have also edited the
latest Macsyma source code into the annotated program listing. All this
is in INTCOM.APX[1,REM]. Together with the main article, in INTCOM.WRU[1,REM],
the first draft of my paper is complete. (Note I haven't yet corrected
the spelling errors noted by you all previously in the main article.)
Remember, limited distribution, for you only, not your friends...
∂12-Mar-78 1043 MRC error handling
To: Ryland at RUTGERS-10, JMC, LES
While writing the pdp10 implementation of the Dialnet routines, I am
coming to realize that a user mode process simply cannot expect to win
with the type of error handling that the line transmission protocol
talks about. Considerable bandwidth would be lost in relatively useless
ERR's. I think that most of the error conditions should be flushed, and
for the most part, a framing error would restart the packet reader
without any diagnostic in the right manner (ie, SOP framing starts at
the packet reader, EOP misframing at the SOP seeker). The SOP seeker
shouldn't complain if a non-SOP comes in; instead it should just flush
it.
The Bell Labs hackers suggested this to me a while back and I agree now
that I see the implementation problems. I propose that ERR should be
used only for checksum error and protocol violation, IFF the protocol
violation occured in a packet with correct checksum. All other garbage
just gets ignored. This will keep the packet reader from blowing its
mind when it gets screwed up in its framing, or rather, it will blow
its mind quietly instead of sending infinite errors.
The only good reason I see for keeping ERR at all is to allow for packet
abortion (to abort a packet if a previous lost). Some standard has to
be set for clean abortion as well (ignoring the right to life arguments!).
I'm sort of thinking of EOP EOP, which seems to be the shortest way to
abort a packet which will work in all cases.
What do you think?
-- Mark
∂13-Mar-78 0720 JRA ruth
Thank you from Ruth! By the way my LISP book is finally being
typeset from paper-tape no less. It should be bound by May 1.
It has been a long process, hopefully worth the hassle. One thing
for sure: That process has made me appreciate your patience,
your laboratory, as well as your language! thank you.
john
∂13-Mar-78 0856 FTP:Levin at SUMEX-AIM national research council
Date: 13 Mar 1978 0856-PST
From: Levin at SUMEX-AIM
Subject: national research council
To: jmc at SAIL
john:
reminder: meeting today at 1:00, polya 204.
...laurie
-------
∂13-Mar-78 1155 DPB Pascal experiment for 105/6
To: JMC
CC: SSO, RWF, GIO, JLH, DWW
∂10-Mar-78 0921 JMC
To: SSO, RWF, JLH, GIO, DPB
1. I have one long term substantive objection to Pascal. It
doesn't have conditional expressions. Therefore, it doesn't allow
conditional expression recursion as a way of defining functions,
and this turns out to be one of the two ways, and in my opinion the
more flexible, of writing programs that can be proved to meet their
specifications. Since we aren't ready to introduce correctness
proofs to CS105 and CS106 students, this is a limited objection
at present.
2. Putting on my LOTS hat, I object to the adoption of a
language without at least a one quarter trial with a smaller group.
Is anyone preparing to make such a trial this Spring, or has
there been one?
[The trial is not planned for Spring. Dave Wall (at least) will do
a section of 105 or 106 during the summer. That should provide the
necessary stats so that we can evaluate. There is a possibility that
more of the summer sections will want to try Pascal, depending on the
instructors. -Denny]
The scale of the trial shouldn't matter during the summer.
∂13-Mar-78 1220 PAT vacation
To: JMC, LES, ZM
I am planning on being away (in France) from May 11 to June 7 for vacation.
∂13-Mar-78 1436 JLH Pascal compiler
To: DPB, SSO, RWF, GIO, JMC
I think the CS140A, cs240A is more than an ample trial,
i.e. the classes are large and they construct reasonably
large programs. (If the trial is from a computer performance
point of view.) From a pedagodical viewpoint, we can
either try Pascal for a first course (105/6), or look at other
univ. employing it. I think if they can learn SAIL, they can
easily work with Pascal.
John
∂13-Mar-78 1501 FTP:Levin at SUMEX-AIM senior faculty mtg
Date: 13 Mar 1978 1434-PST
From: Levin at SUMEX-AIM
Subject: senior faculty mtg
To: buchanan, jmc at SAIL
reminder...reminder
mtg. tomorrow
time: 12:30
place: polya 204
date: tue. 3/14
...ll
-------
∂13-Mar-78 1637 LCW S-1 Seminar and Meeting
To: LES, TED, JBR, ME, JMC
CC: LCW, LLW, TM, PMF
As announced, there will be a seminar about the S-1 on Monday, March 20.
It will deal with the Design System and Multiprocessor and Uniprocessor
architecture and implementation.
Lowell indicated that he would like to talk with the SAIL S-1 group after
the seminar about plans for OS and Switch work.
Curt
∂13-Mar-78 1714 JBR
You have exceeded your disk quota.
The files listed below have been purged to reduce your disk
area to your quota of 2000
Before purging, your files occupied 2907
You have too many files. The purger may not select the
optimum set.
PPSAV.TMP[ESS,JMC]
AIQUAL.LST[W77,JMC]
CRYPT.DMP[ 2,JMC]
CODE.DMP[ 2,JMC]
2P.LAP[W77,JMC]
EXOTIC.XGP[W77,JMC]
FIRST.XGP[W77,JMC]
MAILIN.XGP[SEN,JMC]
LUTZ.XGP[LET,JMC]
III.XGP[LET,JMC]
GRAEVE.XGP[LET,JMC]
TECHNO.XGP[SEN,JMC]
DIALNE.XGP[DIA,JMC]
MINIMA.XGP[S77,JMC]
SUNDES.XGP[SEN,JMC]
LEGISL.XGP[SEN,JMC]
TIMES.XGP[LET,JMC]
ESTRIN.XGP[LET,JMC]
KURGAN.XGP[LET,JMC]
ABRAHA.XGP[LET,JMC]
IJCAI.XGP[E77,JMC]
LISP.XGP[F77,JMC]
MICRO.XGP[F77,JMC]
SCIENC.XGP[LET,JMC]
CARTWR.XGP[LET,JMC]
GORDON.XGP[LET,JMC]
MC[245,JMC]
PATH[245,JMC]
PATH2[245,JMC]
REFER.ENC[225,JMC]
PATHS.RLS[225,JMC]
GRPDAT.RLS[225,JMC]
GRPALG.RLS[225,JMC]
GROPER.RLS[225,JMC]
DIMEN.RLS[225,JMC]
CHAR.RLS[225,JMC]
DEG5.IN[225,JMC]
REPRES.RLS[225,JMC]
S4.REP[225,JMC]
DEG6.IN[225,JMC]
R42.IN[225,JMC]
PUZZ.SAI[225,JMC]
PUZZA.SAI[225,JMC]
PANIC.SOS[225,JMC]
ANNOUN[225,JMC]
PUZZB.SAI[225,JMC]
ENCYC1.PRO[ESS,JMC]
PUZZE.SAI[225,JMC]
MINE.DIR[ AI,JMC]
PTCH2.DIR[ AI,JMC]
DRIVE.DIR[ AI,JMC]
KNOW2.AI[ESS,JMC]
ARTNA1.ART[ESS,JMC]
MEMMTC.QUA[ESS,JMC]
LANG1.AI[ESS,JMC]
MTC71.QUA[ESS,JMC]
S5.QUA[ESS,JMC]
R1301.ART[ESS,JMC]
R1303.ART[ESS,JMC]
QA.PRO[ESS,JMC]
ARTNAT.ART[ESS,JMC]
SEMAN.AI[ESS,JMC]
FRANK.STA[ESS,JMC]
COMPUT.STA[ESS,JMC]
MTCSYL.QUA[ESS,JMC]
AISY69.QUA[ESS,JMC]
FILEC.FAI[ESS,JMC]
LETTVI.REV[ESS,JMC]
R1300.ART[ESS,JMC]
LCFMEM.MTC[ESS,JMC]
PEARL.NOT[ESS,JMC]
NETJO2.PRO[ESS,JMC]
IPT73.REP[ESS,JMC]
NSFX.5[ESS,JMC]
NSFX.7[ESS,JMC]
NSFX.9[ESS,JMC]
NSFX.2[ESS,JMC]
NSFX.6[ESS,JMC]
NSFX.8[ESS,JMC]
NSFX.3[ESS,JMC]
NSFX.4[ESS,JMC]
HAYES.FRM[ESS,JMC]
ADTEL2.NOT[ESS,JMC]
PLATT.NOT[ESS,JMC]
TASK1.MAR[ESS,JMC]
TASK2.MAR[ESS,JMC]
LETTER.HE2[ESS,JMC]
XLET[ESS,JMC]
ZF.AX[ESS,JMC]
FOO73A[ESS,JMC]
FOON.PR1[ESS,JMC]
SWOPSI.RE[ESS,JMC]
HART.ME1[ESS,JMC]
GGGG[ESS,JMC]
PRO[ESS,JMC]
M29AUG.MES[ESS,JMC]
SYMPOS[ESS,JMC]
CH1.BH2[ESS,JMC]
CLOCK.FAI[ESS,JMC]
SLIDES[ESS,JMC]
FOO1[ESS,JMC]
WUTHER.QUE[ESS,JMC]
ROY[LET,JMC]
TOLLES.LET[LET,JMC]
FORDH.LET[LET,JMC]
AD.NET[ESS,JMC]
KNUTH.SAI[225,JMC]
TOLLES.LE1[LET,JMC]
TUCKER.LET[LET,JMC]
TUCKER.LE2[LET,JMC]
TCLUB.MEM[ESS,JMC]
GEN1[ AI,JMC]
DRIVE3[ AI,JMC]
MPRO[ AI,JMC]
PUTCH[ AI,JMC]
PUTCH.DIR[ AI,JMC]
DRIVE[ AI,JMC]
AIBIB[ AI,JMC]
MINE[ AI,JMC]
NOTES2[ AI,JMC]
PUTCH3[ AI,JMC]
RUSS[ AI,JMC]
CHYLIN.LE1[ AI,JMC]
SLAMEC.LE2[LET,JMC]
BOWDEN.LE1[LET,JMC]
CHERN.LE2[LET,JMC]
MAY.ME[LET,JMC]
NSFX.10[ESS,JMC]
MC.AI[ESS,JMC]
HENRY[ESS,JMC]
BIRKHO.LE1[LET,JMC]
FLANAG.LE1[LET,JMC]
ERIK.LE6[LET,JMC]
HARPER.LE1[LET,JMC]
PUZZE.F4[225,JMC]
ERIK.LE7[LET,JMC]
BAJCSY.LE1[LET,JMC]
STARKM.LE1[LET,JMC]
BRIT.LET[LET,JMC]
IRANI.LE1[LET,JMC]
BOWEN.LE1[LET,JMC]
JAF.STA[ESS,JMC]
RECUR[ESS,JMC]
SCIENC.LE1[LET,JMC]
SCRIB.LE1[LET,JMC]
ATKIN.LE1[LET,JMC]
KOCHEN.LE1[LET,JMC]
GREER.FAC[LET,JMC]
CU.LE1[LET,JMC]
RALSTO.LE2[LET,JMC]
MESSAG.SCI[LET,JMC]
MCCRAC.LE2[LET,JMC]
RALSTO.LE3[LET,JMC]
CHRON.LE1[LET,JMC]
LOGIC.QUA[LET,JMC]
SHIGER.LE1[LET,JMC]
PAMELA.LE1[LET,JMC]
TAUGNE.LE1[LET,JMC]
MORGAN.LE2[LET,JMC]
FROOK.LE1[LET,JMC]
LOMONA.LE1[LET,JMC]
APR.ME[LET,JMC]
LOGIC.PUB[LET,JMC]
DAVID.LE1[LET,JMC]
SCHOEL.LE1[LET,JMC]
CALHOU.LE1[LET,JMC]
FINCH.LE1[LET,JMC]
LYKOS.LE1[LET,JMC]
JMILLE.LE1[LET,JMC]
MITCH.LE1[LET,JMC]
JERRY.LE1[LET,JMC]
MCQUIL.LE1[LET,JMC]
PAM.LE1[LET,JMC]
DRLUK.LE1[LET,JMC]
OCT.ME[LET,JMC]
OLSEN.LE1[LET,JMC]
ERIK.LE0[LET,JMC]
SUBCOM.LE1[LET,JMC]
NEWSCI.LE1[LET,JMC]
AHLHAU.LE1[LET,JMC]
MEMOS.CUR[LET,JMC]
PUBINT.LE1[LET,JMC]
HORNIN.LE1[LET,JMC]
FELDST.LE1[LET,JMC]
CHECK.DEP[LET,JMC]
TUCKER.LE1[LET,JMC]
SCANDU.LE1[LET,JMC]
MELTZ.LE1[LET,JMC]
PIASET.LE1[LET,JMC]
MILNER.REC[LET,JMC]
FLESS.LE1[LET,JMC]
TRESSL.LE1[LET,JMC]
PIASET.LE2[LET,JMC]
LICKL.LE1[LET,JMC]
COOP.LE1[LET,JMC]
POE.LE1[LET,JMC]
JMILLE.LE2[LET,JMC]
IGARAS.LE1[LET,JMC]
ITO.LE1[LET,JMC]
CLARK.LE1[LET,JMC]
FLESS.LE2[LET,JMC]
TAVARE.LE1[LET,JMC]
FORSHA.LE1[LET,JMC]
COMPT.LE1[LET,JMC]
STORK.LE1[LET,JMC]
ASTRO2.ART[ESS,JMC]
R1302.ART[ESS,JMC]
FRIEDM.LE1[LET,JMC]
FF[ESS,JMC]
AIRLIN.MTC[ESS,JMC]
JFOL.USE[ESS,JMC]
MEARS.LET[LET,JMC]
GRIMM.BKP[ESS,JMC]
SCIAM.LE1[LET,JMC]
NYT.LE1[LET,JMC]
NAVROZ.LE1[LET,JMC]
NOTES1[ AI,JMC]
STARK.LE1[LET,JMC]
REPRES[LET,JMC]
MDAVIS.LE1[LET,JMC]
IGARAS.LE2[LET,JMC]
GLUGG2[LIT,JMC]
KNOW.ART[ AI,JMC]
FLESS.LE0[LET,JMC]
KNOW.NOT[ AI,JMC]
GLUGG[LIT,JMC]
MCCRAC.LE1[LET,JMC]
NAS.NOT[ESS,JMC]
JAF.LE2[LET,JMC]
TELLER.LE1[LET,JMC]
KILPAT.LE1[LET,JMC]
MEIR.LE1[LET,JMC]
CRYPT.FAI[ 2,JMC]
WEISSK.LE1[LET,JMC]
AAAS.PRO[ESS,JMC]
3D.NSF[ESS,JMC]
ONEILL.NOT[ESS,JMC]
ISRAEL.LE1[LET,JMC]
ISRAEL.ART[LET,JMC]
CORBAT.LE1[LET,JMC]
CHENEY.LE1[LET,JMC]
OCTOBE.OUT[LET,JMC]
CALHOU.LE2[LET,JMC]
EFOO[ESS,JMC]
COMTOP.LIS[ESS,JMC]
NETDOC.PLN[ESS,JMC]
EVANS.LE1[LET,JMC]
EVANS.LE2[LET,JMC]
INFO.NEE[ESS,JMC]
SUFFIC.FRE[LET,JMC]
OCT.USE[ESS,JMC]
OLIVER.LE1[LET,JMC]
INTRO.AI[ 2,JMC]
COMMON.NS[LET,JMC]
DRELL.LE1[LET,JMC]
SIGART.LE2[LET,JMC]
SIGART.BUL[LET,JMC]
SIGART.LE1[LET,JMC]
BOWEN.LE2[LET,JMC]
RUSSOF.NOT[ESS,JMC]
INVOIC[LET,JMC]
ACM.LE2[LET,JMC]
ACM.LE1[LET,JMC]
JOSHI.LE1[LET,JMC]
FERRIS.LE1[LET,JMC]
NYT.LE2[LET,JMC]
BDAY.LIT[LIT,JMC]
BACKUP.MCP[258,JMC]
HOMOMO.NOT[258,JMC]
MCP.PR1[258,JMC]
COND.PRF[258,JMC]
PQR.PRF[258,JMC]
SCHEM.WRU[258,JMC]
MCCUS.PRO[258,JMC]
FOLSCO.MEM[258,JMC]
EXTFOR.MEM[258,JMC]
ACTOR0[258,JMC]
NONTER.PRB[258,JMC]
BRACEW.LE1[LET,JMC]
CORREC.ABS[258,JMC]
NAUR.LE1[LET,JMC]
FEMANO.LE1[LET,JMC]
MTC.GRP[ESS,JMC]
PETERS.LE1[LET,JMC]
JAN.ME[LET,JMC]
LEGUIN.CRI[LIT,JMC]
JAN14.DBX[258,JMC]
GELLMA.LE1[LET,JMC]
ESSEX.LE1[LET,JMC]
BREMER.LE1[LET,JMC]
SHELDO.NS[LET,JMC]
APPEND.CON[258,JMC]
FOLMAN.MOD[258,JMC]
HOBBS.LE1[LET,JMC]
CONSUL.POX[ESS,JMC]
AILET[LET,JMC]
REYNOL.LE1[LET,JMC]
ACM.LE3[LET,JMC]
LEAST.CM1[258,JMC]
NORBYE.LE1[LET,JMC]
JAN.OUT[LET,JMC]
DEC74.IN[LET,JMC]
LOGIC.QUA[258,JMC]
OUTLIN.256[258,JMC]
HAND.2[258,JMC]
JO.NOT[258,JMC]
MTC.BLA[258,JMC]
MANNA.WRU[258,JMC]
DEMO.PR1[258,JMC]
HAND3.W75[258,JMC]
HAND.W75[258,JMC]
MANNA.PR2[258,JMC]
OKAMOT.LE1[LET,JMC]
MANNA.CM2[258,JMC]
HARDPR.OOF[258,JMC]
CHRI.LE1[ 2,JMC]
VIS[ 2,JMC]
VIS1[ 2,JMC]
VIS.CAP[ 2,JMC]
NNN[ 2,JMC]
VITASK[ 2,JMC]
FRAGA.LE1[LET,JMC]
ENERGY.LE1[LET,JMC]
FIELDS.ME2[LET,JMC]
DEC.ME[LET,JMC]
FEB.ME[LET,JMC]
TASK[LIT,JMC]
ADHOC.THE[258,JMC]
SAKAI.LE1[LET,JMC]
LICK.ME1[LET,JMC]
LICK.LE1[LET,JMC]
LICK[ESS,JMC]
HAND.1[258,JMC]
BIBLE[245,JMC]
USER[245,JMC]
IDEAS[225,JMC]
SUZUKI.REC[LET,JMC]
MUFTIC.LE1[LET,JMC]
SAKAI.LE2[LET,JMC]
COMPIL.AX[258,JMC]
INTEGE.AX[258,JMC]
COND.AX[258,JMC]
ZF.AX[258,JMC]
INTEG2.AX[258,JMC]
EQUAL.AX[258,JMC]
MCP.AX[258,JMC]
MCP2.AX[258,JMC]
MCP3.AX[258,JMC]
EXTFOR.AX[258,JMC]
INDUC.AX[258,JMC]
ASHLEY.LE1[LET,JMC]
74DEC1.AJT[LET,JMC]
SMITH.LE1[LET,JMC]
GAREY.LE1[LET,JMC]
BERG.LE2[LET,JMC]
KOREA.POL[S75,JMC]
KOREA.2[S75,JMC]
RUSSIA.LE1[LET,JMC]
GELFAN.LE1[LET,JMC]
MAY75.OUT[MSG,JMC]
JUN75.OUT[MSG,JMC]
JUL75.OUT[MSG,JMC]
AUG75.OUT[MSG,JMC]
SEP75.OUT[MSG,JMC]
JAN75.OUT[MSG,JMC]
DEC74.OUT[MSG,JMC]
MAR75.OUT[MSG,JMC]
APR75.OUT[MSG,JMC]
MAR75.IN[MSG,JMC]
INGERM.LE1[LET,JMC]
APR75.IN[MSG,JMC]
MODEST[S75,JMC]
PROPOS.BLA[CUR,JMC]
PHIL[CUR,JMC]
COMMON[CUR,JMC]
NEWS75.MOD[CUR,JMC]
EXEC.SUM[CUR,JMC]
PRODUC.ESS[CUR,JMC]
STANDA.ESS[CUR,JMC]
ARPMTC[CUR,JMC]
IDEOLO.ESS[CUR,JMC]
CHESS1[CUR,JMC]
SOCIAL.DEF[CUR,JMC]
RAMSEY[CUR,JMC]
HOW.ESS[CUR,JMC]
LEADER.ART[CUR,JMC]
LICK.PRE[CUR,JMC]
REPRES.LIC[CUR,JMC]
REPRES.LI2[CUR,JMC]
GOALS.LIC[CUR,JMC]
TWOMOD.ESS[CUR,JMC]
FORMAL.PRO[CUR,JMC]
SALARY.MEM[CUR,JMC]
SCIP[CUR,JMC]
IDEOLO[CUR,JMC]
ECON.NOT[CUR,JMC]
AICIRC.ABS[CUR,JMC]
CPDUST[CUR,JMC]
IDEOLO.ART[CUR,JMC]
APHOR.AI[CUR,JMC]
ECO.ESS[CUR,JMC]
RACKET.ESS[CUR,JMC]
1974.IN[MSG,JMC]
JAN75.IN[MSG,JMC]
ARPA75.OUT[MSG,JMC]
FEB75.IN[MSG,JMC]
AUG75.IN[MSG,JMC]
JUL75.IN[MSG,JMC]
XGPPUB.1[ESS,JMC]
ADMIT.PUB[CUR,JMC]
KHABAR.AI[S75,JMC]
NEWS75.SUP[ESS,JMC]
NAKATA.LE1[LET,JMC]
REPRES.DOC[ESS,JMC]
REPRES.NOT[ESS,JMC]
HOMER.LE1[LET,JMC]
HENDRI.REV[ESS,JMC]
LEM.LE1[LET,JMC]
REPRES.REP[ESS,JMC]
TWOEAS.MEM[258,JMC]
SCHULT.LE1[LET,JMC]
OPTIM[F75,JMC]
DYSON.LE1[LET,JMC]
LUMELS.LE1[LET,JMC]
NEWBOR.REV[F75,JMC]
NICK.FIL[F75,JMC]
TASK.TA[F75,JMC]
SEXP[F75,JMC]
AI[F75,JMC]
LATIME.LE1[LET,JMC]
GENERA[F75,JMC]
LISP.MAN[S75,JMC]
SANDEW.LE2[LET,JMC]
DEFINE[F75,JMC]
PHIL.ART[F75,JMC]
BULLET[F75,JMC]
DATA3.204[F75,JMC]
DATA1.204[F75,JMC]
DATA4.204[F75,JMC]
ISHII.LE1[LET,JMC]
ED1.DO[ESS,JMC]
TEXTOR.LE1[LET,JMC]
EDH.DO[ESS,JMC]
OFFICE.PLN[F75,JMC]
ASHENH.LE1[LET,JMC]
ASHENH.LE2[LET,JMC]
IJCAI.ENQ[F75,JMC]
IJCAI.DOC[F75,JMC]
LUK.LE1[ 2,JMC]
DCL2.75[LET,JMC]
AAAS[ESS,JMC]
AAAS.76[ESS,JMC]
SCIENC.LE2[LET,JMC]
ABELSO.LE1[LET,JMC]
NILSSO.ME1[LET,JMC]
LUCKHA.LE1[LET,JMC]
HJERPP.LE1[LET,JMC]
BROWN.LE1[LET,JMC]
MOSES.LE1[LET,JMC]
RELATI.POX[CUR,JMC]
GILFIL.REV[F75,JMC]
SCOTT.LE1[LET,JMC]
SENSE.AD[LET,JMC]
PR1.DEC[F75,JMC]
AJT.MEM[ESS,JMC]
RAWLS.REV[ESS,JMC]
CYCLOP[F75,JMC]
GINI.LE1[LET,JMC]
SENSE.POS[F75,JMC]
MARTEL.DOC[LET,JMC]
MARTEL.LE1[LET,JMC]
FRIED.LE1[LET,JMC]
ROTH.LE1[LET,JMC]
FRIED.LE2[LET,JMC]
FIX[F75,JMC]
FIXUP.LBK[F75,JMC]
BRINKE.LE1[LET,JMC]
TABLE[F75,JMC]
FIXUP2.LBK[F75,JMC]
TERMPA.206[F75,JMC]
RYLE.REV[F75,JMC]
TUNNEY.LE1[LET,JMC]
FIXUP[F75,JMC]
FIXUP2[F75,JMC]
BRINKL.LE1[LET,JMC]
NEWS75.PUB[CUR,JMC]
SENSE.LE1[LET,JMC]
MOTIV.ART[ AI,JMC]
GARDNE.LE2[LET,JMC]
DAILY.ART[F75,JMC]
SENSE.LE2[LET,JMC]
REVAL2.LBK[F75,JMC]
REVAL3.LBK[F75,JMC]
AUTOMA.2[F75,JMC]
AUTOMA[F75,JMC]
COUNTE[F75,JMC]
PR1.AX[F75,JMC]
MELTZE.LE[LET,JMC]
HOTER.1[CUR,JMC]
HOTER[CUR,JMC]
NOV.ME[LET,JMC]
HEAD.LET[LET,JMC]
MACHIN[F75,JMC]
BETHE.LE1[LET,JMC]
SENSE.DAT[F75,JMC]
GOLDMA.LE1[ 2,JMC]
BETHE.LE2[LET,JMC]
FINITI[F75,JMC]
SOULE.LE1[LET,JMC]
THESIS[F75,JMC]
SCIENT.LE1[LET,JMC]
MINIMA[F75,JMC]
FIB[F75,JMC]
DEC75.IN[LET,JMC]
BLEDSO.DOC[LET,JMC]
RHODES.LE1[LET,JMC]
FIELDS.ME1[LET,JMC]
BLEDSO.LE1[LET,JMC]
NEWYOR.LE1[LET,JMC]
FIELDS.SUP[LET,JMC]
RAPHAE.LE1[LET,JMC]
GARDNE.LE1[LET,JMC]
KAHN.LE1[LET,JMC]
LISP.INT[F75,JMC]
BLOCKS.226[F75,JMC]
NOTES.226[F75,JMC]
TAUT.PRF[F75,JMC]
UTILI.AIL[F75,JMC]
MOSSIG.LE1[LET,JMC]
PET.LIT[LIT,JMC]
TOLEDO.LE1[LET,JMC]
CBCL[F75,JMC]
SET.AX[F75,JMC]
SET[F75,JMC]
SENSE.LE3[LET,JMC]
FILMAN.RE2[LET,JMC]
GIL1[ 2,JMC]
KNOPOF.LE1[LET,JMC]
YAMADA.LE1[LET,JMC]
CULIK.LE1[LET,JMC]
MILLER.LE0[LET,JMC]
ANDREI.ADR[ESS,JMC]
JAN76.IN[LET,JMC]
LUCKHA.BLA[F75,JMC]
JAN76.OUT[LET,JMC]
ARPA75.IN[MSG,JMC]
ADMIT.TEL[LET,JMC]
LICK.1[LET,JMC]
PHIL.ART[ESS,JMC]
WILSON.LE1[LET,JMC]
WILKIN.LE1[LET,JMC]
OSIS.LE1[LET,JMC]
GERROL.LE1[LET,JMC]
SFWA[S76,JMC]
CSDDIS.N25[ESS,JMC]
COMPUT.CSD[ESS,JMC]
BAT0.LOG[F75,JMC]
CONTEN.MEN[F75,JMC]
NOTE.MEN[F75,JMC]
POLEMI.MEN[F75,JMC]
NOTE1.MEN[F75,JMC]
AI.MEN[F75,JMC]
MOTIVA.MEN[F75,JMC]
SELFCO.MEN[F75,JMC]
CHANGE.MEN[F75,JMC]
TECHNI.MEN[F75,JMC]
AUTOMA.MEN[F75,JMC]
MODEL.MEN[F75,JMC]
EPISTE.MEN[F75,JMC]
CELLUL.MEN[F75,JMC]
RELDEF.MEN[F75,JMC]
MOTIV.MEN[F75,JMC]
CRYPT.MEM[ 2,JMC]
CRYPT.CHK[ 2,JMC]
COMMUN[S76,JMC]
TES[ESS,JMC]
ASSIGN[ESS,JMC]
PDP370.PLN[ESS,JMC]
CONTEN.DOC[ESS,JMC]
FP[ESS,JMC]
TIME.HIS[ESS,JMC]
CODE.PRO[ESS,JMC]
ENERGY.PRO[ESS,JMC]
RATIO.NOT[ESS,JMC]
CHESSP.DIS[ESS,JMC]
TASK[ESS,JMC]
REV[ESS,JMC]
RAY.AP[ESS,JMC]
REFER[ESS,JMC]
REPRES.BIB[ESS,JMC]
ENERGY.QUE[ESS,JMC]
SLAGLE.REC[ESS,JMC]
GENER.AI[ESS,JMC]
ENERGY.FAC[ESS,JMC]
CHARAC.PRO[ESS,JMC]
VAL.JM2[ESS,JMC]
FELDST.REV[ESS,JMC]
MAGS.LIS[ESS,JMC]
ED.DO[ESS,JMC]
SCITOP.LIS[ESS,JMC]
GREET[ESS,JMC]
MUSIC.GRP[ESS,JMC]
EVENT.NOT[ESS,JMC]
GRK.TES[ESS,JMC]
POEM[ESS,JMC]
PSYCHO.AP[ESS,JMC]
TASK.FOO[ESS,JMC]
ANDREI.DAT[ESS,JMC]
SUSAN.TAX[ESS,JMC]
ALL[ESS,JMC]
IGC.PRO[ESS,JMC]
JO.NO2[ESS,JMC]
IGSCS.PLN[ESS,JMC]
DRAW.PRO[ESS,JMC]
VIEW1[ESS,JMC]
RACRUL.MEM[ESS,JMC]
SPUT.PRO[ESS,JMC]
LORE[ESS,JMC]
TECH2.BGL[ESS,JMC]
SUBTES[ESS,JMC]
CSDCOM.REQ[ESS,JMC]
LIST[ESS,JMC]
FREE[ESS,JMC]
GUBBB[ESS,JMC]
BANK[ESS,JMC]
COLE.LET[ESS,JMC]
LIGHT.6[ESS,JMC]
SERA.ZOO[ESS,JMC]
NEWMEX.SPE[ESS,JMC]
PUBLIC.INT[ESS,JMC]
PRO2[ESS,JMC]
JAN75.DBX[ESS,JMC]
LIVER.SAI[ESS,JMC]
IBM.SNG[ESS,JMC]
REGIST[ESS,JMC]
PRIZE.PRO[ESS,JMC]
IBM.ABS[ESS,JMC]
GORIN[ESS,JMC]
NEWREP.LE1[ESS,JMC]
LUCK.REC[ESS,JMC]
CH1.BH[ESS,JMC]
ENERGY.DEM[ESS,JMC]
BROOKS.PLN[ESS,JMC]
NSFX.1[ESS,JMC]
ENERGY.STA[ESS,JMC]
LI[ESS,JMC]
MEMO.PUB[SEN,JMC]
HOOK.LE1[LET,JMC]
WEISKO.LE1[LET,JMC]
AI2.QUA[ESS,JMC]
SPACE.OP[ESS,JMC]
MONOP.STA[ESS,JMC]
BULLET.AP[ESS,JMC]
AISY2.QUA[ESS,JMC]
DATAMA.LE1[LET,JMC]
CHAP1.ART[ESS,JMC]
CHAP2.ART[ESS,JMC]
R1500.2[ESS,JMC]
SUFFIC[ESS,JMC]
INDIA.PRO[ESS,JMC]
PROP.ECO[ESS,JMC]
TECTOP.LIS[ESS,JMC]
R1500.3[ESS,JMC]
R1500.1[ESS,JMC]
TURIN2.LEC[ESS,JMC]
HOTER.PUB[ESS,JMC]
HOTER.DOC[ESS,JMC]
GALLAG.SPI[LET,JMC]
FREGE[W76,JMC]
ENERGY[W76,JMC]
FOL.COM[W76,JMC]
MEET2.SEN[W76,JMC]
IDEOLO[W76,JMC]
HEAVY.SET[W76,JMC]
SENSE.NOT[W76,JMC]
LISP2[W76,JMC]
TASKS.SET[W76,JMC]
CS.226[W76,JMC]
PDP10[W76,JMC]
RUSSEL.PRF[W76,JMC]
LISP.OLD[W76,JMC]
UNIT.PRF[W76,JMC]
Z2.AX[W76,JMC]
HEAVY.AX[W76,JMC]
BUREAU[W76,JMC]
SENSE.PST[W76,JMC]
BLOCKS.AX[W76,JMC]
OBVIOU[W76,JMC]
NUMBER.AX[W76,JMC]
COMMUN[W76,JMC]
HEAVY.THE[W76,JMC]
KOREA[W76,JMC]
SENSE.BLA[W76,JMC]
MOYNIH.1[W76,JMC]
REM.WRU[W76,JMC]
RESOUR[W76,JMC]
AAAS.2[W76,JMC]
SNEKOT.DAT[W76,JMC]
SENSE.1[W76,JMC]
FREEDO[W76,JMC]
FUTURE[W76,JMC]
KNOW[W76,JMC]
ACCOMP[W76,JMC]
SEARCH[W76,JMC]
SUN.OUT[W76,JMC]
PARKER[W76,JMC]
IBM[W76,JMC]
E[W76,JMC]
INFORM[W76,JMC]
SENSE.PS1[W76,JMC]
TELLER[W76,JMC]
WADLER[W76,JMC]
POSTER[SEN,JMC]
SENSE.LAB[W76,JMC]
SENSE.PL1[W76,JMC]
AAAS[W76,JMC]
LICKLI.SPI[LET,JMC]
ASHLEY.COV[SEN,JMC]
SCI[LET,JMC]
COMMEN[S76,JMC]
PROBLE[S76,JMC]
PTCH2[ AI,JMC]
BOARD.ART[ AI,JMC]
PACKAR.LE1[LET,JMC]
SHUTDO[S76,JMC]
PUBLIC.DOC[LET,JMC]
PUBLIC.LE2[LET,JMC]
KNOW.PRF[F75,JMC]
KNOW.AX2[F75,JMC]
KNOW.CM1[F75,JMC]
MORAVE[S76,JMC]
PNUELI.LE1[LET,JMC]
PNUELI.LE2[LET,JMC]
FILE[S76,JMC]
CALIF[SEN,JMC]
STATEM.LET[SEN,JMC]
SAMPLE.LET[LET,JMC]
EXTENS[S76,JMC]
MAY76.IN[LET,JMC]
∂14-Mar-78 0959 LCW Seminar and Meeting
To: LLW, TM, PMF, JBR, LES, ME, TED, JMC
CC: LCW
Lowell advises me that he will not be able to come on Monday after all,
and TED and JBR say they will have a hard time going to the seminar, let
along the meeting, so we will have to forego the S-1 meeting after the
seminar and do it some other time.
Curt
∂14-Mar-78 1450 LCP Names taken in vain
You might be interested to know that the following example operator definition
appears near the top of page 71 of the Revised Report on Algol 68:
OP MC = (COMPL carthy,john) COMPL: (random < .5 | carthy | john);
Larry Paulson
∂14-Mar-78 2103 ME αβ. from line editor in E
∂13-Mar-78 0904 JMC αβ. from the line editor
What about saving a file while in line mode?
ME - Ok, all done, at last.
∂15-Mar-78 0615 MRC WHY254.MEM[DLN,MRC]
To: JMC, LES, Ryland at RUTGERS-10
Please read this file. It explains why 254. channels are better than one,
and describes a major change in the Initial Connection Protocol, and a minor
one in the host-host protocol. Having this feature will make both the design
and the implementation of the file transfer and mail protocols much less
painless and allow more facilities than a single-channel setup would allow.
-- Mark
"much less painless"? What you say is not entirely unmeaningless.
∂15-Mar-78 0935 JJK DEC donation?
The possibility of an equipment donation from DEC has come up in my
dealings with Dick Brucker, the DEC PDP10 and 20 salesman, and Peter
Raulerson, 11 salesman and local manager. Such a donation might be
similar to the lab's KL10 deal, or some other kind of cost splitting over
a Department computer. There are many possibilities in a variety of
machines: 20, 11, VAX, etc. Raulerson wants to talk to appropriate CSD
people to explore what we might want and how it would benefit us and them.
Would you be interested in meeting him for this purpose? He wants a
meeting some time next week. If you can't make it, can we discuss your
ideas about possible DEC donations before then? Thanks. (ps. My Polya
extension is 74269)
I will be available next week, but since we don't have money yet, the
discussion will have to be preliminary.
∂15-Mar-78 1133 LLL Appointments and Promotion mtg
john: am setting up a mtg. re: Doug Lenat. How does
mon 20th look? Tue?...laurie
It looks ok now. Put it in calend[ess,jmc] when you set a time.
∂15-Mar-78 1609 JBR S1 Meeting
To: TED, ME, LES, JMC, LLW, LCW, TM, PMF, JBR
The S1 meeting that had originally been scheduled for after
the S1 seminar has been rescheduled for 1:30PM Friday March 24 at
SAIL.
∂16-Mar-78 0609 MRC ICP protocol
To: Ryland at RUTGERS-10, JMC, LES
I have made major modifications to ICP.PRO[DLN,MRC]. You will probably
want to make a new copy of this file. I have also modified HSTHST.PRO,
but don't bother printing it since it hasn't settled down and is in fact
going to get some more major modifications.
∂16-Mar-78 0616 MRC WHY254.MEM[DLN,MRC]
To: Ryland at RUTGERS-10, LES, JMC
Have you looked at this document yet? I haven't received any comments
from anybody yet. Les, have I answered your objections to your
satisfaction?
∂16-Mar-78 1750 BCM cs226 term project
I will hopefully have my term project ready monday or tuesday.
Ben Moszkowski
∂16-Mar-78 2046 JMC*
Sato thesis for Richard Stark.
al 1,vew
et name2
∂17-Mar-78 0305 MRC new checksum algorithm
To: JMC, LES, Ryland at RUTGERS-10
CHECKSUM ALGORITHM
The packet checksum algorithm used is the result of a conversation with
Knuth. The checksum is 16. bits long and all of the packet header variables and
the entire data area. It does NOT include the packet trailer or the SOP/EOP
packet framing codes. Most data errors are expected to be of the form where
entire bytes will be picked or dropped; hence packet framing will catch such
errors. The checksum is intended to catch the more subtle bit picks/drops.
The algorithm is: (' means the number should be read as octal)
checksum := 1;
while newchar do checksum := (checksum * '013215 + newchar) & '177777;
In PDP-10 assembly code, this is:
; CHKBYT adds a byte to the checksum in SUM. At the beginning of each packet
; SUM must be initialized to 1.
; Call: MOVE BYTE,<byte from data stream>
; PUSHJ P,CHKBYT
; <return>
CHKBYT: IMULI SUM,013215
ADDI SUM,(BYTE)
ANDI SUM,177777
POPJ P,
∂17-Mar-78 0357 MRC host-host protocol
To: JMC, LES, Ryland at RUTGERS-10
Besides making the Knuth algorithm the official algorithm, I have also
added a couple of paragraphs in the introduction to explain the concepts
of the line transmission, host-host, and initial connection protocols,
and to explain why the line transmission and host-host protocols are
closely interrelated and are therefore documented together, while the
initial connection protocol is not.
∂17-Mar-78 0845 CET
To: JMC, EAF, RWF, GHG, JGH, DEK, JMC
February 17, 1978
TO: CSD Faculty and Research Associates
FROM: Gene Golub
SUBJECT: Support for new students.
URGENTLY need to know how many new students you can support.
Please call Gene Golub on 73124.
∂17-Mar-78 0845 CET
To: JMC, EAF, RWF, GHG, JGH, DEK, JMC
February 17, 1978
TO: CSD Faculty and Research Associates
FROM: Gene Golub
SUBJECT: Support for new students.
URGENTLY need to know how many new students you can support.
Please call Gene Golub on 73124.
∂17-Mar-78 2228 CLT
I had to special order it. You are welcome to borrow my copy, at least
until I get 206 organized and my term project finished.
Thanks, I would like to borrow it.
I have first comments on "Transforming LISP programs". This chapter
should have two aspects, theoretical (which you have) and computational.
In the published book, I think the computational aspect should dominate,
because I consider the book to be about how to program supplemented
with techniques for proving the programs correct - rather than
primarily about proving.
This chapter should take advantage of the fact that transformations
of LISP programs are conveniently expressed in LISP, because the
internal form of data matches the form of program and because many
operations on functions are conveniently described as functions.
The transformations you discuss can be made more concrete by giving
some of the the LISP programs that carry out the transformations
and giving some others as exercises.
I am also a bit concerned that you might go overboard on theory in
teaching the course.
What are the expressions like (IV.*.*) on p. 3?
Incidentally, what time are you leaving tonight, so I can get back
to you before then if my reading results in more for you to think
about?
∂17-Mar-78 2259 CLT
I was unsure whether or not to put in some of transforming programs.
I have an implementations of the purification program, one in the
style of EVAL as it appears in the notes and one base on abstract
syntax primitives. Or perhaps I could lift Ben's Cost Compiler.
I was basically trying to get the top level of the chapter down.
I don't expect there will be time to really polish to much before
it goes to press. More examples of programs can be handed out
in class and put on the system at LOTS for the students to
play with.
I will try not to be so abstract in my lecturing.
The expressions (IV.*.*) are a hack to stop the undefined label
message. They are generated by a reference to a label defined
in another chapter and will disapper when the whole thing gets
compiled at once. (Theoretically)
I'm not sure how long I will last tonight. For a hour or two anyway.
I will send you a message when I get ready to leave, in case you have
something to say.
∂18-Mar-78 0115 ME justification in E
∂17-Mar-78 2348 JMC
A small bug has crept into justification in E. It is my custom to
use Xjf 8 the first time and Xjf subsequently to get the paragraphing I
like. Now, curiously, the jf 8 doesn't work the first time although it
works on all subsequent occasions. The bug is but a day or so old.
ME - This is probably a result of my changing the default justification
mode from N to A. You can force it to be N by saying Xjf N 8 the first
time, in which case everything should be as it used to thereafter.
Possibly there might be some real bug, although I doubt it. Exactly what
do you mean by "doesn't work the first time"? Does it simply fail to
indent the first line of a paragraph? If so, that is probably because in
A mode it decides what you think is a new paragraph isn't one; switching
to N mode may fix things for you. If it doesn't, let me know. Note that
no comma should appear between the N and the 8 in the command.
∂18-Mar-78 0116 CLT
ABSTRA looks good. The Boyer Moore example(2) is now an exercise
(last section). I take it you want to delete the current section 1
and the LISP progam syntax section. I am not sure what you mean
by the example 3 comment.
As to more exercises and substantial problems, I planned not to
do much more for the current edition. I thought that during the
quarter I could work on collecting some and make up a few
handouts. They could be incorporated in the next round.
The algebraic expression reading and printing sounds interesting.
The example of simplification of algebraic expressions
containing differential operators (in ND and ZM draft) might
be a good extension of your example 1 for the substantial
problems section of ABSNTX.
∂18-Mar-78 0218 CLT
i am leaving soon. back early sunday afternoon
∂18-Mar-78 0422 MRC when our Dialnet modem gets installed
To: LES, JMC
We should try to get a 497 number, so calling other centrex numbers (eg,
LOTS) won't cost anything. Or at least so Roode claims (he suggested
that the current dialer should get moved, which I doubt will happen, but
it seems like a good idea for the Dialnet modem).
∂18-Mar-78 1016 MRC
∂18-Mar-78 1014 JMC Busy signals
Do you think it is feasible to tell whether the line is busy or
must Dialnet simply time out?
MRC - That sort of stuff is usually done in the hardware. I don't
know if the hardware we ordered is smart enough.
∂18-Mar-78 1354 SIRBU at MIT-MC (Marvin A. Sirbu, Jr.)
Date: 18 MAR 1978 1655-EST
From: SIRBU at MIT-MC (Marvin A. Sirbu, Jr.)
To: JMC at SU-AI
CC: SIRBU at MIT-MC
Dear Professor McCarthy:
Having nearly completed the study of electronic message
systems which brought me to see you last fall, I am now engaged
in a very different study where I also need your help.
M.I.T. has recently established a Committee on Future Computational
Needs and Resources chaired by Mike Dertouzos and Wes Burner to
prepare a 5-7 year plan for satisfying educational, research and
administrative computing needs at the Institute.
I have been asked by the Committee to visit several Universities
which have developed innovative schemes for providing computing resources
to students, and of course the LOTS facility at Stanford is high on
the list. Would you be able to see me sometime on Wednesday March 29
to discuss the reasoning behind Stanford's decision to acquire LOTS?
I would also be interested in speaking to one of the maintainers
of the system about experiences in using it.
I look forward to seeing you again,
Sincerely,
Marvin Sirbu
I plan to be at Stanford on that day and can see you. I suggest you
call Ralph Gorin 415 497-1360 to arrange a discussion of operational
matters and a visit.
∂18-Mar-78 1621 JJK Meeting with Raulerson of DEC
To: JMB, JMC, wiederhold at SUMEX-AIM
Still no date and time set for the meeting because Raulerson has
been out sick. I will let you know when he is back and a time is
set.
∂19-Mar-78 1145 MRC ERR code is too hairy
To: Ryland at RUTGERS-10, JMC, LES
I am becoming quite unhappy with the design of ERR. While the information
might be good for debugging, it is virtually useless for a production
mode, since most of the codes can't happen given debugged DCP's at either
end. I have already flushed the codes having to do with framing errors.
Considering the remaining codes:
000 Allocation has reached zero.
This is super random. Why the heck would any reasonable DCP need
to be told about this? It should know what it allocated, and what
it was allocated.
001-004 Sent xxx when {a|no} connection exists
These too are super random. No debugged DCP should cause these to
happen. It isn't clear what a DCP should do if it gets these.
005 Packet checksum error
This should be rare; it means framing was alright, but the
checksum lost. It might save the other DCP some work to know
not to continue down the window.
006 Illegal op code
Super random right now, but maybe useful later if more op codes
are added and a new guy talks to an old one (?).
007 Illegal window size
Super random; no debugged DCP should cause this.
010 Packet timeout
Is this really a good idea?
011 Allocation overflow
Another loser; no debugged DCP should cause this.
I propose flushing most of these, and replace with 000 Packet bad, re-
transmit starting at this packet (ie, checksum error), 001 Packet violates
protocol (you are losing). Maybe include 002 Illegal op code and
003 Packet timeout? What do you think?
It's still unclear what to do when you are told you are violating protocol.
Presumably with illegal op code if you are a version n DCP and you did a
frob that version n-1 doesn't have, you would try doing what you want some
other way. Suppose the other guy was reloaded but you didn't realize that.
Now all your packets going to (gone) connections will get packet violates
protocol. But a reloaded system is supposed to send resets to say it's
been reloaded.
Packet timeout seems to be in retrospect a bad idea. Suppose I become
catatonic. Eventually the other guy will give up on me because I'm not
talking to it any more. I am after all supposed to send those NOP
packets every so often. Suppose I really wasn't catatonic but line noise
zapped the end of my packet in an obscure. The other guy will see my NOP
packet and after flushing the mis-framed old packet will win again.
It seems that even with very large windows you can win without an
explicit error for bad sequence or timeouts or the like. Suppose I get
a packet with a bad sequence number. Now, if I know I haven't complained
about missing the packet I want already, then I can figure out that I
can lie and send a checksum ERR, without the danger of multiple errors,
because I won't ask for it again. A DCP can be written to have such
smarts in it without the protocol demanding them.
What do you think of making ERRs work like ACKs? In other words, they
would have packet number 0, and not be acknowledged.
-- Mark
∂19-Mar-78 1351 ZM
John,
I have a standing invitation to give a series of talks in Rio de Janeiro.
It seems that the best time for the trip is April. Would you recommend that I
stick around while the Dean "thinks" about my appointment, or is it irrelevant ?
Zohar
I think it's irrelevant, but I'll check.
∂20-Mar-78 0038 JBR My Datamedia
To: LES
CC: JMC
Martin Morf says that DCA will probably want it back sometime in June.
How about starting now on getting me a replacement?
∂20-Mar-78 0241 WD
To: JMC, WD
I might need your viewgraph projector. I wonder if you might send
me a note saying where it is so I could drop by an pick it up tomorrow
afternoon. It will return tomorrow evening.
Thanks,
Whit
The projector is in Susie's room.
∂20-Mar-78 0518 RUSSELL at USC-ISI Visit to Washington
Date: 20 MAR 1978 0509-PST
Sender: RUSSELL at USC-ISI
Subject: Visit to Washington
From: RUSSELL at USC-ISI
To: JMC at SU-AI
Cc: Carlson, Carlstrom, Hilda, Russell
Message-ID: <[USC-ISI]20-MAR-78 05:09:55.RUSSELL>
John,
I will be Dallas on 3 and 4 April. I expect to be in the
office on Wednesday, 5 April and would suggest a time either first
thing in the morning of else after lunch.
Will check with Dr Fossum, but you should know he is
in the midst of his congressional testiomony and extremely hard to
catch.
Will be happy to hear your proposals, but Congress
still expects us to paint a path toward the pay-off of any research
program we support. That continues to be the an important
consideration in any new thrust.
Looking forward to meeting with you and in
particular hearing about new ideas ready for exploitation.
Dave
-------
How about first thing in the morning on the 5th then? What time is
most convenient for you?
∂20-Mar-78 0558 CLT abstract syntax
I have begun incorporating your intro into the chapter. I still
need to complete example 1.
∂20-Mar-78 0924 LCW S1 SAIL Meeting
To: LLW, TM, PMF, JBR, ME, TED, LES, JMC, EJG
CC: LCW
LLW can't come to a meeting Friday, 24 March. Let's have the meeting
on Monday, 27 March, at 1330. Does anyone object?
Curt
∂20-Mar-78 1550 BS Ralph Gorin
Denny told me that Ralph Gorin will teach CS109A during Spring. I am
preparing appointment papers, but need to know whether we will pay him
for this teaching and, if so, how much. Denny said he would talk to you
about this, but in view of the admissions crunch he may not have had time.
Betty
I have not been involved in arranging this, but it seems to me that the
department should transfer to the LOTS budget whatever it normally pays
to a lecturer. This is under the assumption that Ralph cannot be paid
extra for the work.
∂20-Mar-78 1830 DCL
To: GROUP.DIS[VCG,DCL]:
***********************************************************************
VERIFICATION GROUP SEMINAR TUESDAY 21ST. MARCH
IS CANCELLED THIS WEEK FOR SPRING BREAK
****************************************
THERE WILL BE A SYSTEM MEETING ON THURSDAY 23RD; PLEASE SEND PREFERENCE
FOR TIME AND PLACE (AI LAB. or ERL) TO DCL.
∂20-Mar-78 2334 CLT INTRO2
I have copied intro2 into PREFAC[lsp,clt], and put in the pub
header. Perhaps if you make more modifications you could work
on this file. I wanted to make sure things didn't disintegrate
when I added this to the list of files to assemble. I dom't
plan to give a chapter number, just the unnumbered heading.
Is that ok?
That's all fine.
∂21-Mar-78 0449 RUSSELL at USC-ISI
Date: 21 MAR 1978 0444-PST
Sender: RUSSELL at USC-ISI
From: RUSSELL at USC-ISI
To: JMC at SU-AI
Cc: russell, carlson,
Cc: Carlstrom, Hilda
Message-ID: <[USC-ISI]21-MAR-78 04:44:31.RUSSELL>
In-Reply-To: Your message of MARCH 20, 1978
John,
Suggest 9 am on Wednesday morning, 5 April as a good time
to meet.
Dave
-------
∂21-Mar-78 0733 CARLSON at USC-ISI Your visit
Date: 21 MAR 1978 0733-PST
From: CARLSON at USC-ISI
Subject: Your visit
To: jmc at SU-AI
cc: carlson, russell, carlstrom, hilda
Sorry, but I will be on travel all that week. I am
very interested in hearing more about SU-AI plans
for the S1. Perhaps I can stop by the AI Lab and
talk to you on a trip in the near future.
Bill
-------
∂21-Mar-78 0939 Levin at SUMEX-AIM misc.
Date: 21 Mar 1978 0938-PST
From: Levin at SUMEX-AIM
Subject: misc.
To: jmc at SU-AI
john:
when you come to polya today for the a&p mtg at 1:30
could you stop by my office...will give you green forms, + i need
your signature....thanks....laurie
-------
∂21-Mar-78 1357 Levinthal at SUMEX-AIM EPRI meeting
Date: 21 Mar 1978 1356-PST
From: Levinthal at SUMEX-AIM
Subject: EPRI meeting
To: jmc at SU-AI
cc: tob at SU-AI
John, The meeting withMilton Levenson is wednesday march 22nd
9 am 3120 Hillview Bldg 3 top floor. Let me know if you will
be able to come. Elliott
-------
∂21-Mar-78 1420 STT CS224
Ed is thinking of three different CS224 talks he would like you to help with-
1) Tues May 9 History of AI panel disc. McCarthy, Samuel, Feigenbaum.
2) Thurs May 18 Talk on formalisms for representing knowledge. McCarthy.
3) Tues May 23 Panel disc. on Representation. McCarthy, Hendrix, Winograd,
Bobrow, Feigenbaum.
Are these all right with you? Schedule is still tentative, as you might expect.
All class meetings to be 11:00-12:15.
Also, could you recommend a paper for students to read as background for the
talk on knowledge formalisms, something on the order of 30 pages? Margaret
Boden's book Artificial Intelligence and Natural Man will be the main text for
the course, and if I remember right she doesn't say much about formalisms for
knowledge and belief.
Thank you
∂21-Mar-78 1729 CLT book
There is a copy of the current assembled version on my desk. (Complete
with a few pub bugs that are now fixed.) I have changed the Transformations
chapter a bit. I plan to return around 11 or 12 pm and repub
a final version. The hope is to get a printable copy to the printers
tomorrow early. A objections or corrections??
∂21-Mar-78 1748 LCW S1
To: JMC
CC: LCW
Relative to our talk this afternoon: The most effective thing that you
could do now to aid in the progress of the S1 Project towards a useful
Mark I and thence to a successful Mark II is to channel recruits to us.
Our primary bottleneck is lack of qualified, highly motivated people to do
either hardware or software design and analysis; interesting projects are
available in both areas. We have tapped the CS department, and now the AI
Lab, but have only approximately half-filled our current capacity. GSRA
or LLL employee positions are available. I would appreciate it if you
would tell your friends.
Curt
∂21-Mar-78 1823 DCL
To: JMC, LES
CC: TOB, JC
REPLY TO:
*****
∂22-Jan-78 1630 JMC Time spent waiting for the computer.
To: DCL, SMG, RAK
CC: LES
According to my computation, DCL spent 8.64% of the time he was
logged in waiting for the computer during the first half of January,
and SMG spent 2.1% of the time he was logged in waiting. The crude
calculation involves just multiplying the CPU% over logged in time
by the load factor. The champion waiter is Leland Smith who spent
15% of his logged in time waiting for the computer.
I regard this estimate as rather crude and would welcome suggestions
as to other ways of measuring the quantity, but it certainly indicates
that no-one would be much more productive if the computer were
faster except that maybe people are deterred by objectively small
but subjectively large amounts of waiting. In particular, time
spent waiting for an outcome one fears will be unsuccessful,
whether because of a machine or system failure or one's own bug,
must subjectively count as a larger amount of waiting.
*****
REPLY:
John,
In response to your message of Jan 22nd I have polled a small sample
of big job users and have prepared a resume of their suggestions
under the four headings below. The raw reactions are saved on
BUREAU[VCG,DCL] in case you wish to refer to them for details.
1.CHANGES IN STATISTICS GATHERING METHODS.
There is general agreement that the current scheme of dividing A,B,C time usage
by relation 1,2,3 is not realistic and tends to hide the real gluttons.
It is argued that our facilities are so heavily overloaded that A time is
useless for any real computation at all, that B time is getting that way often,
and that resort to C time is all that is open to many of us. Therefore there
should be no reduction factor in machine use statistics for use at any time.
The misleading use of our statistics to do anything more than measure
cycle-soaking can be illustrated as follows:
In your message above you estimate that LCS is the champion at waiting.
The truth is that he always has at least two jobs running simultaneously,
thus contributing to his own waiting-time statistic (as computed above)
while not waiting necessarily at all:
∂16-Feb-78 1740 SMG Where have all the cycles gone?
.WHE LCS
22 MLG,LCS XM 242P IOWQ* 5:12 0:01 III-20
25 MLG,LCS XM 228P DWQ * 0:19 0:00 DD-36
55 MLG,LCS XM 228P DWQ * 1:18 0:01 III-22
2.WHAT SHOULD BE POLICED NOW
There is general agreement against policing as a means of reducing machine usage.
It is argued that certain changes in the system (software and hardware, some
of which you are already contemplating) together with a policy for restricting
future usage so that demand does not again outstrip any increased supply, is best.
Requests for immediate policing seem mainly concerned with:
A. Games.
B. Encouraging courses to smooth out the usage over the whole quarter and not
bunch up in panic mode at the quarter's end.
We could put a CPU-time/month limit on each course user, based on our statistics.
C. Courses perhaps to be restricted to certain times.
D. Datadisk use during the afternoon, A-time, to be restricted.
3.SYSTEM CHANGES THAT WOULD HELP
A.Changes to the monitor scheduler.
The scheduler doesn't seem to know how to handle multi-segmented jobs.
Watching the two segments (80P and 100P) alternating residency in core is
very frustrating.
B. More core.
C. Demand Paging. Even at the risk of initial period of unreliabilty.
4.A POLICY FOR THE FUTURE (SO POLICING IS NOT NECESSARY)
There is general agreement that we should devise a policy of allowable machine
users in various categories (Arpa project, NSF project, courses, department,
unpaid visitors, etc.), limits etc. now.
The hope is that this will stop improvements in system capacity from being
soaked up and our present demise from occuring all too soon again.
We should meet NOW on this.
∂21-Mar-78 2301 MRC ARPAnet Resource Handbook
To: LES, JMC, JBR, ME, BH, JC, PMF, RPG, ES, LCW, REM
Our entry in the ARPAnet Resource Handbook is online as SUAI.TXT[NET,MRC].
I have already substancially updated it to reflect many system changes,
particularly in the network software.
You are invited to submit contributions to this write-up. In particular,
I'm interested in program descriptions for subsystems which we are willing
to make available to network users for usage here or or export. For
example, programs like the music group's and S-1 group's software aren't
represented at all, neither are some commonly used utilities like POX.
Please do NOT update SUAI.TXT itself--it's a readonly file (all writes are
reserved). MAIL your updates to MRC. This isn't for any bureaucratic
reason; it's just to prevent screws with two people trying to update the
file at the same time, and to make sure all entries are in proper format.
General rules:
(1) Keep it short and simple.
(2) You will simplify my task greatly if you send it to me in
the proper format to begin with. See the example below
for "proper format".
(3) Stick to standard ASCII (no SAIL graphics). Also, tabs are
a no-no (I will replace tabs with the right number of
spaces) since they blow NLS' mind.
(4) I will not make any editorial changes other than making sure
the format is right. Please make sure you want to have
it published as you said it! Only in an exceptional case
will I change anything.
A sample program description is:
(E)
TYPE: Display text editor
CONTACT: Arthur Samuel (ALS@SU-AI) (415) 497-3330
DESCRIPTION: E is a text editor used from local display terminals
at Stanford. It uses random access, edit-in-place operation, so
small changes to large files are done very much faster than in
recopying editors such as SOS or TECO. Changes to the text within a
line are made using the line editing facilities in the timesharing
monitor, which gives fast response to editing commands since E need
not be swapped in and scheduled for these commands.
E is currently the popular text editor at SU-AI.
ACCESS:
[.]etv <SP> FILENAME <CR>
NETWORK USE PARAMETERS:
Although E can be run from a terminal which is not a
local Stanford display, it is not designed for such
use; many of its facilities are not available in that
mode of operation.
Datamedia 2500 terminals (and SUPDUP terminals) are
"local Stanford displays" and enjoy E's full facilities.
DOCUMENTATION:
1. Online: E.ALS[UP,DOC]
∂22-Mar-78 0004 MRC Announcement and request for comments
First, Dialnet@SU-AI sends a message to the currently active Dialnet
people (mrc, jmc, les, ryland) and more can be added as time progreses.
Next, two closely related issues that I think should be considered in
file transfer are data compression and data encryption. REM claims he
has an all new, wonderful algorithm for data compression which he is
patenting/copyrighting/going to become rich selling... I don't know
how much of this is real and how much is REM. But clearly we are going
to have to consider some form of data compression. I really don't know
where to look in poking around on that.
Data encryption is another thing that should be considered. The really
tough questions are how it's to be done. It's easy to say Rivest codes,
but there's more to it than that. In particular, suppose we want this
to be good enough for the military to use (and I see no reason why it
shouldn't be in principle). Communications between two red points over
a black medium require extensive encryption to ensure that a tapper can
not obtain any useful information. To start off with, this means that
not just the data part has to be encrypted--the entire packet must be
encrypted! In theory, a spy might get some useful information from
knowing which packets are data and which are acknowledgements, which
are commands, allocates, etc; although Rivest codes do a good job of
zapping that. Of course, the flexibility that black sites have in
their packets is a luxury which red sites cannot afford--all packets
must be fixed sized, and sent out at a regular interval, to avoid
traffic analysis. Dialnet helpfully provides no-op packets which
will help in this, but the result means (of course) a lower bandwidth
and more overhead. A magic hardware box, such as are installed on
Pluribus IMPs on red sites on the ARPAnet, does help a good deal.
The biggest problem is the no-op. All no-op packets are equal, and
it spazzes up the line transmission protocol to make the encryption
depend upon the packet sequence number (the error checking can't win
if it changes in that way). This might be a defect in the line
transmission protcol, but my proposed solution is to send random
data in the data part of the packet instead of zeros. Thus no two
no-op packets will be alike, therefore it should be impossible for a
villian to tell which is a no-op and which is a real packet.
SOP/EOP markers are another problem. I am not familiar with military
security procedures for data transmission. I presume that it is
acceptable in a packet setup to allow a spy to know where a packet
begins and ends? I can't see how SOP/EOP information can be
encrypted, and also a spy can figure that out for himself unless
the lines in both directions are continuously transmitting without
a break.
For interesting reading, refer to BBN Report 1822, the IMP-Host
protocols, specificlly the Appendix dealing with the Pluribus IMP
and the data encryption options for red/black/red communications.
-- Mark
∂22-Mar-78 0614 RWG on TTY26 (at TV-61) 0614
what have you heard of "Death from Falling Watermelons", a book by C Adler?
∂22-Mar-78 1319 SIRBU at MIT-MC (Marvin A. Sirbu, Jr.) Re: meeting
Date: 22 MAR 1978 1619-EST
From: SIRBU at MIT-MC (Marvin A. Sirbu, Jr.)
Subject: Re: meeting
To: JMC at SU-AI
I'm sorry that we won't be able to get together as I had
hoped to be able to share with you the results of our study
for the FCC, and the implications for efforts such as Dialnet.
Thank you for putting me in touch with Mr. Gorin.
Marvin Sirbu
∂22-Mar-78 1223 RWW NETWORK DIRECTRY
ARPA.DIR[W78,RWW] contains a write up about FOL for the arpa net directory.
If you think it's ok I'll give it to MRC to send off for inclusion.
rww
∂22-Mar-78 1430 DCL
To: GROUP.DIS[VCG,DCL]:
***********************************************************************
VERIFICATION GROUP SYSTEM MEETING THURSDAY 23RD MARCH
PLACE: AI LAB
TIME: 3:00 pm.
TITLE: TAKING STOCK: WHERE ARE WE NOW?
DISCUSSION 1. Rule language
2. Top Level
3. User manual
4. Facilities for Runcheck
5. What next?
∂23-Mar-78 0314 MRC Privilege issue comes up again
To: GRIPE, JMC, LES, JBR, ME, BH
AS A RESULT OF THE XGP LOSSAGE, I PURGED THE QUEUE AND PUT UP A MESSAGE
SAYING NOT TO QUEUE ANYTHING. I AM UNABLE TO BRING THE XGP SPOOLER DOWN
BY DETACHING THE XGP BECAUSE THAT IS A PRIVILEGED FUNCTION, ONE WHICH THE
POWERS THAT BE HAVE CHOSEN NOT TO GRANT ME.
THIS MEANS THAT ANYBODY WHO TRIES TO USE THE XGP ANYWAY (SAY WHO LOGS IN
WITH THE . OPTION AND DOESN'T SEE THE WARNING) WILL BE, ALONG WITH THE
CTY, HARASSED FOR AN INFINITE PERIOD OF TIME TO FIX THE XGP, WHICH CANNOT
BE DONE. I WAS UNABLE, EVEN BY BABYSITTING IT, TO KEEP IT FROM GETTING A
PAPER JAM ENOUGH TO BE ABLE TO GET MY (SMALL) LISTING PRINTED.
THERE IS AN ALTERNATIVE WAY; TO KILL THE SPOOLER AND TO TELL THE SYSTEM
NOT TO REVIVE IT. THIS IS NOT PRIVILEGED, BUT IT ALSO MEANS THAT THE
PERSON WHO FIXES THE XGP HAS TO KNOW THE UUO MAGIC TO RE-ENABLE THE XGP
SPOOLER. I HAVE NOT DONE THIS. THIS IS ALSO LESS DESIREABLE THAN
DETACHING THE XGP, SINCE THE QUEUEING PROGRAM SAYS THAT THE XGP IS DOWN
WHEN IT IS DETACHED.
SURE I HAD (SEVERAL) WAYS TO GIVE MY JOB THE DEV PRIVILEGE. HOWEVER, I AM
WORKING WITHIN THE SYSTEM, NO MATTER HOW SCREWED UP THE SYSTEM IS. I
SUGGEST THOUGH THAT THE SYSTEM BE RECONSIDERED. HOW MANY MORE TIMES ARE
THINGS LIKE THIS GOING TO HAVE TO HAPPEN BEFORE THIS UNREASONABLE POLICY
GETS CHANGED?
∂23-Mar-78 0434 MRC more host-host protocol changes
I have removed all timeouts. I feel that a timeout interval should be
left up to the implementors, to allow him/her to select an optimal time-
out interval.
The requirement that NOPs be sent every five seconds is enough to take
care of missing packet timeouts. I have changed the way an out-of-sequence
packet should be sent. A packet that is too low gets an ACK for the packet
we are at now, a packet which is too high gets an ERR for the packet we are
expecting. So whether an ACK or a packet is missed, it's gonna get caught
when the next good packet comes through, be it a real packet or just a NOP.
Eventually things will settle down. I think the Bell people even suggested
something like this, but I didn't quite understand what they meant at the
time (I wish I had more time alone with their hackers).
The other two time-outs are for host dead and connection crufty. I figure
each host can decide for themselves on that. If they make an unreasonable
decision, they are only screwing themselves.
HSTHST.PRO has been suitably updated.
-- Mark
∂23-Mar-78 0451 MRC ACK vs. NOP
I am on the verge of considering flushing these as two separate entities.
A no-op can be done quite effectively by re-acknowledging the last packet
you got! It also makes it simpler for the DCP not to have to worry about
two types of packets.
This means that the initial connection protocol will need to be changed.
Suggestions?
∂23-Mar-78 0458 100 : J. Weizenbaum via MIT-ML
where will I find the Lederberg review you had filed? Joe W.
∂23-Mar-78 1217 PAT airplane reservations
Don't forget to tell me your flight preferences...patte
∂23-Mar-78 1417 Raphael at SRI-KL (Bert Raphael) Grigory Goldshtein
Date: 23 Mar 1978 1416-PST
From: Raphael at SRI-KL (Bert Raphael)
Subject: Grigory Goldshtein
To: Hart, Sacerdoti, Bolles, McCarthy at SU-AI, PHW at MIT-AI
cc: Raphael
I have just learned that Grigory Goldshtein has been found guilty of
"parasitism" and sentenced to one year. (He was fired from his job
years ago and is now guilty of the "crime" of not working!) The nature
of the sentence-- where he will be imprisoned, etc.-- is unclear. His
brother Isai has not yet been arrested, but the threat is there.
It is believed that letters from here can be of tremendous help. We
should inquire for explanations and details, mention our concern, past
contact, and personal interest, etc. Letters may be addressed to:
Mr. Pakidze, Procuratur
Tbilisi, Georgia, USSR
Alexei Inauri
Chief, KGB
Tbilisi, Geogia, USSR
Ilia Vekua
Vice-Chairman, Georgian Academy of Science
S. Euly Street 5
Tbilisi 380086, Georgia, USSR
Prof. V. V. Chavchanidze
Director of Cybernetics Institute
S. Euly Street 5
Tbilisi 380086, Georgia, USSR
and any other people you may know of any stature in the Tiblisi
community.
Please send copies of any letters or wires you send, to:
Regina Waldman
Bay Area Council on Soviet Jewry
106 Bazen St.
San Francisco, CA 94131
Thanks!
Bert Raphael
-------
∂23-Mar-78 1532 JJK Meeting with Raulerson of DEC
To: JMC, wiederhold at SUMEX-AIM
We have set up a meeting for 1:30pm, Monday, March 27, in my office
in Polya. Does that fit your schedule? We can reschedule if not.
∂23-Mar-78 1613 JJK New times for Raulerson
To: JMC, JMB, wiederhold at SUMEX-AIM
Because of JMC conflict, Monday time is rejected. Raulerson can go
for some time on Thursday or Friday, March 30/31. Please let me know
what times of either day you can make it so we can reach an agreeable
intersection. Polya Hall site preferred.
∂24-Mar-78 0934 TOB
John
Your measure of waiting time is interesting. I would propose another
which is more important to me. The ratio of real time to machine
time from time of finishing a command string to time of response.
Console time includes talking with students, looking up references,
etc.
Particularly in editing, I find the response so slow that it is no
longer possible to interact by waiting for response. I need to calculate
the number of lines and proceed by dead reckoning.
I disagree with the conclusion. "No one ...."
Tom
∂24-Mar-78 1016 Tajnai at SUMEX-AIM Bengt Aspvall
Date: 24 Mar 1978 1016-PST
From: Tajnai at SUMEX-AIM
Subject: Bengt Aspvall
To: JMC at SAIL
cc: DPB at SAIL, Tajnai
I need help with a problem.
Bengt Aspvall entered the MS program autumn 1976-77. He was admitted
to the Ph.D. program Autumn 1977-78. Don Knuth was his advisor last
year and Bengt said Prof. Knuth treated him as a first-year Ph.D.
student.
Bengt got a Masters pass on the comp Jan. 1977. He was only going to
get a Masters pass on the May 1977 comp. so he did not take the
programming project and therefore his record shows "fail."
He passed the Comp with a Ph.D. pass February 1978. I gave him his
application for Ph.D. candidacy. He put down his intention to pass
a Qual by Spring 1979. He plans to take a qual spring 1978, but
doesn't want to be dismissed if he fails. All second year students
are given until Spring 1979 to pass the Qual.
THE PROBLEM: Bob Floyd is acting as Bengt's advisor in Don Knuth's
absence. Bob feels Bengt should be treated as a third year student
with the requirement that he pass the Qual Spring 1978. He refuses
to sign the petition and said to refer it to you. He asked Betty to
look it up in the rule book and all she could find is,
9/30/71 Faculty Minutes, p. 320 policy is quoted
"If admitted only for the M.S., a student must reapply for admission
after he takes an M.S."
Bengt has a letter from Prof. Feigenbaum June 1977 saying he has
several more chances to pass the comp. Obviously a form letter, but
nevertheless signed.
Bengt needs to get his application for candidacy signed in order to
be eligible for the higher salary he is entitled to as of April 1
(admitted to candidacy rate).
There are other students in the same situation, and I do not feel that
Bengt should be a test case.
Prof. Floyd thinks this should be resolved. His point is that the
students should be able to pass these exams quicker if they have been
here as a Masters student. He also said students "who had any smarts
would come in as masters students then transfer to the ph.d. program
in order to have more time to pass all the exams."
I have probably given you too much detail, but I wanted you to have
as much background as possible.
Will you sign Bengt's Ph.D. application for candidacy?
Thank you for your assistance,
Carolyn
-------
∂24-Mar-78 1249 DPB Aspvall's candidacy
To: JMC, tajnai at SUMEX-AIM, RWF
∂24-Mar-78 1016 Tajnai at SUMEX-AIM Bengt Aspvall
Date: 24 Mar 1978 1016-PST
From: Tajnai at SUMEX-AIM
Subject: Bengt Aspvall
To: JMC at SAIL
cc: DPB at SAIL, Tajnai
I need help with a problem.
Bengt Aspvall entered the MS program autumn 1976-77. He was admitted
to the Ph.D. program Autumn 1977-78. Don Knuth was his advisor last
year and Bengt said Prof. Knuth treated him as a first-year Ph.D.
student.
Bengt got a Masters pass on the comp Jan. 1977. He was only going to
get a Masters pass on the May 1977 comp. so he did not take the
programming project and therefore his record shows "fail."
He passed the Comp with a Ph.D. pass February 1978. I gave him his
application for Ph.D. candidacy. He put down his intention to pass
a Qual by Spring 1979. He plans to take a qual spring 1978, but
doesn't want to be dismissed if he fails. All second year students
are given until Spring 1979 to pass the Qual.
THE PROBLEM: Bob Floyd is acting as Bengt's advisor in Don Knuth's
absence. Bob feels Bengt should be treated as a third year student
with the requirement that he pass the Qual Spring 1978. He refuses
to sign the petition and said to refer it to you. He asked Betty to
look it up in the rule book and all she could find is,
9/30/71 Faculty Minutes, p. 320 policy is quoted
"If admitted only for the M.S., a student must reapply for admission
after he takes an M.S."
Bengt has a letter from Prof. Feigenbaum June 1977 saying he has
several more chances to pass the comp. Obviously a form letter, but
nevertheless signed.
Bengt needs to get his application for candidacy signed in order to
be eligible for the higher salary he is entitled to as of April 1
(admitted to candidacy rate).
There are other students in the same situation, and I do not feel that
Bengt should be a test case.
Prof. Floyd thinks this should be resolved. His point is that the
students should be able to pass these exams quicker if they have been
here as a Masters student. He also said students "who had any smarts
would come in as masters students then transfer to the ph.d. program
in order to have more time to pass all the exams."
I have probably given you too much detail, but I wanted you to have
as much background as possible.
Will you sign Bengt's Ph.D. application for candidacy?
Thank you for your assistance,
Carolyn
-------
[Policy is: Bengt is now a second-year PhD student. He should be treated
like one. His previous year as a Masters student is not counted against
him. It is true that most students will be able to pass the exams earlier
if they are masters students first. The counter to Floyd's suggestion that
students come as MS students first and then transfer to the PhD program is
that one cannot "transfer to the PhD program." The full admissions process
is applied to EVERY PhD admission. Bengt is the only such from his entering
class. Carolyn Talcott is the only one from the new (78) admittees.
Don last year at the Gray Tuesday meetings made a point that our records
show that Bengt entered as a PhD student in Fall 76. Bob should sign the
Candidacy application, unless he has other objections to Bengt's application.
I will talk to Bob at the meeting this afternoon of the theory faculty.
-Denny]
∂24-Mar-78 1614 Levin at SUMEX-AIM misc.
Date: 24 Mar 1978 1614-PST
From: Levin at SUMEX-AIM
Subject: misc.
To: jmc at SU-AI
hi john:
1. i have a form for you to sign.
2. a student of yours, Ben Moszkowski, left a paper @ polya
that he did to be graded.
are you planning to be here next week...or shall i send it to AI lab?
...laurie
-------
∂24-Mar-78 1703 MRC Another incompatable change!
Per suggestion by Jeff Rubin, EOP has been changed from 226 to 256.
∂24-Mar-78 1727 Wiederhold at SUMEX-AIM Re: Meeting with Raulerson of DEC
Date: 24 Mar 1978 1717-PST
From: Wiederhold at SUMEX-AIM
Subject: Re: Meeting with Raulerson of DEC
To: JJK at SU-AI
cc: jmc at SAIL
In response to your message sent 23 Mar 1978 1532-PST
Friday or Thursday are fine, Thursday is preferred. I guess Jeff Barth
wont be back in time. Maybe Susan Owicki can join ? Gio
-------
∂25-Mar-78 1732 PAM
Thanks... his apparently pre-school kid called here for him, and I was trying to
find him...Paul
∂25-Mar-78 2305 MRC Proposed change in Dialnet line transmission protocols
To: Moon at MIT-MC, Ryland at RUTGERS-10, JMC, LES, JBR
The ARPAnet does packet framing using ASCII DLE STX for start of
packet, and DLE ETX for end of packet. This of course is what
ASCII considers to be "right" and may be a bit better than having
a single byte start and end of packet marker. What do you think?
Also, the CHAOS net (and I think the ARPAnet) has an acknowledgement
packet number as part of the packet header. This is used to
have the highest acknowledged packet number (and, as in Dialnet,
automagically acknowledges any lower packets), so the functions of
acknowledgement and data/command are combined. This way, the
ACK packet can be flushed. If one wants to just acknowledge, they
can use a NOP.
Opinions? To make quoting less obnoxious, I would use DLE with the
200 bit on, so only that flavor of DLE need be quoted, so a 7-bit
communication won't be slowed down if it has a lot of control-P's
in it.
∂26-Mar-78 0450 REM Another winning thing about my new compression algorithm
To: CRUNCH.DIS[1,REM]: CRUPUB.DIS[1,REM]:
I have just discovered that my algorithm is truly finite-state for
both crunch (new result) and uncrunch (known before). For crunch
it is impossible to have 0 followed by an infinite string of 111111...
because that would imply always selecting the least-used character,
which would thus cycle thru all tokens, one of which is <EOF> which
would immediately terminate. (All this depends on my particular way
of tabulating statistics in adaptive-algorithm, but it's a reasonable
way to do it.)
∂26-Mar-78 0917 REM via AMES-TIP#20 Working FAIL program
To: CRUNCH.DIS[1,REM]:
RU C[1,REM] for demo of crunch and uncrunch using my new algorithm INTCOM.
It uses single-character perfect-code (better than Huffman code) updating
the histogram (used for encoding or decoding) with each character processed,
initialized with all counts equal. It uses Pseudo-Baudot coding.
Type "C" to crunch, you type plaintext, it types bits.
Type "U" to uncrunch, you type bits, it types plaintext.
∂26-Mar-78 2151 KRD lisp history
Is there available a written version of the lisp history you were talking
about tonite at Ed's house? A file I can list?
Randy
Yes, but it needs a new PUB preamble and to be repubbed. I'll ask Patte
to do it and send you a copy. Please MAIL her where you want it sent.
Please put LISP[f77,jmc] back in a form where it can be pubbed for
8 1/2 by 11 printing and print some copies for local distribution.
Randy Davis wants one and will MAIL you where to send it.
∂26-Mar-78 2258 MRC via SU-TIP#1 Nova AI program
To: JMC, phw at MIT-AI
i saw the repeat of it on tonight. i thought it was pretty good, although
it seemed rather unfair to give joe w. as much attention as he got. i
guess they were trying to give the opposing viewpoint equal billing by
having him and dreyfuss on...although i sort of thing it might have been
better to let his viewpoint be seen as the minority it is (well, dreyfuss
didn't really say anything that seemed to claim it was impossible, just
that it was very hard to do).
i thought that terry's presentation was perhaps the most impressive of
the whole show (well, after all the blocks world was a major event). i'm
glad that there wasn't an excessive amount of the crowd pleasing stuff.
too bad as much time was spent on DOCTOR as it was.
-- mark
Weizenbaum's main role seemed to be to provide contrast, and some of
the juxtapositions were quite artificial. For example, the statement
of Weizenbaum's just after mine was obviously referring to something
different than what I had said. The program didn't seem to explain much.
∂27-Mar-78 0532 RDR via AMES-TIP#51
May I list you as an employment reference concerning my work at LOTS?
All in all, would your comments be favorable?
They would mention both strengths and weaknesses, but I couldn't say
much one way or the other about your programming ability if that is
relevant.
∂27-Mar-78 0941 JJK
Thursday seems better for Raulerson meeting. Time preference?
Anytime.
Second thought: morning better.
∂27-Mar-78 1217 JJK Raulerson/DEC
To: JMC, wiederhold at SUMEX-AIM
Meeting will be Thursday March 30 at 10am in my office, 214 Polya.
∂27-Mar-78 1304 RAK
To: GROUP.DIS[VCG,DCL]:
***********************************************************************
VERIFICATION GROUP SYSTEM MEETING TUESDAY 28TH MARCH
PLACE: DSL meeting room on campus and subsequent
adjournment locations
TIME: 2:30 pm. SHARP
TITLE: PIZZA LUNCH
DISCUSSION IMPORTANT--Attendance at this important seminar is restricted
to those who have placed $3 in an envelope (small bills please) with their
name on it in my box at the AI lab by NOON on Tuesday. I will NOT buy
more pizza and beer than the total amount I have received will justify.
Exception--people at DSL who want to come may let me know by a note;
with payment at the seminar -- none of you have exhausted your credit
rating with me yet. But for everyone else -- and I do mean all you
deadbeats in the verification group, not excluding academic staff --
I want that green in advance.
Dick
∂27-Mar-78 1552 100 : REM via SU-TIP#1 In reply to your question about what compression ratios I'm getting
The test program in FAIL completed a day ago uses TTY (manual) input and
output, thus cannot be used on "typical" data such as a disk file or a
session full of computer output. Next stage is to install disk I/O
so it can crunch a file. It starts out with no knowledge, and quickly
learns the type of data it is processing (but null context at present).
If you type English, after you get tired of typing (about 100 characters)
it hasn't learned much, your type-in is probably mostly non-repeats hence
not much if any compression. If you type in lots of repeats, it quickly
learns, and for example might generate one bit per 20 characters if data
consists all of same character. But, the real test will be with files
and with terminal output....
I tried repeating the same character at it, and it seemed to get into
a periodic state where it generated about one bit for every 10 characters.
I typed a lot of sentences at it, and didn't get as good a result as
I would have expected. There may be a bug, but it will surely turn up
when you can point it at files.
∂27-Mar-78 1807 Feigenbaum at SUMEX-AIM Re: Manna
Date: 27 Mar 1978 1807-PST
From: Feigenbaum at SUMEX-AIM
Subject: Re: Manna
To: JMC at SU-AI
In response to your message sent 27 Mar 1978 1718-PST
John, concerning the Manna material, will you mail it to me electronically
at SUMEX? I'll look at it on Tuesday (tomorrow)....Ed
-------
∂27-Mar-78 1808 100 : REM
Hmmm, depends on parameters. I have INTCOM set to N=11 and J=100, and
I guess that only gives a 70:1 compression ratio on series of same character.
Still that's a lot better than Huffman limit of one bit percharacter (7:1).
With N=12 or more should do better. Try uncrunching and it should exactly
invert (modulo a delay of about 11 bits due to N-bit internal state N=11),
if so no actual bug exists.
I don't know how to uncrunch, there not being a file.
∂28-Mar-78 0122 LES EFIND
If this is what you want, we can ask him to put it up.
∂25-Mar-78 0254 RSP EFIND
[1,RSP] contains EFIND.SAI, EFIND.DMP, AND EFIND.RSP (documenttation!).
In addition, PHONES.PTY[1,RSP] has been kludged into the appropriate
format as a test case.
I've run it out of TEST.DMP[1,3] and as a user program. Let me know
if you find any problems or have further suggestions.
∂28-Mar-78 1025 Wiederhold at SUMEX-AIM Re: Raulerson/DEC
Date: 28 Mar 1978 1024-PST
From: Wiederhold at SUMEX-AIM
Subject: Re: Raulerson/DEC
To: JJK at SU-AI, JMC at SU-AI
In response to the message sent 27 Mar 1978 1531-PST from JJK at SU-AI (Jonathan King)
10 am it is. Do you expect we will be free before lunch? Gio
-------
∂28-Mar-78 1141 100 : REM via SU-TIP#2 HOW TO UNCRUNCH USING C.DMP
Give U command at start, then each time it gives you comma-backspace
(comma conrol-H) (comma lambda on displays) you overprint the comma
with a "0" or "1" character. After the 11-bit buffer is full it
will start emitting plaintext as you feed it additional bits. For
fun, crunch long strings of the same character, then restart program
and uncrunch and watch it magically uncrunch each bit (after the
first eleven) into lots of characters. (To get backspace, instead of
ugly lambda, ask JBR or MRC how to SUPDUP to SU-AI using a DM simulator
mode which converts λ into backspace effect on datadisk.)
∂28-Mar-78 1657 FEINLER at SRI-KL SU-AI write-up - Request for update
Date: 28 Mar 1978 1651-PST
Sender: FEINLER at SRI-KL
Subject: SU-AI write-up - Request for update
From: FEINLER at SRI-KL
To: mrc@SU-AI
Cc: jmc@SU-AI
Message-ID: <[SRI-KL]28-Mar-78 16:51:40-PST.FEINLER>
Dear Liaison,
It is time to revise the Resource Handbook again. The SERVER write-ups
for the host or hosts for which your are Liaison are located in the
files
<NETINFO>SU-AI.TXT on SRI-KL (66 dec)
and may be FTPed by invoking FTP at your local host, logging in as
'anonymous' followed by your lastname as password, and pulling the file
to your directory or lineprinter.
These write-ups are to be included in the new edition of the Resource
Handbook. Please make any changes, additions, or deletions and return
as soon as possible. Deadline for changes is no later than
*FIRM!******************* April 14, 1978 *******************FIRM!*
Return write-ups online to
FEINLER@SRI-KL
Return hardcopy via U.S. mail to
Elizabeth Feinler
Network Information Center, J2021
SRI International
Menlo Park, California 94025
Users would particularly like to know: 1) who are the the proper
contacts for your host, 2) what are the network interests and/or
research at your site, and 3) what software do you have that others can
use on your computer or transport to their own computers.
NOTE: In the past several Liaison mistakenly thought that announcing
software as 'available' meant it was free and maintained for novice
users. This does not have to be the case (although it certainly is
not discouraged!) 'Available' means only that the software has been
implemented and that it can be shared in some fashion across the
network. 'Software' can be anything from operating system releases
to user programs. It is understood that the people controlling the
software may set whatever conditions seem appropriate for an
interchange with other users.
Last year some Liaison, due to local difficulties, were unable to obtain
the online version of their resource handbook host write-ups in time.
Therefore, in addition to the online notice and write-up, we are sending
each of you the same information in hardcopy via U.S. mail. We hope
that this will cover most contingencies, and will make it as convenient
as possible for you to provide the NIC with the needed information.
More and more users are coming onto the ARPANET. They are frequently
dependent upon the information supplied by the Liaison to find their way
around the net. I hope that all of you will provide write-ups for them
to reference, and use this opportunity to publicize interesting work
going on on your hosts and useful services that are available.
Regards,
Elizabeth Feinler, Mgr.
Arpanet Network Information Center
Les, Will you help Mark do the right thing?
∂28-Mar-78 1703 MRC via SU-TIP#1
I already updated the resource handbook entry.
It is on SUAI.TXT[NET,MRC]. I haven't sent it to Jake yet.
-- Mark
∂28-Mar-78 2102 MRC
To: JMC, LES
∂28-Mar-78 2057 MOON at MIT-MC (David A. Moon) [SAIL] HSTHST.PRO[DLN,MRC]
Seems simple and straight-forward enough. It probably even should work.
mrc - Nice to get a comment like that!
∂28-Mar-78 2128 ZM
I left in your mailbox all the data you needed. I hope it will be of some help.
∂28-Mar-78 2159 KRD sigma chi
Can I interest you in seconding my nomination to sigma xi? (Lederberg
has agreed to nominate me). The form requires only a brief paragraph
evaluating research contributions and potential.
Randy
If you write the paragraph not too outlandishly, I'll be glad to second
the nomination.
∂28-Mar-78 2224 KRD
I shall be a master of restraint..
∂29-Mar-78 1100 JMC*
Call Pat D., limb, nsf, colgate
∂30-Mar-78 0804 BS Zohar Manna
I have checked every possible source--CS evaluations (not begun until
1972-73), department files, etc., and there are no teaching evaluations for
Zohar except for random comments in reference letters.
Betty
∂30-Mar-78 1100 JMC*
bionic.ns
∂31-Mar-78 0916 RCM via SRI-KL
To: JMC, RWW
Now that my thesis is essentially done, I expect to be spending much less
time at SAIL. Any messages for me should be sent to BMOORE%SRI-KL.
Bob
∂31-Mar-78 1115 LCW S-1 Report
To: LES, TED, JMC, JBR, DWW, GIO, BTH, ME
CC: LCW
We now have available the new double-sided bound reports with color pictures.
If you need one, send me a note.
Curt
∂31-Mar-78 1330 Sacerdoti at SRI-KL (Earl Sacerdoti) Re: letter about Manna
Date: 31 Mar 1978 1328-PST
From: Sacerdoti at SRI-KL (Earl Sacerdoti)
Subject: Re: letter about Manna
To: JMC at SU-AI
cc: SACERDOTI
In response to your message sent 30 Mar 1978 1401-PST
John,
Here's my personal impression of Zohar's teaching; it was easy to
write since he's among the best teachers I encountered either at Stanford
or at Yale. I'm sending it via the ARPAnet as you indicated speed was
important; I'll be glad to have it typed up if you'd prefer to wait.
--Earl
I attended CS156, taught by Professor Manna, in the fall of 1971.
His teaching can be described in a word: formidable. He is efficient,
enthusiastic, and supremely well organized. He has a talent for balancing
the conflict between responding to students' requests for clarification
and covering the material. He has high expectations of his students, and
he seems successful at getting more than his share from them.
Of all the classes I attended at Stanford, I would rank Professor
Manna's as the clear best.
Professor Manna's teaching ability has, in an odd way, had a major
impact on my career. In the fall of 1969 I was visiting a friend who was
a graduate student in the Computer Science Department, and I went along
with him to a class taught by Professor Manna. I felt that if that was
the way computer science was taught at Stanford, then Stanford was where I
wanted to do my graduate work. In my opinion, adding Professor Manna to
the tenured faculty will help to ensure that the teaching quality in
computer science remains high.
-------
∂31-Mar-78 1605 Creary at SUMEX-AIM Any Developments?
Date: 31 Mar 1978 1605-PST
From: Creary at SUMEX-AIM
Subject: Any Developments?
To: JMC at SU-AI
Have there been any developments concerning the funding of your grant
proposal that would permit us to resume our discussion of my working for you
as a research associate, investigating epistemological problems of AI? If not,
can you tell me when the relevant information is likely to become available?
I am appending to this message a version of my resume that has no
underlining or pagemarks; I'll be happy to send you a printed version if that
would be useful to you. In any case, I'll send you a copy of a review that I
wrote (with a colleague) concerning a book on counterfactual conditionals.
-- Lew Creary
************
LEWIS G. CREARY
RESUME
March, 1978
PERSONAL
Born on May 8, 1937, in Detroit, Michigan.
Married on June 15, 1963; one daughter, born on January 26, 1974.
EDUCATION AND FELLOWSHIPS
General Motors Institute (Flint, Michigan) 1955-1960; received the
degree of Bachelor of Mechanical Engineering.
Wayne State University (Detroit, Michigan) 1960-1963; studied
mathematics and philosophy; received the degree of Master of Arts
(in philosophy).
Princeton University (Princeton, New Jersey) 1963-1966; Department
of Philosophy: Program in the History and Philosophy of Science;
Bamford Fellow, 1963-1964; Council of the Humanities Fellow, 1964-
1965 and 1965-1966; received the Ph.D. degree in June, 1969.
ENGINEERING EXPERIENCE
Received a total of three years of full-time work experience as a
junior metallurgical engineer at General Motors Research
Laboratories (Warren, Michigan), in conjunction with the education
at General Motors Institute.
ACADEMIC EXPERIENCE
Lecturer in Philosophy at the University of Maryland (College
Park, Maryland) 1966-1967. Courses taught: Elementary logic and
Semantics, Symbolic Logic, Philosophy of Science.
Assistant Professor of Philosophy at Case Western Reserve
University (Cleveland, Ohio) 1967-1976. Courses taught:
Introduction to Logic, Epistemology, Introduction to Philosophy,
Philosophy of Language, Mathematical Logic, Seminar in Inductive
Logic, Probability and Induction, Philosophy of Mathematics
(Graduate Tutorials), Philosophy of Science, Seminar in
Epistemology, Current Topics in Philosophy: Philosophy of Logic,
Seminar in Philosophy of Language: Foundations of Linguistic
Theory, Foundations of Cognitive Psychology (Graduate Tutorial).
1
LEWIS G. CREARY Page 2
COMPUTER EXPERIENCE
Research Associate (half-time), Wayne State University Computing
Center (Detroit, Michigan) 1961-1963. Worked mainly on two
projects:
1) Debugging and system subroutine writing for the list
processing language IPL-V, as implemented on the IBM 650. This
work led to my being designated as the person responsible for
maintenance of the 650 IPL-V system (see Communications of the
ACM, Vol. 6 (March 1963), 86-87).
2) Design and implementation of programs for the IBM 7070 and
IBM 1401 to compute, summarize, and report measurements of air
pollution made by a group of eight different atmospheric analyzing
instruments.
Research Associate in Computer Science, Heuristic Programming
Project, Stanford University (Stanford, California) 1976-present.
Have worked mainly on two projects:
1) Design and implementation of new capabilities for Meta-
DENDRAL, a system of applied artificial intelligence programs for
automatic inductive inference (from appropriate data) of laws
governing the fragmentation of organic molecules in a mass
spectrometer.
2) Assisting in the design of a MYCIN-like automated consultant
for users of a large general purpose finite-element structural
analysis program.
MAIN PHILOSOPHICAL INTERESTS
Epistemology, semantics and logic, probability and induction,
information-processing theories of cognition, philosophy of
science, philosophy of mathematics.
MAIN INTERESTS IN ARTIFICIAL INTELLIGENCE
Representation and maintenance of knowledge; development of
knowledge-based systems for intelligent inference, deliberation,
problem-solving, and decision-making; computer processing of
natural language at the semantic level; and the relation of work
in these areas to problems and results in information-processing
psychology.
DOCTORAL DISSERTATION
The Pragmatic Justification of Induction: A Critical Examination,
(1969). A detailed critique of Hans Reichenbach's pragmatic
treatment of the problem of induction.
Advisor: Professor Carl G. Hempel.
LEWIS G. CREARY Page 3
PUBLICATIONS
"Empiricism and Rationality," Synthese, Vol. 23 (1971), Nos. 2/3,
234-265.
"For the Compleat Logical Empiricist: 'Non-Cognitive' Foundations
for Inductive Logic," American Philosophical Quarterly, Vol. 10
(1973), No. 2, 123-131.
Review and critical discussion of David Lewis's book
Counterfactuals (Cambridge, Mass.: Harvard University Press,
1973), Philosophy of Science, Vol. XLIII, No. 3 (September 1975),
341-344. (Jointly authored with Christopher Hill.)
ADDRESSES
"The Concept of Semantic Information," to an interdisciplinary
seminar on "The Notion of Information in Different Fields," School
of Library Science, Case Western Reserve University, January 1969.
"On Frege's Conception of Meaning," Cleveland Linguistics Society,
May 1970.
"Chomsky, Language, and Rational Man," in the series "Contemporary
Views of Humanity," Center for Continuing Education, Case Western
Reserve University, October 1972.
PROFESSIONAL ACTIVITIES
Member, American Philosophical Association, and Special Interest
Group on Artificial Intelligence of the Association for Computing
Machinery.
Commentator on R. Giere's paper "The Epistemological Foundations
of Objective Statistical Methods," American Philosophical
Association, Western Division, May 1970, St. Louis.
Referee for the I.B.M. Journal of Research and Development, Fall
1975.
-------
∂02-Apr-78 1107 ZM
I have sent the body of my Talcott letter to Pat. Zohar
∂03-Apr-78 0016 LES Bulnes
Just remembered a loose string: you didn't say what should be done about
Bulnes. When last we talked, there was some uncertainty, based on RWW's
input. Should we offer regular support for the Spring Quarter?
For Spring, yes.
∂03-Apr-78 0032 LES
∂01-Apr-78 1225 JMC EFIND
It occurs to me that EFIND would be most effective if merged into FIND.
This oould be done by having FIND look at the directory and if the first
line of the directory was AAAAA, assume the file was in EFIND format
and act accordingly. Then our users wouldn't have to be persuaded
to change their habits in using FIND - only in creating new files,
which they must do anyway.
Unfortunately, this scheme doesn't work in general: the existance of
a directory page and an alphabetized file does NOT insure that it
is alphabetized by the field you are searching for. For example,
if you do a "FIND JMC", it will fail (i.e. it will look on the "J"
page and not find you there).
∂03-Apr-78 0910 Levin at SUMEX-AIM tour of duty officially ended!
Date: 3 Apr 1978 0910-PST
From: Levin at SUMEX-AIM
Subject: tour of duty officially ended!
To: jmc at SAIL
well john.....less paperwork to deal with...congratulations!!!!
hope you'll come around and say hi every now and then....!!!
...laurie
-------
∂03-Apr-78 1017 DAVESMITH at PARC-MAXC VEL (LISP70)
Date: 3 Apr 1978 10:05 am (Monday)
From: DAVESMITH at PARC-MAXC
Subject: VEL (LISP70)
To: Barrow @ SRI-KL, McCarthy @ SU-AI, Earnest @ SU-AI
cc: DAVESMITH
Barrow @ SRI-KL, McCarthy @ SU-AI, Earnest @ SU-AI
It is with extreme regret that I am forced to announce that I will be unable to
finish the VEL (formerly known as LISP70) project by myself. I simply do not
have the time. I am involved in a major project this year at Xerox, and I feel I
must give my job here top priority. It has even been requiring a few weekends
lately. In addition we have a new daughter (Erin, 5 mos now) with whom I
want to spend some time, and also a new house which needs some fixing up.
I am unwilling to simply kill the VEL project (VEL stands for Very Extensible
Language) because:
(1) I think it has a great deal to contribute to AI programming in general and
list-manipulation and LISP in particular. I believe the language
represents a quantum jump in representational clarity for symbolic
computations. It also combines for the first time efficient number
crunching with LISP, since it has data types.
(2) I have spent three years working on it, and it is in a state where 6-9
months of full-time work would complete it. Had Colby not left to go to
UCLA, it would now be available for the AI community.
My current feeling is that I will be unable to finish the project unless I am
given help by someone. The help would have to be in the form of one or two
competent full-time programmers. I would act as consultant and provide
documentation, but not do any programming. If anyone feels that this is worth
spending time on to complete, please let me know. I will be available in the
above capacity for the forseeable future. In the meantime, if I don't hear from
anyone within the month, I plan to archive the existing source files on a tape
and remove them from the Stanford system.
John and Les, thank you sincerely for the support you have provided in the
form of computer time and disk resources up to now. You have been most
generous.
-- Dave
∂03-Apr-78 1123 Tajnai at SUMEX-AIM Faculty Meeting 4/4 2:15
Date: 3 Apr 1978 1122-PST
From: Tajnai at SUMEX-AIM
Subject: Faculty Meeting 4/4 2:15
To: faculty.list:
cc: Tajnai
CSD Faculty Meeting Tuesday, April 4, 2:15 p.m., Polya 204
Agenda:
1. Degree Candidates
2. Admissions Results
3. Curriculum Results
4. Non Academic Council Members as Principal Dissertation Advisors
5. Comprehensive Examination
6. Definition of normal progress for Ph.D. students with previous
graduate education
-------
∂03-Apr-78 1715 DCL
To: GROUP.DIS[VCG,DCL]:
***********************************************************************
VERIFICATION GROUP SYSTEM MEETING TUESDAY 11TH APRIL
PLACE: ERL 237
TIME: 2:30 pm
TITLE: ORGANIZATIONAL MEETING
SPEAKER: To be announced
Please note: No seminar tomorrow, Tuesday April 4th.
∂04-Apr-78 1731 BCM appointment
Can i make an appointment with you w. r. t. advising me on PhD goodies?
spasibo
Ben Moszkowski
P. S. I realize you are currently on a trip. Have fun!!
Sure, after April 15. One commetn on your term paper in
which you didn't actually sucfceed in making progress in
formalization. I think "IMYA" WAS the wrong word
in the title. It should have been "NAZVANIYE". The
former means only first name of a person.
∂04-Apr-78 1806 RSMITH at RUTGERS-10 Group at the Center concerning AI and Philosophy
Date: 4 Apr 1978 (Tuesday) 2109-EST
From: RSMITH at RUTGERS-10
Subject: Group at the Center concerning AI and Philosophy
To: MCCARTHY at SU-AI
You had mentioned some time ago that you were organizing
a group at the Center on the relation between AI and philosophy.
I was wondering if there might be a chance for me to participate
in that group, perhaps for one semester. I have made some
inquiries and it appears it is not impossible that I might
be able to get away.
If you would be interested, I would like to explore further
how I might fit into the group, and some of the possible
research topics to which I might be able to make a contribution.
Regards,
Bob Smith
rember Bob Smith.
∂05-Apr-78 1311 Raphael at SRI-KL (Bert Raphael) Re: Grigory Goldshtein
Date: 5 Apr 1978 1310-PST
From: Raphael at SRI-KL (Bert Raphael)
Subject: Re: Grigory Goldshtein
To: Hart, Sacerdoti, Bolles, McCarthy at SU-AI, PHW at MIT-AI
cc: RAPHAEL
In response to the message sent 23 Mar 1978 1417-PST from Raphael
Grigory Goldshtein's appeal is scheduled for TOMORROW, April 6. If you
want to help, send a Western Union "night letter" right away. The most
important recipient is
Mr. A. E. Batidze, Procurator
Lenin District
Tbilisi, Georgia
USSR
(This apparently is a better spelling of the prosecutor's name than the one
I sent in the previous message. Also, please note that the Bay Area Council
address is 106 Baden St., not Bazen St., San Francisco. Thanks.)
Bert
-------
∂05-Apr-78 1721 Davis at SUMEX-AIM your encyclopedia article
Date: 5 Apr 1978 1721-PST
From: Davis at SUMEX-AIM
Subject: your encyclopedia article
To: jmc at SAIL
I think the organization and content are very good. It's hard to provide
a coherent overview in so few pages, but I think you've done it nicely.
I have just a few comments on possible additional content you might
consider.
I have extensive comments regarding style, tho. In general I think you
indulged too much jargon and wrote text that requires more background
than the average encyclopedia reader will have. The result is that some
very good ideas are rendered cryptic.
The easiest way to give you the comments would be for me to edit a
copy of your source file (I could type in all uppercase to make clear
what I suggest changing), but that would take on the order of
an hour or two (my comments re-write almost the entire thing).
I'm wary of doing so before finding out whether the result will actually
be used. That is, will you really have a chance to get those changes
(the ones you agree to, anyway) into the article, or is it too late 'cause
the editors already have their hands on it? When you know for sure
whether the comments might be useful, let me know and we can decide
what to do from there.
Randy
-------
Thanks much for the comments. The file is ARTIFI[W78,JMC].
Ask Patte to telephone to determine whether revisions are
possible.
∂06-Apr-78 0828 MRC "Dialnet box"
To: TED, LES, JMC
I think that the line transmission protocol (including packet framing and
checksum computation) in Dialnet can be reasonably implemented by hardware
instead of software and should be. What is required is to search for a
DLE followed by STX, a 5-byte packet header, a 0 to 255 byte packet data
area (the last word of the header says how large it is), a two byte packet
checksum, and DLE followed by ETX. DLE followed by DLE means a single DLE
to be passed as part of the packet, anything after DLE other than DLE, STX,
and ETX is illegal (I haven't decided what to do about that case yet).
DLE, STX, and ETX all have the 200 bit set; 220, not 020, is DLE.
In the event of a framing error it just discards the packet it is accumulating
with no error CONI bit. A checksum error probably should set some random CONI
bit. Data should be transfered as 8-bit bytes, with each byte being stored in
a PDP-10 9-bit quarterword; this makes my debugging MUCH easier than storing
as 8-bit bytes with four free bits at the end. Of course, it only interrupts
for an entire packet.
How feasible do you think this will be? Les, do you think this IAZ person is
a hardware type who might be interested in designing the thing? I'm convinced
this is much better than implementing it in software on a PDP-10 or on a front
end PDP-11.
-- Mark
∂06-Apr-78 0901 Hart at SRI-KL (Peter Hart) Panel discussion
Date: 6 Apr 1978 0900-PST
From: Hart at SRI-KL (Peter Hart)
Subject: Panel discussion
To: Feigenbaum at SUMEX-AIM, JMC at SAIL, Bobrow at XEROX-PARC,
To: TENENBAUM
cc: hart
I would like to invite you to be on a panel to discuss "The Future of AI."
The panel discussion will be held at Stanford on Thursday, May 11, at
3:45 (room to be arranged) as the last of a mini-series of one hour
seminars on AI.
Although one hour is obviously insufficient to examine issues in detail,
I would hope that the panel could address the following types of questions:
1. What are the most significant technical problems facing AI?
2. What progress is likely to be made on these problems in the
next 10 years?
3. What effect would such progress have on the world outside
AI (practical systems, perhaps societal implications, etc.)
4. What, if anything, could be done in an institutional or
political vein to speed up progress beyond its present rate
("more of the same," or is something more imaginable possible)?
I believe that an occasional panel of this type is valuable to get
graduate students thinking about longer-term issues, and I hope that you
will be able to participate. Please let me know by Tuesday, April 11,
if you can attend.
Cheers,
Peter
-------
∂06-Apr-78 0916 Feigenbaum at SUMEX-AIM Re: Panel discussion
Date: 6 Apr 1978 0908-PST
From: Feigenbaum at SUMEX-AIM
Subject: Re: Panel discussion
To: Hart at SRI-KL, JMC at SAIL, Bobrow at XEROX-PARC,
To: TENENBAUM at SRI-KL
In response to the message sent 6 Apr 1978 0900-PST from Hart at SRI-KL (Peter Hart)
I aCCept panel invitation. looking forward to it.
Ed
-------
∂06-Apr-78 1232 RPG New Maclsp
To: MACLSP.DIS[AID,RPG]:
Warning! LISP 1525 has been installed, the first new assembly
since September 1977. At the next opportunity you should re-make any
systems based on Maclisp as well as reassemble any LAP code which
does file IO. Also, you should READ LSPARC (the first 3-4 pages) to
learn of the changes in this version. Good luck.
-rpg-
∂06-Apr-78 1307 bobrow at PARC-MAXC Re: Panel discussion
Date: 6 Apr 1978 12:59 pm (Thursday)
From: bobrow at PARC-MAXC
Subject: Re: Panel discussion
In-reply-to: Your message of 6 Apr 1978 0901-PST.
To: Hart at SRI-KL (Peter Hart)
cc: Feigenbaum at SUMEX-AIM, JMC at SAIL,
TENENBAUM at SRI-KL,
I accept. See you at the panel.
danny
∂07-Apr-78 1018 RPG *rset
To: MACLSP.DIS[AID,RPG]:
Remember, the default for *rset in MacLisp 1525 is t, not nil as
in all previous versions. Also, LISP.INI files, if present, are automatically
loaded.
-rpg-
∂07-Apr-78 1044 CCG ibm fellowship nomination
To: JMC, eaf at SUMEX-AIM, tw at PARC-MAXC, buchanan at SUMEX-AIM, TOB
To: DCL, ZM
There is yet another ibm fellowhip available from IBM San Jose. I talked
with Pat Mantey to see what they were looking for. They want to award
it to someone in AI,probably. Suggestions included natural language,
practical program correctness, and high level progam specifications
aids. Robotics was out, and anything close to the work in DSL will
be supported by another fellowship given thru dsl. I would like to
see it go to needy projects. Send me the names of your favorite
nominees by tuesday. I will select according to fellowship
distribution patterns, need, and likelihood to win the fellowship.
We will then prepare recommendations.
Cordell
An immediate suggestion is Dave Wilkins. Another is Ben Moszkowitz,
but probably he still has an NSF.
∂07-Apr-78 1330 STT
stt/cc
∂07-Apr-78 1330 STT CS224 schedule changes
To: JMC
CC: STT
If it is all right with you, we would like to change the date of your lecture
on formal methods for representing knowledge from May 18 to May 9 (Tuesday.)
We'd like to have the history of AI panel on the 18th. Thus:
May 9 Formal Methods for Representing Knowledge
May 18 Panel on the History of AI
May 23 Panel on Representation
Please let me know if these dates are OK with you. And do you feel these three
are too many?
Note: CS244 meets TuTh 11:00-12:15 in 111 Polya. There are about 50 students.
∂07-Apr-78 1509 LES
.onecol; cb Budget
.begin "budget" verbatim select 5;
PERIOD COVERED: 3 Years: 1 June 1978 through 31 December 1981.
Dates: 6/1/78-5/31/79 6/1/79-5/31/80 6/1/80-5/31/81
Person- Person- Person-
A. SALARIES AND WAGES months months months
1. Senior Personnel:
a. John McCarthy, 24,257. 6.5 27,007. 6.5 29,168. 6.5
Professor
Summer 75%(2 mos.)
Acad. Yr. 50%
2. Other Personnel
a. Student Research
Assistants
(50% Acad.Yr.;
100% Summer)
(1) 7,155. 7.5 7,704. 7.5 8,320. 7.5
(2) 7,155. 7.5 7,704. 7.5 8,320. 7.5
b. Support Personnel
(1) Sec'y (20%) 2,092. 2.4 2,259. 2.4 2,440. 2.4
(2) Elec.Tech.(15%) 2,937. 1.8 3,172. 1.8 3,426. 1.8
_______ _______ _______
Total Salaries & Wages 43,596. 47,846. 51,674.
B. STAFF BENEFITS
9/1/77-8/31/78:19.0%
9/1/78-8/31/79:20.3%
9/1/79-8/31/80;21.6%
9/1/80-8/31/81;22.4%
8,708. 10,179. 11,472.
_______ ________ ________
C. TOTAL SALARIES, WAGES,
AND STAFF BENEFITS 52,304. 58,025. 63,146.
.next page
D. PERMANENT EQUIPMENT 5,000 -- --
(2 Datamedia terminals)
E. EXPENDABLE SUPPLIES 1,632. 1,730. 1,834.
& EQUIPMENT(e.g.,copying,
office supplies, postage,
freight, consulting,
honoraria)
F. TRAVEL 1,840. 1,950. 2,067.
All Domestic Travel
G. PUBLICATIONS 1,000. 1,060. 1,124.
H. OTHER COSTS 6,640. 7,038. 7,460.
1.Communication 1,600.
(telephone)
2. Computer 5,040.
Equip. Maint.
_______ ________ _______
I. TOTAL COSTS 68,416. 69,803. 75,631.
(A through H)
J. INDIRECT COSTS:58% of 36,781. 40,486. 43,866.
A through H, less D. ________ ________ ________
K. TOTAL COSTS 105,197. 110,289. 119,497.
L. THREE YEAR TOTAL 334,983.
.end "budget"
If it is just as suitable for the lab, putting system programmer instead
of electronic technician might work better with NSF. The final total
still might be more than they will go for, but you might telephone
Eamon Barret at NSF and see what he thinks of this budget.
∂07-Apr-78 1624 RPG Lisp 1525
To: MACLSP.DIS[AID,RPG]:
Compiled hunk code appears to lose in MacLisp 1525.
If this will effect you, use sys:maclsp.old.
-rpg-
∂07-Apr-78 1809 RPG Hunk bug in MacLisp 1525
To: MACLSP.DIS[AID,RPG]:
Re-compile all files which use hunks.
-rpg-
∂09-Apr-78 1657 BAK at MIT-AI (William A. Kornfeld)
Date: 9 APR 1978 1956-EST
From: BAK at MIT-AI (William A. Kornfeld)
To: feigenbaum at SU-AI, jmc at SU-AI, nilsson at SU-AI
I am interested in collecting examples of the following sorts of phenomena
for my thesis research.
1. Where there are several methods that can be applied to solve a certain
subproblem. Some of these methods may diverge or require intractable
amounts of computation time on certain examples and there is no way to tell
a priori which method will be the right one for a particular example.
2. Where a computation is involved in a losing depth first search and the
control structure makes it difficult to recover and try another method.
I am making a study of such problems which I believe occur quite commonly in
AI programming. If you have any favorite examples of this from your
research I would appreciate hearing about them. Distribution of this note
to others who might be able to supply input is greatly appreciated.
Thank you,
William Kornfeld (BAK@MIT-AI)
∂10-Apr-78 0940 CET
To: JMC, TOB, HPM, DPB
Memo to Professors McCarthy and Binford, Hans Moravec
copy to Denny Brown
Subject: Conditional on Qualifying Exam.
Ref: Letter to Mr. Moravec from Prof. Floyd daed 7/1/75
"This letter is to inform you that your AI Qual.Exam. committee has been
unable to determine a grade for you based on your oral presentation on
June 10, 1975.
It is the recommendation of this committee that you continue this exam
during the summer quarter 1974-75. Specifically, the committee found that
you must prepare a survey of the "state of the science" with broad coverage,
but also with specific emphasis on perception research.
It is up to you to contact the memers of your committee to set up a new time
and place. They ae: Prof. Feigenbaum, L. Quam and R. Duda or T. Binford."
Hans said he has completed this work. I need a statement in his folder
confirming this. Please advise.
∂10-Apr-78 1117 JBR
You have exceeded your disk quota.
The files listed below have been purged to reduce your disk
area to your quota of 2000
Before purging, your files occupied 3010
You have too many files. The purger may not select the
optimum set.
PPSAV.TMP[ESS,JMC]
AIQUAL.LST[W77,JMC]
DIALNE.LST[DIA,JMC]
CRYPT.DMP[ 2,JMC]
CODE.DMP[ 2,JMC]
2P.LAP[W77,JMC]
FIRST.XGP[W77,JMC]
IJCAI.XGP[E77,JMC]
HOTER.XGP[ESS,JMC]
NAUR.XGP[LET,JMC]
BERNFI.XGP[LET,JMC]
LIBURD.XGP[LET,JMC]
COMMEN.XGP[LET,JMC]
MANNA.XGP[LET,JMC]
LISP.XGP[F77,JMC]
MOORE.XGP[LET,JMC]
TALCOT.XGP[LET,JMC]
CONCEP.XGP[ESS,JMC]
MENTAL.XGP[F76,JMC]
MINIMA.XGP[ESS,JMC]
ESTRIN.XGP[LET,JMC]
MC[245,JMC]
PATH[245,JMC]
PATH2[245,JMC]
REFER.ENC[225,JMC]
PATHS.RLS[225,JMC]
GRPALG.RLS[225,JMC]
GRPDAT.RLS[225,JMC]
GROPER.RLS[225,JMC]
DIMEN.RLS[225,JMC]
CHAR.RLS[225,JMC]
DEG5.IN[225,JMC]
S4.REP[225,JMC]
REPRES.RLS[225,JMC]
DEG6.IN[225,JMC]
R42.IN[225,JMC]
ANNOUN[225,JMC]
PUZZ.SAI[225,JMC]
PUZZA.SAI[225,JMC]
PANIC.SOS[225,JMC]
PUZZB.SAI[225,JMC]
ENCYC1.PRO[ESS,JMC]
PUZZE.SAI[225,JMC]
PTCH2.DIR[ AI,JMC]
DRIVE.DIR[ AI,JMC]
MINE.DIR[ AI,JMC]
KNOW2.AI[ESS,JMC]
ARTNA1.ART[ESS,JMC]
MEMMTC.QUA[ESS,JMC]
LANG1.AI[ESS,JMC]
S5.QUA[ESS,JMC]
MTC71.QUA[ESS,JMC]
R1303.ART[ESS,JMC]
R1301.ART[ESS,JMC]
QA.PRO[ESS,JMC]
ARTNAT.ART[ESS,JMC]
SEMAN.AI[ESS,JMC]
FRANK.STA[ESS,JMC]
MTCSYL.QUA[ESS,JMC]
COMPUT.STA[ESS,JMC]
AISY69.QUA[ESS,JMC]
FILEC.FAI[ESS,JMC]
LETTVI.REV[ESS,JMC]
R1300.ART[ESS,JMC]
LCFMEM.MTC[ESS,JMC]
PEARL.NOT[ESS,JMC]
NETJO2.PRO[ESS,JMC]
IPT73.REP[ESS,JMC]
NSFX.5[ESS,JMC]
NSFX.7[ESS,JMC]
NSFX.9[ESS,JMC]
NSFX.2[ESS,JMC]
NSFX.6[ESS,JMC]
NSFX.8[ESS,JMC]
NSFX.4[ESS,JMC]
NSFX.3[ESS,JMC]
HAYES.FRM[ESS,JMC]
ADTEL2.NOT[ESS,JMC]
PLATT.NOT[ESS,JMC]
TASK2.MAR[ESS,JMC]
TASK1.MAR[ESS,JMC]
LETTER.HE2[ESS,JMC]
XLET[ESS,JMC]
ZF.AX[ESS,JMC]
FOO73A[ESS,JMC]
FOON.PR1[ESS,JMC]
SWOPSI.RE[ESS,JMC]
HART.ME1[ESS,JMC]
GGGG[ESS,JMC]
PRO[ESS,JMC]
M29AUG.MES[ESS,JMC]
SYMPOS[ESS,JMC]
CH1.BH2[ESS,JMC]
CLOCK.FAI[ESS,JMC]
SLIDES[ESS,JMC]
FOO1[ESS,JMC]
WUTHER.QUE[ESS,JMC]
ROY[LET,JMC]
TOLLES.LET[LET,JMC]
FORDH.LET[LET,JMC]
AD.NET[ESS,JMC]
KNUTH.SAI[225,JMC]
TUCKER.LET[LET,JMC]
TUCKER.LE2[LET,JMC]
TOLLES.LE1[LET,JMC]
TCLUB.MEM[ESS,JMC]
MINE[ AI,JMC]
NOTES2[ AI,JMC]
PUTCH3[ AI,JMC]
RUSS[ AI,JMC]
CHYLIN.LE1[ AI,JMC]
GEN1[ AI,JMC]
DRIVE3[ AI,JMC]
MPRO[ AI,JMC]
PUTCH[ AI,JMC]
PUTCH.DIR[ AI,JMC]
DRIVE[ AI,JMC]
AIBIB[ AI,JMC]
SLAMEC.LE2[LET,JMC]
BOWDEN.LE1[LET,JMC]
CHERN.LE2[LET,JMC]
MAY.ME[LET,JMC]
NSFX.10[ESS,JMC]
HENRY[ESS,JMC]
MC.AI[ESS,JMC]
BIRKHO.LE1[LET,JMC]
FLANAG.LE1[LET,JMC]
ERIK.LE6[LET,JMC]
HARPER.LE1[LET,JMC]
PUZZE.F4[225,JMC]
ERIK.LE7[LET,JMC]
BAJCSY.LE1[LET,JMC]
STARKM.LE1[LET,JMC]
BRIT.LET[LET,JMC]
IRANI.LE1[LET,JMC]
BOWEN.LE1[LET,JMC]
JAF.STA[ESS,JMC]
RECUR[ESS,JMC]
COOP.LE1[LET,JMC]
POE.LE1[LET,JMC]
JMILLE.LE2[LET,JMC]
IGARAS.LE1[LET,JMC]
ITO.LE1[LET,JMC]
CLARK.LE1[LET,JMC]
FLESS.LE2[LET,JMC]
TAVARE.LE1[LET,JMC]
SCIENC.LE1[LET,JMC]
SCRIB.LE1[LET,JMC]
ATKIN.LE1[LET,JMC]
KOCHEN.LE1[LET,JMC]
GREER.FAC[LET,JMC]
CU.LE1[LET,JMC]
RALSTO.LE2[LET,JMC]
MESSAG.SCI[LET,JMC]
MCCRAC.LE2[LET,JMC]
RALSTO.LE3[LET,JMC]
CHRON.LE1[LET,JMC]
LOGIC.QUA[LET,JMC]
SHIGER.LE1[LET,JMC]
PAMELA.LE1[LET,JMC]
TAUGNE.LE1[LET,JMC]
MORGAN.LE2[LET,JMC]
FROOK.LE1[LET,JMC]
LOMONA.LE1[LET,JMC]
APR.ME[LET,JMC]
LOGIC.PUB[LET,JMC]
DAVID.LE1[LET,JMC]
SCHOEL.LE1[LET,JMC]
CALHOU.LE1[LET,JMC]
FINCH.LE1[LET,JMC]
LYKOS.LE1[LET,JMC]
JMILLE.LE1[LET,JMC]
MITCH.LE1[LET,JMC]
JERRY.LE1[LET,JMC]
MCQUIL.LE1[LET,JMC]
PAM.LE1[LET,JMC]
DRLUK.LE1[LET,JMC]
OCT.ME[LET,JMC]
OLSEN.LE1[LET,JMC]
ERIK.LE0[LET,JMC]
SUBCOM.LE1[LET,JMC]
NEWSCI.LE1[LET,JMC]
AHLHAU.LE1[LET,JMC]
MEMOS.CUR[LET,JMC]
PUBINT.LE1[LET,JMC]
HORNIN.LE1[LET,JMC]
FELDST.LE1[LET,JMC]
CHECK.DEP[LET,JMC]
TUCKER.LE1[LET,JMC]
SCANDU.LE1[LET,JMC]
MELTZ.LE1[LET,JMC]
PIASET.LE1[LET,JMC]
MILNER.REC[LET,JMC]
FLESS.LE1[LET,JMC]
TRESSL.LE1[LET,JMC]
PIASET.LE2[LET,JMC]
LICKL.LE1[LET,JMC]
FORSHA.LE1[LET,JMC]
COMPT.LE1[LET,JMC]
STORK.LE1[LET,JMC]
R1302.ART[ESS,JMC]
ASTRO2.ART[ESS,JMC]
FRIEDM.LE1[LET,JMC]
FF[ESS,JMC]
AIRLIN.MTC[ESS,JMC]
JFOL.USE[ESS,JMC]
MEARS.LET[LET,JMC]
GRIMM.BKP[ESS,JMC]
SCIAM.LE1[LET,JMC]
NYT.LE1[LET,JMC]
NAVROZ.LE1[LET,JMC]
NOTES1[ AI,JMC]
STARK.LE1[LET,JMC]
MDAVIS.LE1[LET,JMC]
IGARAS.LE2[LET,JMC]
REPRES[LET,JMC]
GLUGG[LIT,JMC]
GLUGG2[LIT,JMC]
FLESS.LE0[LET,JMC]
KNOW.ART[ AI,JMC]
KNOW.NOT[ AI,JMC]
KILPAT.LE1[LET,JMC]
MEIR.LE1[LET,JMC]
MCCRAC.LE1[LET,JMC]
NAS.NOT[ESS,JMC]
JAF.LE2[LET,JMC]
TELLER.LE1[LET,JMC]
CRYPT.FAI[ 2,JMC]
WEISSK.LE1[LET,JMC]
3D.NSF[ESS,JMC]
ONEILL.NOT[ESS,JMC]
AAAS.PRO[ESS,JMC]
ISRAEL.LE1[LET,JMC]
ISRAEL.ART[LET,JMC]
CHENEY.LE1[LET,JMC]
OCTOBE.OUT[LET,JMC]
CORBAT.LE1[LET,JMC]
COMTOP.LIS[ESS,JMC]
EFOO[ESS,JMC]
CALHOU.LE2[LET,JMC]
INFO.NEE[ESS,JMC]
EVANS.LE1[LET,JMC]
EVANS.LE2[LET,JMC]
NETDOC.PLN[ESS,JMC]
SUFFIC.FRE[LET,JMC]
OCT.USE[ESS,JMC]
OLIVER.LE1[LET,JMC]
INTRO.AI[ 2,JMC]
COMMON.NS[LET,JMC]
DRELL.LE1[LET,JMC]
SIGART.BUL[LET,JMC]
SIGART.LE1[LET,JMC]
SIGART.LE2[LET,JMC]
BOWEN.LE2[LET,JMC]
RUSSOF.NOT[ESS,JMC]
INVOIC[LET,JMC]
ACM.LE2[LET,JMC]
ACM.LE1[LET,JMC]
JOSHI.LE1[LET,JMC]
FERRIS.LE1[LET,JMC]
NYT.LE2[LET,JMC]
BDAY.LIT[LIT,JMC]
EXTFOR.MEM[258,JMC]
ACTOR0[258,JMC]
NONTER.PRB[258,JMC]
CORREC.ABS[258,JMC]
BACKUP.MCP[258,JMC]
HOMOMO.NOT[258,JMC]
BRACEW.LE1[LET,JMC]
MCP.PR1[258,JMC]
COND.PRF[258,JMC]
PQR.PRF[258,JMC]
SCHEM.WRU[258,JMC]
MCCUS.PRO[258,JMC]
FOLSCO.MEM[258,JMC]
FEMANO.LE1[LET,JMC]
MTC.GRP[ESS,JMC]
PETERS.LE1[LET,JMC]
JAN.ME[LET,JMC]
LEGUIN.CRI[LIT,JMC]
JAN14.DBX[258,JMC]
ESSEX.LE1[LET,JMC]
GELLMA.LE1[LET,JMC]
APPEND.CON[258,JMC]
FOLMAN.MOD[258,JMC]
HOBBS.LE1[LET,JMC]
BREMER.LE1[LET,JMC]
SHELDO.NS[LET,JMC]
AILET[LET,JMC]
CONSUL.POX[ESS,JMC]
REYNOL.LE1[LET,JMC]
ACM.LE3[LET,JMC]
LEAST.CM1[258,JMC]
NORBYE.LE1[LET,JMC]
JAN.OUT[LET,JMC]
DEC74.IN[LET,JMC]
HAND.2[258,JMC]
MTC.BLA[258,JMC]
JO.NOT[258,JMC]
LOGIC.QUA[258,JMC]
HAND3.W75[258,JMC]
HAND.W75[258,JMC]
MANNA.PR2[258,JMC]
MANNA.CM2[258,JMC]
HARDPR.OOF[258,JMC]
OKAMOT.LE1[LET,JMC]
MANNA.WRU[258,JMC]
DEMO.PR1[258,JMC]
VITASK[ 2,JMC]
CHRI.LE1[ 2,JMC]
VIS[ 2,JMC]
VIS1[ 2,JMC]
VIS.CAP[ 2,JMC]
NNN[ 2,JMC]
FRAGA.LE1[LET,JMC]
ENERGY.LE1[LET,JMC]
FIELDS.ME2[LET,JMC]
FEB.ME[LET,JMC]
DEC.ME[LET,JMC]
TASK[LIT,JMC]
ADHOC.THE[258,JMC]
SAKAI.LE1[LET,JMC]
LICK.ME1[LET,JMC]
LICK.LE1[LET,JMC]
LICK[ESS,JMC]
HAND.1[258,JMC]
BIBLE[245,JMC]
USER[245,JMC]
IDEAS[225,JMC]
SUZUKI.REC[LET,JMC]
MUFTIC.LE1[LET,JMC]
SAKAI.LE2[LET,JMC]
EQUAL.AX[258,JMC]
MCP.AX[258,JMC]
MCP2.AX[258,JMC]
MCP3.AX[258,JMC]
EXTFOR.AX[258,JMC]
INDUC.AX[258,JMC]
COMPIL.AX[258,JMC]
INTEGE.AX[258,JMC]
COND.AX[258,JMC]
ZF.AX[258,JMC]
INTEG2.AX[258,JMC]
ASHLEY.LE1[LET,JMC]
74DEC1.AJT[LET,JMC]
SMITH.LE1[LET,JMC]
GAREY.LE1[LET,JMC]
BERG.LE2[LET,JMC]
KOREA.POL[S75,JMC]
KOREA.2[S75,JMC]
RUSSIA.LE1[LET,JMC]
GELFAN.LE1[LET,JMC]
AUG75.OUT[MSG,JMC]
SEP75.OUT[MSG,JMC]
JAN75.OUT[MSG,JMC]
DEC74.OUT[MSG,JMC]
MAR75.OUT[MSG,JMC]
APR75.OUT[MSG,JMC]
MAY75.OUT[MSG,JMC]
JUN75.OUT[MSG,JMC]
JUL75.OUT[MSG,JMC]
MAR75.IN[MSG,JMC]
INGERM.LE1[LET,JMC]
APR75.IN[MSG,JMC]
MODEST[S75,JMC]
AICIRC.ABS[CUR,JMC]
CPDUST[CUR,JMC]
IDEOLO.ART[CUR,JMC]
APHOR.AI[CUR,JMC]
ECO.ESS[CUR,JMC]
RACKET.ESS[CUR,JMC]
PROPOS.BLA[CUR,JMC]
PHIL[CUR,JMC]
COMMON[CUR,JMC]
NEWS75.MOD[CUR,JMC]
EXEC.SUM[CUR,JMC]
PRODUC.ESS[CUR,JMC]
STANDA.ESS[CUR,JMC]
ARPMTC[CUR,JMC]
IDEOLO.ESS[CUR,JMC]
CHESS1[CUR,JMC]
SOCIAL.DEF[CUR,JMC]
RAMSEY[CUR,JMC]
HOW.ESS[CUR,JMC]
LEADER.ART[CUR,JMC]
LICK.PRE[CUR,JMC]
REPRES.LIC[CUR,JMC]
REPRES.LI2[CUR,JMC]
GOALS.LIC[CUR,JMC]
TWOMOD.ESS[CUR,JMC]
FORMAL.PRO[CUR,JMC]
SALARY.MEM[CUR,JMC]
SCIP[CUR,JMC]
IDEOLO[CUR,JMC]
ECON.NOT[CUR,JMC]
1974.IN[MSG,JMC]
JAN75.IN[MSG,JMC]
ARPA75.OUT[MSG,JMC]
FEB75.IN[MSG,JMC]
AUG75.IN[MSG,JMC]
JUL75.IN[MSG,JMC]
XGPPUB.1[ESS,JMC]
ADMIT.PUB[CUR,JMC]
KHABAR.AI[S75,JMC]
NEWS75.SUP[ESS,JMC]
NAKATA.LE1[LET,JMC]
SCHULT.LE1[LET,JMC]
REPRES.REP[ESS,JMC]
TWOEAS.MEM[258,JMC]
HOMER.LE1[LET,JMC]
REPRES.DOC[ESS,JMC]
LEM.LE1[LET,JMC]
REPRES.NOT[ESS,JMC]
HENDRI.REV[ESS,JMC]
OPTIM[F75,JMC]
LUMELS.LE1[LET,JMC]
DYSON.LE1[LET,JMC]
NEWBOR.REV[F75,JMC]
TASK.TA[F75,JMC]
NICK.FIL[F75,JMC]
SEXP[F75,JMC]
AI[F75,JMC]
LATIME.LE1[LET,JMC]
GENERA[F75,JMC]
LISP.MAN[S75,JMC]
SANDEW.LE2[LET,JMC]
PHIL.ART[F75,JMC]
DEFINE[F75,JMC]
BULLET[F75,JMC]
DATA3.204[F75,JMC]
DATA1.204[F75,JMC]
DATA4.204[F75,JMC]
ISHII.LE1[LET,JMC]
TEXTOR.LE1[LET,JMC]
EDH.DO[ESS,JMC]
ED1.DO[ESS,JMC]
OFFICE.PLN[F75,JMC]
ASHENH.LE1[LET,JMC]
ASHENH.LE2[LET,JMC]
IJCAI.DOC[F75,JMC]
IJCAI.ENQ[F75,JMC]
AAAS.76[ESS,JMC]
DCL2.75[LET,JMC]
LUK.LE1[ 2,JMC]
AAAS[ESS,JMC]
LUCKHA.LE1[LET,JMC]
ABELSO.LE1[LET,JMC]
SCIENC.LE2[LET,JMC]
NILSSO.ME1[LET,JMC]
HJERPP.LE1[LET,JMC]
MOSES.LE1[LET,JMC]
RELATI.POX[CUR,JMC]
BROWN.LE1[LET,JMC]
GILFIL.REV[F75,JMC]
SCOTT.LE1[LET,JMC]
SENSE.AD[LET,JMC]
PR1.DEC[F75,JMC]
AJT.MEM[ESS,JMC]
RAWLS.REV[ESS,JMC]
CYCLOP[F75,JMC]
GINI.LE1[LET,JMC]
SENSE.POS[F75,JMC]
MARTEL.DOC[LET,JMC]
MARTEL.LE1[LET,JMC]
FRIED.LE1[LET,JMC]
FRIED.LE2[LET,JMC]
ROTH.LE1[LET,JMC]
FIX[F75,JMC]
BRINKE.LE1[LET,JMC]
FIXUP.LBK[F75,JMC]
TABLE[F75,JMC]
TERMPA.206[F75,JMC]
FIXUP2.LBK[F75,JMC]
RYLE.REV[F75,JMC]
TUNNEY.LE1[LET,JMC]
FIXUP[F75,JMC]
BRINKL.LE1[LET,JMC]
FIXUP2[F75,JMC]
NEWS75.PUB[CUR,JMC]
MOTIV.ART[ AI,JMC]
GARDNE.LE2[LET,JMC]
SENSE.LE1[LET,JMC]
DAILY.ART[F75,JMC]
SENSE.LE2[LET,JMC]
REVAL3.LBK[F75,JMC]
REVAL2.LBK[F75,JMC]
AUTOMA[F75,JMC]
AUTOMA.2[F75,JMC]
COUNTE[F75,JMC]
PR1.AX[F75,JMC]
HOTER.1[CUR,JMC]
HOTER[CUR,JMC]
MELTZE.LE[LET,JMC]
NOV.ME[LET,JMC]
HEAD.LET[LET,JMC]
MACHIN[F75,JMC]
BETHE.LE1[LET,JMC]
SENSE.DAT[F75,JMC]
BETHE.LE2[LET,JMC]
GOLDMA.LE1[ 2,JMC]
FINITI[F75,JMC]
SOULE.LE1[LET,JMC]
THESIS[F75,JMC]
SCIENT.LE1[LET,JMC]
MINIMA[F75,JMC]
FIB[F75,JMC]
DEC75.IN[LET,JMC]
BLEDSO.DOC[LET,JMC]
RHODES.LE1[LET,JMC]
FIELDS.ME1[LET,JMC]
BLEDSO.LE1[LET,JMC]
NEWYOR.LE1[LET,JMC]
RAPHAE.LE1[LET,JMC]
GARDNE.LE1[LET,JMC]
FIELDS.SUP[LET,JMC]
KAHN.LE1[LET,JMC]
LISP.INT[F75,JMC]
BLOCKS.226[F75,JMC]
NOTES.226[F75,JMC]
TAUT.PRF[F75,JMC]
UTILI.AIL[F75,JMC]
MOSSIG.LE1[LET,JMC]
PET.LIT[LIT,JMC]
TOLEDO.LE1[LET,JMC]
CBCL[F75,JMC]
SET[F75,JMC]
SET.AX[F75,JMC]
SENSE.LE3[LET,JMC]
FILMAN.RE2[LET,JMC]
GIL1[ 2,JMC]
KNOPOF.LE1[LET,JMC]
YAMADA.LE1[LET,JMC]
CULIK.LE1[LET,JMC]
ANDREI.ADR[ESS,JMC]
JAN76.IN[LET,JMC]
MILLER.LE0[LET,JMC]
LUCKHA.BLA[F75,JMC]
JAN76.OUT[LET,JMC]
ARPA75.IN[MSG,JMC]
ADMIT.TEL[LET,JMC]
LICK.1[LET,JMC]
PHIL.ART[ESS,JMC]
WILSON.LE1[LET,JMC]
WILKIN.LE1[LET,JMC]
OSIS.LE1[LET,JMC]
GERROL.LE1[LET,JMC]
SFWA[S76,JMC]
CSDDIS.N25[ESS,JMC]
COMPUT.CSD[ESS,JMC]
BAT0.LOG[F75,JMC]
SELFCO.MEN[F75,JMC]
CHANGE.MEN[F75,JMC]
TECHNI.MEN[F75,JMC]
AUTOMA.MEN[F75,JMC]
MODEL.MEN[F75,JMC]
EPISTE.MEN[F75,JMC]
CELLUL.MEN[F75,JMC]
RELDEF.MEN[F75,JMC]
MOTIV.MEN[F75,JMC]
CONTEN.MEN[F75,JMC]
NOTE.MEN[F75,JMC]
POLEMI.MEN[F75,JMC]
NOTE1.MEN[F75,JMC]
AI.MEN[F75,JMC]
MOTIVA.MEN[F75,JMC]
CRYPT.CHK[ 2,JMC]
CRYPT.MEM[ 2,JMC]
COMMUN[S76,JMC]
LI[ESS,JMC]
TES[ESS,JMC]
ASSIGN[ESS,JMC]
CONTEN.DOC[ESS,JMC]
FP[ESS,JMC]
ENERGY.PRO[ESS,JMC]
RACRUL.MEM[ESS,JMC]
CODE.PRO[ESS,JMC]
RATIO.NOT[ESS,JMC]
RAY.AP[ESS,JMC]
CHESSP.DIS[ESS,JMC]
TASK[ESS,JMC]
REV[ESS,JMC]
REFER[ESS,JMC]
REPRES.BIB[ESS,JMC]
SLAGLE.REC[ESS,JMC]
FELDST.REV[ESS,JMC]
CHARAC.PRO[ESS,JMC]
ENERGY.FAC[ESS,JMC]
ED.DO[ESS,JMC]
MAGS.LIS[ESS,JMC]
GENER.AI[ESS,JMC]
NSFX.1[ESS,JMC]
TASK.FOO[ESS,JMC]
SUSAN.TAX[ESS,JMC]
GREET[ESS,JMC]
MUSIC.GRP[ESS,JMC]
EVENT.NOT[ESS,JMC]
GRK.TES[ESS,JMC]
POEM[ESS,JMC]
PSYCHO.AP[ESS,JMC]
NEWMEX.SPE[ESS,JMC]
TIME.HIS[ESS,JMC]
IGSCS.PLN[ESS,JMC]
ALL[ESS,JMC]
IGC.PRO[ESS,JMC]
JO.NO2[ESS,JMC]
FREE[ESS,JMC]
DRAW.PRO[ESS,JMC]
VIEW1[ESS,JMC]
SPUT.PRO[ESS,JMC]
TECH2.BGL[ESS,JMC]
LORE[ESS,JMC]
SERA.ZOO[ESS,JMC]
SUBTES[ESS,JMC]
CSDCOM.REQ[ESS,JMC]
LIST[ESS,JMC]
BANK[ESS,JMC]
GUBBB[ESS,JMC]
ENERGY.DEM[ESS,JMC]
COLE.LET[ESS,JMC]
LIGHT.6[ESS,JMC]
PRIZE.PRO[ESS,JMC]
IBM.SNG[ESS,JMC]
PUBLIC.INT[ESS,JMC]
PRO2[ESS,JMC]
JAN75.DBX[ESS,JMC]
LIVER.SAI[ESS,JMC]
ANDREI.DAT[ESS,JMC]
REGIST[ESS,JMC]
LUCK.REC[ESS,JMC]
IBM.ABS[ESS,JMC]
GORIN[ESS,JMC]
NEWREP.LE1[ESS,JMC]
ENERGY.QUE[ESS,JMC]
CH1.BH[ESS,JMC]
SCITOP.LIS[ESS,JMC]
BROOKS.PLN[ESS,JMC]
PDP370.PLN[ESS,JMC]
ENERGY.STA[ESS,JMC]
MEMO.PUB[SEN,JMC]
HOOK.LE1[LET,JMC]
WEISKO.LE1[LET,JMC]
MONOP.STA[ESS,JMC]
BULLET.AP[ESS,JMC]
AI2.QUA[ESS,JMC]
SPACE.OP[ESS,JMC]
AISY2.QUA[ESS,JMC]
DATAMA.LE1[LET,JMC]
PROP.ECO[ESS,JMC]
INDIA.PRO[ESS,JMC]
TECTOP.LIS[ESS,JMC]
R1500.3[ESS,JMC]
R1500.1[ESS,JMC]
TURIN2.LEC[ESS,JMC]
CHAP1.ART[ESS,JMC]
CHAP2.ART[ESS,JMC]
R1500.2[ESS,JMC]
SUFFIC[ESS,JMC]
HOTER.PUB[ESS,JMC]
HOTER.DOC[ESS,JMC]
GALLAG.SPI[LET,JMC]
FREGE[W76,JMC]
ENERGY[W76,JMC]
FOL.COM[W76,JMC]
MEET2.SEN[W76,JMC]
IDEOLO[W76,JMC]
HEAVY.SET[W76,JMC]
SENSE.NOT[W76,JMC]
LISP2[W76,JMC]
TASKS.SET[W76,JMC]
CS.226[W76,JMC]
PDP10[W76,JMC]
LISP.OLD[W76,JMC]
UNIT.PRF[W76,JMC]
Z2.AX[W76,JMC]
HEAVY.AX[W76,JMC]
BUREAU[W76,JMC]
SENSE.PST[W76,JMC]
BLOCKS.AX[W76,JMC]
OBVIOU[W76,JMC]
NUMBER.AX[W76,JMC]
COMMUN[W76,JMC]
HEAVY.THE[W76,JMC]
KOREA[W76,JMC]
SENSE.BLA[W76,JMC]
MOYNIH.1[W76,JMC]
REM.WRU[W76,JMC]
RESOUR[W76,JMC]
AAAS.2[W76,JMC]
SNEKOT.DAT[W76,JMC]
SENSE.1[W76,JMC]
FREEDO[W76,JMC]
FUTURE[W76,JMC]
KNOW[W76,JMC]
ACCOMP[W76,JMC]
SEARCH[W76,JMC]
SUN.OUT[W76,JMC]
PARKER[W76,JMC]
IBM[W76,JMC]
E[W76,JMC]
INFORM[W76,JMC]
SENSE.PS1[W76,JMC]
TELLER[W76,JMC]
WADLER[W76,JMC]
SENSE.LAB[W76,JMC]
SENSE.PL1[W76,JMC]
AAAS[W76,JMC]
SCI[LET,JMC]
LICKLI.SPI[LET,JMC]
COMMEN[S76,JMC]
PROBLE[S76,JMC]
PTCH2[ AI,JMC]
BOARD.ART[ AI,JMC]
PACKAR.LE1[LET,JMC]
SHUTDO[S76,JMC]
PUBLIC.LE2[LET,JMC]
PUBLIC.DOC[LET,JMC]
KNOW.PRF[F75,JMC]
KNOW.AX2[F75,JMC]
KNOW.CM1[F75,JMC]
MORAVE[S76,JMC]
PNUELI.LE1[LET,JMC]
PNUELI.LE2[LET,JMC]
FILE[S76,JMC]
CALIF[SEN,JMC]
SAMPLE.LET[LET,JMC]
EXTENS[S76,JMC]
MAY76.IN[LET,JMC]
JUN76.IN[LET,JMC]
PACIFI.LE2[LET,JMC]
PLF.LE1[LET,JMC]
CARNAP[S76,JMC]
HAYASH.LE1[LET,JMC]
HOTER.POX[PUB,JMC]
ADVISO[S76,JMC]
LENG.LE1[LOT,JMC]
JUL10[S76,JMC]
LEM.LE3[LET,JMC]
ROYDEN.LE1[LET,JMC]
HARDY.LE1[LET,JMC]
MCCART.LE1[LET,JMC]
LEVENT.LE1[LET,JMC]
CONSID.LE1[LET,JMC]
FREY.LE1[LET,JMC]
DEC.OUT[LET,JMC]
LEDERB.LE1[LET,JMC]
STELMA.LE1[LET,JMC]
WILHEL.LE1[LET,JMC]
TABLE[LET,JMC]
SLAMEC.LE1[LET,JMC]
MAR.ME[LET,JMC]
PUBLIC.LE1[LET,JMC]
STREET.LE1[LET,JMC]
LUCKHA.LE2[LET,JMC]
KQED.LE1[LET,JMC]
EXAMIN.LE1[LET,JMC]
MILLER.LE2[LET,JMC]
SHOCKL.LE1[LET,JMC]
BERG.LE1[LET,JMC]
RACING.LE1[LET,JMC]
FRANKL.ME1[LET,JMC]
ROY.LE1[LET,JMC]
MILLER.LE1[LET,JMC]
LENG.SPI[LET,JMC]
DWORKI.LE1[LET,JMC]
FRANKL.DOC[LET,JMC]
FU.LE1[LET,JMC]
LEM.LE2[LET,JMC]
HOFFMA.LE1[LET,JMC]
DONIAC.LE1[LET,JMC]
TONGE.LE1[LET,JMC]
COMPUT.REQ[LET,JMC]
SORRY[S76,JMC]
UCLA.LE1[LET,JMC]
KQED.LE2[LET,JMC]
AMPEX.LE1[LET,JMC]
COMMON.LE1[LET,JMC]
HODGSO.LE1[LET,JMC]
DICKEN.LE1[LET,JMC]
JUILLA.LE1[LET,JMC]
KNUTH.LE1[LET,JMC]
MCCLOS.LE1[LET,JMC]
∂10-Apr-78 2254 DCL Organizational meeting ERL 237 at 2.30pm Tuesday 11th. April
To: GROUP.DIS[VCG,DCL]:
We are going to try to organize this quarter's seminars; some of them
will be joint with Sue Owicki's seminar.
Tomorrow talks are:
Tennant's Language Design Methods based on Semantic Principles - Wolf Polak
Class types vs. Modules, an imminent choice - discussion
∂11-Apr-78 1135 DEW Hans Berliner's visit
Hans is planning to come out the week of May 15 through 22. He would like
to know if he should rent a car and stay in a motel or if there are
some apartments on campus to rent (a la IJCAI 73) so he can walk to campus.
He's wants to make reservations soon if he needs to get a motel and car.
Thanks, Dave
He should certainly rent a car unless you intend to chauffeur him.
Come to think of it, there is a chance he could stay at the faculty
club with this much notice. Please ask Patte right away to try to
make a reservation.
∂11-Apr-78 1238 CLT
I haven't done much more than skim through Winston. I will take a look
at the LISP chapters this weekend. The robot is about to take his
first step!
∂12-Apr-78 1105 CCG fellowship
Moszkowitz is out,with the nsf fellowship. It's probably a bad policy to
give fellowships to Wilkins or Filman, since there is too good a chance
they wouldn't get to spend it, and it's not transferable. About three
competitive candidates have arisen in verification and natural language
so we seem to have enough candidates. I'll assume you will go along with this
unless I hear from you.
Cordell
∂14-Apr-78 1351 DPB Comprehensive Exam
To: JMC, REF
Since you were unable to attend the first Comp. Exam
meeting on 4/14 at 1 pm, Frances asked me to send a note
summarizing what happened. Each area was assigned to one
or more members of the committee. Hardware was a problem,
since nobody on the committee is primarily interested in
hardware. Kevin Karplus volunteered to help. It was suggested
that JMC might be interested. Tentatively, JMC and KJK will
do the hardware part. It is also assumed that REF and JMC will
handle the AI part of the exam.
Dates:
Friday Apr. 21, 3:30 Meeting to discuss rough drafts of questions.
6 copies of the drafts should be available for discussion.
Friday Apr. 28. Final drafts of questions due. (There may be a meeting
if open issues exist.)
Saturday May 13. Date of written exam.
Wednesday May 17. Grading of written exam finished. Distribute
programming project to students still alive.
Monday May 22. Programming project due.
Above dates are approximate. P.S. Everybody is supposed to be
thinking up ideas for the programming problem. Discuss next Friday.
(Other Committee members: Frances Yao - Chair, Gene Golub, Jeff Barth,
Kevin Karplus, who are respectively: FFY, GHG, JMB, KJK. Frances and
Gene should not be assumed to log in often.)
-Denny
∂14-Apr-78 1410 CLT Thank you
To: JMC, DEK, ZM, RWW
@ibm
∂14-Apr-78 1502 TOB student for summer
To: LES, JMC
Mark Faust, a student at CMU,approached me
to work here this summer. He seems to be a
good programmer, works as a programmer on the
staff at CMU. I propose to hire him for
summer only.
Tom
∂15-Apr-78 0000 JMC*
rember Bob Smith.
∂15-Apr-78 0000 JMC*
Michel Guerard's Cuisine Minceur Morrow $12.95
∂15-Apr-78 0000 JMC Expired plan
Your plan has just expired. Maybe you should make a new one.
∂15-Apr-78 1538 CARL at MIT-AI (Carl Hewitt)
Date: 15 APR 1978 1838-EST
From: CARL at MIT-AI (Carl Hewitt)
To: jmc at SU-AI
CC: CARL at MIT-AI
John,
Could you please give me the reference to the paper in which you introduce
the "amb" construct also the reference to the paper by yourself and Zohar
in which you dicuss the parallel versions of "and" and "or". I need the references
for a paper which I am finishing.
Thanks,
Carl
∂15-Apr-78 2216 BCM appointment
When is a good time to meet with you? I'm free, say, mon/wed/fir after 3:15
although I would needa way to get to the lab. Is there a time when we could
meet on campus?
Joke: man walks into store in east germany and asks salesgirl, "Is this
the department that sells no stockings?" She answers, "No, that's on the next
floor. This is the department that sells no dresses."
∂16-Apr-78 0234 host MIT-AI
JONL@MIT-MC 04/16/78 05:34:06 Re: New Implementation of LISP
To: (*MSG *ITS) at MIT-MC
CC: lisp-discussion at MIT-AI
"NIL" is an acronym for the new implementation of LISP, currently
being done at MIT's Laboratory for Computer Science. Initial versions
will come up (simulated) under MACLISP on the PDP-10, and soon on
the VAX11/780. Plans are under way to bring up a version on the S1 too.
Currently, NIL is somewhat of a cross between MACLISP and the LISP machine
versions of LISP. There is a NIL directory on MC, and suggestions can
be mailed there; also, development programs and commentary will probably
go on there too.
∂17-Apr-78 0430 LLW Contract Exception Snag
To: LES
CC: LCW, LLW, JMC
Completion of the SAIL-LLL contract execution has run into a major snag
over here centering on the exception you asked the Sponsored Projects
Office to take to the prior-approval-of-travel clause. As I predicted
when I last discussed the matter with you, Milt Eaton was willing to
neither-agree-nor-disagree with the exception, and get the contract signed
off by senior LLL bureaucrats. However, his boss, recalling the messy
squabble over a start-date ambiguity in a contract of last Summer between
the Project and the SU CS Department, insisted that the exception matter
be explicitly resolved, before he would put the first concurring signature
on the contract.
The problem is that LLL's basic contract with the Department of Energy,
which dominates all other derivitive contracts, such as that between the
S-1 Project and the Navy, requires that all per-diem (out-of-Bay-Area)
travel be approved in advance by the Lab Director or his designated
alternate, and that all such travel on any LLL sub-contract be likewise
approved. LLL is totally lacking in capability to waive this requirement
on either itself or its sub-contractors, I have been repeatedly told, and
a request to DoE to DoE for an ad hoc waiver will be routinely rejected,
since the requirement is imposed on DoE by OMB regulations. My hopes for
accommodating your objections have thus been based on getting the LLL
bureaucracy to overlook the exception you took to this clause, rather than
on their explicitly agreeing to it; these hopes have been dashed by Betty
Scott's actions of last Summer, which led to DoE auditors demanding strict
compliance by LLL with DoE sub-contracting regulations with respect to SU,
I was informed this past week.
These developments force me to ask you to reconsider your position on the
prior-approval-of-per-diem-travel matter. Milt Eaton has offered to work
with me to set up a short approval-request form for your use, which would
have only traveller name, travel dates and travel destination blanks for
you or your secretary to fill in and mail to him; verbal approval (as
would suffice for travel on short notice, when mail delay would be too
great) could be requested by your secretary of Mr. Eaton as she dropped
the confirming approval-request form in the mail to him. I suggest that
your personal overhead per trip could thus be kept to under a minute, and
that of your secretary to less than 5 minutes. Since prior approval needs
to be obtained only for out-of-Bay-Area trips by SAIL personnel, I would
expect that the time required of you for the entire contract period for
the travel approval process would be less than 5 minutes.
I continue to be in thoroughgoing sympathy with your desire to keep SAIL
overhead activity to a minimum, and to step heavily on creeping
bureaucracy whenever it encroaches. However, the present problem seems to
arise simply from the basic difference between a government grant and a
government contract with respect to approval of per-diem travel. LLL is
prohibited by its master contract from awarding grants, while SAIL's
present administrative procedures deal only with grants. I continue to
very earnestly hope that the S-1 Project (which is bound by government
contract regulations on LLL) and SAIL can work together formally on the
only basis available to us, that of contract. Would you and John
therefore please reconsider extending SAIL administrative procedures to
include accepting government contract-type funding, particularly as it
affects per-diem travel?
Lowell
∂17-Apr-78 1241 DEW Hans Berliner's talk
Hans is expecting to give a talk when he's here but does not know anything
more about it. I have an abstract of the talk. Are you going to take
care of scheduling it? If so, I have the title, if not should I be doing
anything? Dave
Please arrange scheduling it with Bob Filman.
∂17-Apr-78 1439 REM
To: JMC, MRC
∂16-Apr-78 2253 PBARAN at USC-ISI Cal CBBS
Date: 16 APR 1978 2249-PST
From: PBARAN at USC-ISI
Subject: Cal CBBS
To: PCINNER:
Today SILVER and I picked up the gear
(Cromemco Z2, Persci floppy drive,etc)
for the Cal CBBS from Jim Warren.
As of 10:42PM SILVER had it mostly
working. Hopefully we get
the diskette full of software in the next
day or so. The only real delay looks
like the DAA.
Dave C
-------
[REM - When this Community Bulletin Board System is up and running
under auspices of Byte Shop of Palo Alto and the PCNet Committee,
we may have the first of "useful to Arpanet and SU-AI community"
services available for all to enjoy, thus begin to return some
payment for use of Arpanet and SU-AI during PCNet startup.]
∂17-Apr-78 1613 REM via AMES-TIP#22 Progress on interval compression
ck to working on INNTCOM. Using BISUB (a subroutine to facilitate
establishing a co-routine link between any two programs) I have written
and made-working the following modules:
IN7 Input from file in binary mode, extracts 7-bit bytes but
if line-number bit is on it prefaces with a '205 byte,
and at end of file it returns '200 bytes.
C7OUT Output of 7-bit bytes together with '205 bytes, constructing
a binary file. This is the inverse of IN7, even for non-text files.
ASCBAU Input a stream of 7-bit ASCII bytes together with '205 tokens
and a '200 at EOF, and convert to Pseudo-Baudot bytes.
BAUASC Input a stream of Pseudo-Baudot bytes and convert to ASCII.
This is the inverse of ASCBAU, even for binary data.
MAINLINE -- Patches the above IN7 --> ASCBAU --> OUTCHR --> BAUASC --> C7OUT
and thus copies any file (even binary) by treating it internally
as a text file with '205 tokens for the 36th bit where necessary.
Next to install the remaining modules to do INTCOM of files...
∂17-Apr-78 1643 DCL
To: GROUP.DIS[VCG,DCL]:
***********************************************************************
VERIFICATION GROUP SYSTEM MEETING TUESDAY 18TH APRIL
PLACE: ERL 237
TIME: 2:30 pm
TITLE: SPECIFICATION AND VERIFICATION
OF
MODULES AND CLASSES
SPEAKER: David Luckham and Wolf Polak
We compare Class and Module constructs and discuss the requirements for
"encapsulation". We propose a method of specifying Modules and proof
rules for verifying both modules themselves and programs using them.
A write-up is in preparation!
∂18-Apr-78 0537 REM via AMES-TIP#20 Further progress report
Binary I/O (varying sizes of bytes overlapping word boundaries, used to
read and write crunched files) have been copied from CRU2.FAI and
modified to all run as co-routines using BISUB. Dummy crunch and
uncrunch routines have been added. The resultant program is able to
"crunch" a file and to "uncrunch" it back to be BINCOM equivalent to
the original, an improvement over CRU2, SPINDL and friends which
could only handle clean ascii files with no strangness. Next to
install some actual crunch/uncrunch modules...
∂18-Apr-78 1041 Feigenbaum at SUMEX-AIM Margaret would like to speak with you
Date: 18 Apr 1978 1041-PST
From: Feigenbaum at SUMEX-AIM
Subject: Margaret would like to speak with you
To: jmc at SAIL
John, could you call as soon as you read this? Margaret would like to talk
with you re the AI year at the Center.
Could you join us for lunch at the faculty club , about 12:30pm?
Ed
-------
∂18-Apr-78 1320 Barr at SUMEX-AIM CS Dept. Picnic
Date: 18 Apr 1978 1317-PST
From: Barr at SUMEX-AIM
Subject: CS Dept. Picnic
To: jed at SAIL, les at SAIL, jmc at SAIL,
To: FEIGENBAUM at SUMEX-AIM, TAJNAI at SUMEX-AIM, bs at SAIL
This is a preliminary announcement of the date for the Department
Picnic, May 19, so that maybe we can avoid major conflicts for that
afternoon. The picnic will be held in Huddart Park in Woodside,
where a nice picnic area has been reserved. More later.
/Av
-------
∂18-Apr-78 1620 Feigenbaum at SUMEX-AIM dinner plans
Date: 18 Apr 1978 1619-PST
From: Feigenbaum at SUMEX-AIM
Subject: dinner plans
To: jmc at SAIL
We're getting together at 7pm in the lobby of the Palo Alto
Holiday Inn, then will proceed to dinner next door at the
Chinese Restaurant.
Ed
-------
∂18-Apr-78 1705 Tajnai at SUMEX-AIM Ph.D. Minor Exam
Date: 18 Apr 1978 1704-PST
From: Tajnai at SUMEX-AIM
Subject: Ph.D. Minor Exam
To: JMC at SAIL, RAK at SAIL
cc: Tajnai, DPB at SAIL
John,
We are trying to spread out the job of administering the Ph.D. CS minor
oral exam (under the old rules). Dileep Divekar has requested an exam
and Richard Karp has agreed to assist. Will you please be the professor
on the exam committee?
He has taken 111, 211, 135, 206 211B, 212.
Dick said he can cover 135 and you can cover 206.
Please advise.
Carolyn
-------
OK, if it can be here at the AI Lab at a mutually convenient time.
∂18-Apr-78 2029 REF
We seek a metric to prove the termination of the production system:
1] ¬¬α → α
2] ¬(α∧β) → ¬α ∨ ¬β
3] ¬(α∨β) → ¬α ∧ ¬β
4] α∧(β∨∂) → (α∧β)∨(α∧∂)
5] (α∨β)∧∂ → (α∧∂)∨(β∧∂)
We consider the following metric.
¬E = 3↑E - 1 (the negation of an expression is 3 to the expression, less one)
E1∧E2 = E1*E2 -1
E1∨E2 = E1+E2
and the value of any atomic expression is 2.
Thus, we have
1] 3↑(3↑α-1) -1 > α, trivially, for all α≥2
2] 3↑(α*β-1) -1 > 3↑α + 3↑β for all α,β≥2
3] 3↑(α+β) - 1 = 3↑α*3↑β-1 > (3↑α - 1)*(3↑β - 1) for α,β≥2
4] α*(β+∂) - 1 = α*β+α*∂-1 > α*β-1+α*∂ - 1
5] (α+β)*∂ - 1 = α*∂ + β*∂ - 1 > (α*∂-1) + (β*∂-1)
∂19-Apr-78 0903 Tajnai at SUMEX-AIM (Response to message)
Date: 19 Apr 1978 0902-PST
From: Tajnai at SUMEX-AIM
Subject: (Response to message)
To: JMC at SU-AI
In response to your message sent 19 Apr 1978 0833-PST
Thank you. I'll give him your number and he will contact you direct.
Carolyn
-------
∂19-Apr-78 1400 BCM akademicheskie soveti
kogda vam vozmozhno co mnoi vstrechat'c'a?
venjamin stepanovich moszkowski
Any afternoon you can make it up here. I can probably meet you
on campus on some mornings or evenings. Telephoning in the afternoon,
like now, is a good way to make an appointment.
∂19-Apr-78 1506 AJT THE END??
NOW THAT BRUCE BUCHANAN AND ROGER SHEPARD HAVE
BOTH GIVEN THEIR BLESSING, AND YOU'VE SEEMED TO SAY OK, AND
I'VE FIXED UP TYPOS AND SO ON, CAN I GO AHEAD AND
GET COPIES MADE OF MY THESIS FOR SIGNATURE. I THINK I'LL
GET IT COPIED PRIVATELY AND WORRY ABOUT A PSSIBLE AI MEMO(??????)
LATER.
RSVP ARTHUR
Go ahead full steam.
∂19-Apr-78 1709 Hart at SRI-KL (Peter Hart) Panel Discussion
Date: 19 Apr 1978 1711-PST
From: Hart at SRI-KL (Peter Hart)
Subject: Panel Discussion
To: JMC at SU-AI
John,
As you have probably seen by now, you are announced as
being a member of a panel discussion on the future of AI on May 11.
A secretarial error was responsible for your being listed; however,
I do hope that you will be able to participate, and that you faild
merely through an oversight to respond to my earlier msg.
Peter
-------
∂19-Apr-78 1727 JB MY REPORT.
HELLO, JOHN. HOW WAS YOUR TRIP?
I WOULD LIKE TO HEAR YOUR COMMENTS ABOUT MY REPORT.
∂19-Apr-78 2111 Geoff at SRI-KA (Geoffrey S. Goodfellow) [lauren at UCLA-Security: MSGGROUP# 662 More ''advances'' from ''Quasar'']
Date: 19 Apr 1978 2059-PST
Sender: GEOFF at SRI-KA
Subject: [lauren at UCLA-Security: MSGGROUP# 662 More "advances" from "Quasar"]
From: Geoff at SRI-KA (Geoffrey S. Goodfellow)
To: JMC at SAIL, LES at SAIL
Message-ID: <[SRI-KA]19-Apr-78 20:59:50.GEOFF>
Gents, FYI -
Begin forwarded message
===========================
Mail from USC-ISI rcvd at 19-Apr-78 1925-PST
Mail-from: UCLA-SECURITY rcvd at 1-APR-78 0254-PST
Date: 1 Apr 1978 0252-PST
Sender: lauren at UCLA-Security
Subject: MSGGROUP# 662 More "advances" from "Quasar"
From: lauren at UCLA-Security
To: MSGGROUP at ISI
Redistributed-To: [ISI]<MsgGroup>Mailing.List;152:
Redistributed-By: STEFFERUD (connected to MSGGROUP)
Redistributed-Date: 19 APR 1978
It's been sometime now since we've heard anything new from
"Quasar Industries," the firm which has made all of those
marvelous advances in AI research and robotics while the rest
of the computer science community struggles along with basic
research.
As you may recall, "Quasar" is the firm claiming to market
"intelligent household androids," capable of washing dishes,
babysitting, serving you drinks, and all sorts of other
wondrous things. A "prototype" unit, variously known
as "Sam Strugglegear" and "Klatu" has been making the rounds of
shopping centers in major cities throughout the U.S. over the
past few months.
As the "Quasar" claims have spread, various network-related groups have
actually observed "Sam" in action, and have generally given rather,
uh, unfavorable reports. (I should add that recently a number
of us here in Southern California saw the unit in action at a
local department store; the group included people from UCLA, Rand,
ISI, and other local sites. What we saw so closely resembled
previous reports that it does not merit further discussion
here, other than to say that when "Quasar" felt "threatened" during
their demonstration, their immediate response was to try have us
ejected from the store!)
In any case, some new information has begun circulating concerning
our friends at "Quasar." They have now announced some new
products that will accompany their household androids line.
These include a "robot guard" supposedly for industrial and
governmental use, and maybe even a "robot paramedic" unit.
The guard is touted as a highly versatile unit, and will
be equipped with all sorts of "non-lethal" restraining/disrupting
devices, such as strobe lights to temporarily blind the intruder,
and an ultrasonic unit to "confuse" the unfortunate person.
As usual, these announcements are being stated as fact by the
mass media, talked about in much the same manner as the announcement
of a new type of steel-belted tire. It is indeed unfortunate
that this situation has not improved; indeed it appears to be
getting worse.
The most lengthy article I have seen to date concerning "Quasar"
and their products is in the April 1978 edition of "Interface
Age" magazine (an issue devoted to robotics, what else?)
This article seems to give more information about the internals
of "Quasar" than any previous publications. "Quasar" appears to be
claiming that the department store demos have just been a means to
help provide R&D funds. The implication from Quasar continues
to be that the "household android," "robot guard," and other
forthcoming products, are real, have all of the amazing capabilities
they claim, and will be marketed. I suggest that anyone wishing
to gain a little more insight into "Quasar" should read
the "Interface" article.
Well, that about does it for this installment. Isn't it nice
to know that we no longer need work on all of these
AI and related projects; that "Quasar" has already solved
all the problems and is ready to market the fruits of their
labors?
I welcome comments on the whole "Quasar" business from any
interested parties.
--Lauren--
[Lauren@UCLA-SECURITY]
-------
===========================
End forwarded message
∂19-Apr-78 2110 Geoff at SRI-KA (Geoffrey S. Goodfellow) [FARBER at USC-ISI: Re: MSGGROUP# 662 More ''advances'' from ''Quasar'']
Date: 19 Apr 1978 2101-PST
Sender: GEOFF at SRI-KA
Subject: [FARBER at USC-ISI: Re: MSGGROUP# 662 More "advances" from "Quasar"]
From: Geoff at SRI-KA (Geoffrey S. Goodfellow)
To: JMC at SAIL, LES at SAIL
Message-ID: <[SRI-KA]19-Apr-78 21:01:03.GEOFF>
P.S.
Begin forwarded message
===========================
Mail from USC-ISI rcvd at 19-Apr-78 2009-PST
Date: 19 APR 1978 1830-PST
Sender: FARBER at USC-ISI
Subject: Re: MSGGROUP# 662 More "advances" from "Quasar"
From: FARBER at USC-ISI
To: lauren at UCLA-SECURITY
Cc: [ISI]<MsgGroup>Mailing.List;152:
Message-ID: <[USC-ISI]19-APR-78 18:30:08.FARBER>
In-Reply-To: Your message of APRIL 1, 1978
Lauren
Indeed I wish to relate more on the story of the Quasar saga.
When I was in beautiful Boca Raton Florida ( temperature 85 and
the same humidity), I heard on the local radio and newspaper that
the Dade county police were considering the purchase of a Quasar
robot to act as a guard in the county jail. It costs $70000 and
was full of marvelous facilities including a large vocabulary and
various strobe devices designed to subdue the poor escaping
inmate. The basic issue that the paper reporte4d on was the
study by the county as to whether the purchase of such a robot
was cost effective .
Anyone for a good product liability lawsuit the first time that
robot hurts an inmate.
Dave
===========================
End forwarded message
∂20-Apr-78 1650 BCM appointment
i tried to call you by phone today but I couldn't reach you. when is
a good time to call?
Afternoon is good or come by tomorrow afternoon if you can. I'll get
you a ride back.
∂20-Apr-78 2009 DEW ARPA TALK
Sorry my talk wasn't more coherent. I could do better if I had a few days to
prepare and work out some examples and slides. Since the time was so short
I tried to pick one point rather than explain evrything and it didn't sound very
good.
I think everyone understood the problem.
∂20-Apr-78 2023 Geoff at SRI-KA (Geoffrey S. Goodfellow) NS
Date: 20 Apr 1978 2022-PST
Sender: GEOFF at SRI-KA
Subject: NS
From: Geoff at SRI-KA (Geoffrey S. Goodfellow)
To: JMC at SAIL
Cc: Greg at SRI-KL
Message-ID: <[SRI-KA]20-Apr-78 20:22:38.GEOFF>
One of our local whiz kids, Greg Hinchliffe, is interested in
doing some hacking on the long time unsupported News Service
system at SAIL; Could something be arranged of mutual benefit
along these lines?
I suspect we could come up to SAIL sometime early next week or
this weekend to talk over ideas, etc. if you'd like.
[Geoff]
Sounds like a good idea. Martin Frost should be part of it too. I
suggest some afternoon. Please telephone.
∂20-Apr-78 2039 REM Working Interval-Compression program for files
Today I finally got everything working, IC1.FAI[1,REM] is source,
IC1.DMP[1,REM] is working adaptive cruncher and uncrunch. Warning,
I haven't yet installed headers to identify the version of the cruncher
and the parameters used in crunching, so don't delete the source and
expect to be able to get it back. This is just a test version!
First major test, crunching the source for the program that did the
crunching, plaintext 4.9k --> crunched 3.3k (single-character optimum-code
using adaptive Markov model and Pseudo-Baudot coding). With any file
that contains tremendous redundancy it should do much better, adapting
itself to any character frequencies, assuming independence between characters.
∂20-Apr-78 2051 REM More IC1 results
(1) Using teco, creating 200 identical lines "FOOOOOOOOOOOO" <CRLF>
IC1 achieves a 10:1 compression ratio.
(2) Using teco, creating a file consisting of nothing but 20000 letter "E"
IC1 crunches the 1.9 k file down to 5 words!!!!!
Pretty adaptive, don't you say?
∂20-Apr-78 2154 REM
New version of IC1.DMP is 41% faster for crunch. Same for uncrunch as before.
∂20-Apr-78 2234 REM
Ok, I've added my usual password header at front of crunched file, feel
free to play with it although at present I can't guarantee compatibility
of cruncher with uncruncher until it's been thoroughally tested.
Note that it works on binary files as well as text files, although the
Markov model is designed for text files. -- Current program has no context,
just one top-level histogram that is adaptive and optimum (better than
Huffman code, virtual exact match to entropy even if entropy is much less
than one bit per token crunched).
∂21-Apr-78 0525 Feldman at SUMEX-AIM Mike Gordon's visit
Date: 21 Apr 1978 0525-PST
From: Feldman at SUMEX-AIM
Subject: Mike Gordon's visit
To: FWH at SU-AI
cc: jmc at SU-AI
Hello, I'm flying in from Rochester today and may appear at the AI Lab
on Monday. Will be staying chez Waldinger.
Looking forward to seeing you
................. Mike Gordon
-------
∂21-Apr-78 1054 STT 224 lecture
Your lecture on "Formal Methods for Representing Knowledge" is scheduled for
Tuesday, May 9 at 11:00. (If there is any problem please let me or Ed know.)
What would you like students to read before your lecture, as background? It
should be about 30 pages, plus or minus 15.
Thank you.
∂21-Apr-78 1115 Geoff at SRI-KA (Geoffrey S. Goodfellow) [ Robert M. Frankston <Frankston at MIT-Multics>: MSGGROUP# 669 Quasar]
Date: 21 Apr 1978 1105-PST
Sender: GEOFF at SRI-KA
Subject: [ Robert M. Frankston <Frankston at MIT-Multics>: MSGGROUP# 669 Quasar]
From: Geoff at SRI-KA (Geoffrey S. Goodfellow)
To: JMC at SAIL, LES at SAIL
Message-ID: <[SRI-KA]21-Apr-78 11:05:30.GEOFF>
Begin forwarded message
===========================
Mail from USC-ISI rcvd at 21-Apr-78 0409-PST
Mail-from: MIT-MULTICS rcvd at 19-APR-78 2128-PST
Date: 20 April 1978 0020-est
From: Robert M. Frankston <Frankston at MIT-Multics>
Subject: MSGGROUP# 669 Quasar
To: msggroup at USC-ISI
cc: lauren at UCLA-Security, gumpertz at CMU-10a
Redistributed-To: [ISI]<MsgGroup>Mailing.List;152:
Redistributed-By: STEFFERUD (connected to MSGGROUP)
Redistributed-Date: 20 APR 1978
First, a comment, since there appears to be a general inerest in Quasar
(and maybe related hacks) and since the delay via msggroup is so long,
it would be nice if someone on a forwarding host (such as ITS) could
establish an appropriate mailing list ("QUASAR-ETC"?) so that
distribution can be more efficient.
For those who are unaware, the Boston Globe had an article in their
magainze section March 12, 1978. It was featured on the cover of the
magazine section. It did include quotes from people such as Minsky and
Fredkin, but more with the attitude of "it just goes to show you that
those academicians can't do anything practical, and all you need is some
guy working in the back of a garage to put them to shame" (this is not a
quote, just an attempt to convey the tone of the article. If yo are
ierested, perhaps you can write to the Globe for a copy of the issue.
From the index to the magazine section:
After R2-D2, Klatu?
There could be a robot around the house -- possibly as soon as next
year -- to dust, vacuum, babysit or event serve cocktails if the
prediction of one manufacturer comes true. By Andrew Blake.
Andrew Blake is a staff writer for new england magazine.
-------
===========================
End forwarded message
∂21-Apr-78 1348 BCM appointment
I can't make it today. I'll try next week.
∂21-Apr-78 1537 CLT
FILES USED IN PREPARATION OF S78 VERSION OF BOOK
Pub macros:
BKMAC PUB LSPCLT 2.0 22-MAR-78 0122 000 1CLT E 03-APR-78 04-APR-78 P1101>
Text:
PREFAC LSPCLT 1.1 22-MAR-78 0128 000 1CLT E 31-MAR-78 04-APR-78 P1101>
INTRO LSPCLT 15.5 21-MAR-78 2322 000 1CLT E 22-MAR-78 04-APR-78 P1101>
RECFUN LSPCLT 11.5 22-MAR-78 0015 005 1CLT E 22-MAR-78 04-APR-78 P1101>
FOLISP LSPCLT 15.6 22-MAR-78 0252 005 1CLT E 22-MAR-78 04-APR-78 P1101>
IMPURE LSPCLT 7.3 22-MAR-78 0148 000 ESSJMC E 22-MAR-78 04-APR-78 P1101>
EXTEND LSPCLT 4.6 22-MAR-78 0254 000 1CLT E 22-MAR-78 04-APR-78 P1101>
ABSNTX LSPCLT 7.5 22-MAR-78 0230 000 ESSJMC E 22-MAR-78 04-APR-78 P1101>
COMPIL LSPCLT 8.0 22-MAR-78 0257 005 1CLT E 22-MAR-78 04-APR-78 P1101>
MTLISP LSPCLT 2.3 21-MAR-78 1348 005 1CLT E 22-MAR-78 04-APR-78 P1101>
LSPBIB PUB LSPCLT 1.0 22-MAR-78 0303 005 1CLT E 22-MAR-78 04-APR-78 P1101>
FOLAPX LSPCLT 9.2 22-MAR-78 0304 005 1CLT E 31-MAR-78 04-APR-78 P1101>
LCOMAP LSPCLT 2.8 20-MAR-78 0250 005 1CLT E 22-MAR-78 04-APR-78 P1101>
Assemblers:
LSPDOC PUB LSPCLT 640 21-MAR-78 0453 005 1CLT E 21-MAR-78 04-APR-78 P1101>
LSPDC1 PUB LSPCLT 512 21-MAR-78 0451 000 1CLT E 31-MAR-78 04-APR-78 P1101>
TOTAL= 113.8
LSPCMT on LSPCLT p7 has annotated directory of chapters.
LSPCMT generally has documentation, ideas and pointers.
∂22-Apr-78 0541 MRC For your information
By JEROME IDASZAK
(c) 1978 Chicago Sun-Times
CHICAGO - Western Union by 1990 might be providing
message service by terminal directly into people's homes,
according to the company chairman.
No distant dream, R. W. McFall said it might happen
because of competition from the U.S. Postal Service.
The postmaster general told a congressional committee
that the post office plans by 1985 to test
10 cities for overnight delivery of mail through
electronic transmission.
McFall told shareholders at Western Union's annual meeting in
Chicago that the postal system's plans pose no
immediate threat to Western Union's popular Mailgram service.
''They don't expect full electronic delivery until the
1990s, and we're not going to stand still ourselves,''
McFall told the meeting at the Continental Plaza hotel.
''I think by 1990 you'll see a lot of terminal-equipped
homes. You can go to a Radio Shack today and buy a home
computer for $500, and that's not much more than a color
television costs.''
Western Union's Mailgram service started in 1970 and
transmits messages to post offices for delivery with the
next business day's mail. It's been a growth area for the
company, and McFall said that during this year's first quarter,
600,000 Mailgrams were sent each week, an increase of
10 per cent over the same period last year and a company
record.
The question about competition from the U.S. government came
from a shareholder after McFall said the tariff rate
structure has brought a lower rate of return on Mailgrams
to Western Union so far but that the rate would increase
as more years pass.
Though Mailgram service, telegrams and money orders
are the most visible of Western Union's revenue generators,
those combined in 1977 accounted for one-fourth of the
company's $650,460,000 revenue. The bulk came from
teletypewriter networks provided business firms and from
leased communication systems to private business and the
federal government.
The latter included installation last year for the U.S.
Department of Agriculture of a data system using meteor burst
technology to collect and communicate snow and rainfall
information in Western states.
At the first annual meeting in 10 years in Chicago of
the New Jersey-based company, shareholders were told
first quarter results were slightly above a year
ago. The company posted net income of $11 million,
or 56 cents a share, on revenues of $164 million. That
compares with a year ago net of $9.6 million, or 55
cents a share, on revenues of $157 million.
The company faces a major problem in 1979, McFall
said, when it might have to pay substantially higher rates
for facilities that Western Union leases from American
Telephone and Telegraph because long-term contracts
expire this year. The leasing fee was $76 million last year,
should be $90 million this year and higher in 1979.
By 1980, McFall said the growth of Western Union's own
transmission capacity should reduce such leasing fees.
McFall also told shareholders the company is pleased
that the Federal Communications Commission has begun
an inquiry into telegram competition. ''We are not
afraid of competition so long as we are allowed to
compete on the same basis as everyone else,'' McFall said.
He added he would also like to see legislation that
would eliminate a ban on Western Union from providing
its new communication services internationally.
In voting at the meeting, two shareholder proposals from
Lewis D. Gilbert were defeated but received significant
support. A proposal to allow shareholders to buy new
stock before being offered through brokerage firms
- so-called pre-emptive rights - gained 16.3 per cent in favor.
And a proposal that no profit-sharing compensation be
paid in a year when no cash dividend on common stock is
declared won 17.9 per cent of those voting.
∂22-Apr-78 1035 Feigenbaum at SUMEX-AIM (Response to message)
Date: 22 Apr 1978 1035-PST
From: Feigenbaum at SUMEX-AIM
Subject: (Response to message)
To: JMC at SU-AI
In response to your message sent 22 Apr 1978 0956-PST
John, my initial reaction is that it's a great idea. I can think of
many other reasons why it should be at the 100 level and part of the
comprehensive. LISP is one of the essential building blocks of
literacy in computer science.
I would like to submit the McCarthy-Talcott manuscript to McGraw-Hill for
review leading to inclusion in my Computer Science Series.
Is there a manuscript ready for review? Are you agreeable? I would
appreciate it. I am particularly motivated becuase I do not have a
book by you in my series and would very much like one.
::::::::::::::::::::::::::::::::::::::::
new subject:
did Laurie get hold of you re Manna papers?
We forgot one thing: a short description of the professional activities and
stature in the field of each of the external letter writers (so they can
calibrate our letters). They are not holding up the papers for this, but would
like the info just as soon as possible. Could you write this up--one
short paragraph for each external letter writer?
Ed
-------
There is a current edition being used in CS206. It will be greatly
revised and enlarged. I have just put a copy of the present version
into interdepartmental mail. I will have several questions:
1. How does McGraw-Hill compare financially with others?
2. Can they do a good production job? I'm not too impressed with
the job they did on Manna's book. We might like to use Knuth's
new system or at least his new fonts.
3. How would they feel about a preliminary edition? There is already
some demand. I wouldn't want the present edition to be the preliminary
edition, but next Fall's edition might serve as its basis.
4. Isn't John Allen's book in your series? What about competition
within the line?
I'll send Laurie the description of the letter writers.
∂22-Apr-78 1057 Feigenbaum at SUMEX-AIM good questions
Date: 22 Apr 1978 1058-PST
From: Feigenbaum at SUMEX-AIM
Subject: good questions
To: jmc at SAIL
I'll come up with some good answers and get back to you,
Ed
-------
∂22-Apr-78 1115 REM via AMES-TIP#66 Working version of my new data-compression program, for disk files
To: CRUNCH.DIS[1,REM]: CRUPUB.DIS[1,REM]:
The file SU-AI: IC1.DMP[1,REM] is an experimental apparantly-working
implementation of my Interval-Subdivision data-compression (also known
as my proposal for the "Data-Compression Standard") algorithm, invented
by me 1978.Feb.12 -- At present this program uses an adaptive Markov
process, starting out with a uniform histogram of 108 Pseudo-Baudot
tokens, adding J (default 105) for each occurrance of a token, keeping
N (default 16) bits of information, using Pseudo-Baudot conversion between
the actual file and the compression-per-se. Program is 8.9 k, which
includes RAID (like DDT only better). Without RAID it is 1.6 k.
∂22-Apr-78 1142 LES Overhead money
Interesting idea. I'll look into it.
∂22-Apr-78 1155 REM
∂22-Apr-78 1145 JMC
What's your plan for context? Without it the compression is inadequate.
[REM -- (1) For ultra-redundant files such as a few thousand "E"
this code is much better than any previous code that I know of
that is general-purpose, although indeed where context is useful
this program compares unfavorably with CRU2 and CRU3/SPINDL.
(2) I am at a "benchmark" now and will allocate the next week or
so to other things such as IMSSS and PCNet before returning to
IntCom to work out a left-context Markov-model. The major
question is which of three ways to handle escaping from a context
to its parent (to handle a rare character), (a) pure-escape with
wasted addressing space (b) escape with repeated tokens eliminated
(c) weighted average between context and its parent-chain. The
latter is probably too expensive, the former wasteful but fastest,
the middle not sure how expensive...]
∂22-Apr-78 1602 REM via AMES-TIP#66
New IC1.DMP has (1) user typein of parameters without going thru RAID (2)
I is now a variable instead of constant 1 (3) misc.improvements - still
null context only.
∂22-Apr-78 1704 BCM appointment
is monday, wednes or fri afternoon ok? have you any ideas on how i can
get to the lab. Or we could meet for lunch tues,wed,thur,fri.
Any of the above days is ok. I don't know how to get you to Lab,
but many people go up from campus. Perhaps you could put up a message
asking for a ride. mail * does it.
∂23-Apr-78 1017 LEFAIVRE at RUTGERS-10 QUASAR ROBOT STILL LIVES
Date: 23 Apr 1978 (Sunday) 1319-EST
From: LEFAIVRE at RUTGERS-10
Subject: QUASAR ROBOT STILL LIVES
To: AMAREL, RSMITH, SRIDHARAN, KAPLAN, DSMITH,
To: JOSH, GOLDBERG
cc: MINSKY at MIT-AI, HENRY at MIT-AI, DHT at MIT-AI,
cc: KEN at MIT-AI, KLH at MIT-AI, JMC at SU-AI,
cc: REID at CMU-10A, BAK at MIT-ML
Much to my surprise, given the flurry of activity last November/December,
the Quasar "robot" appears to be alive and well. Following is an article
which appeared in the New Brunswick HOME NEWS today (Sunday, April 23):
---------------------------------------------------------------
TALKING ROBOT TO BE SHOWN
Its name is Klatu. It can walk your dog, answer your doorbell, vacuum the
rugs and babysit your kids.
Quasar Industries of Rutherford claims that its creation, Klatu, a 5-foot
2-inch, 180-pound talking robot, can perform all these tasks and more.
Critics are skeptical, however, of Quasar's claims. Several computer experts
have expressed doubt that technology is far enough advanced at present to
make the development of such an android possible.
The public is invited to judge for itself during an outdoor demonstration
of Klatu's abilities beginning at noon on Wednesday, April 26, at Middlesex
County College. The demonstration is sponsored by the Forum Committee of
the College Center Program Board and is open to the general public free of
charge. The rain location will be inside the College Center.
Called a "Domestic Android", Klatu will be marketed in less than two years
for about $4000, according to Quasar President Anthony Reichelt, and could
be the ultimate appliance of the near future. On the other hand, its critics
wonder if it is indeed a self-contained functional robot or a machine
operated by humans using remote controls.
For further information on the demonstration, call the college Office of
Student Activities, 548-6000, extension 327.
------------------------------------------------------------------
I have a class to teach on Wednesday afternoon, but hope that some other
interested faculty or graduate students will be able to attend this
"demonstration" of Klatu's abilities.
∂23-Apr-78 1143 DCL
To: GROUP.DIS[VCG,DCL]:
***********************************************************************
VERIFICATION GROUP SYSTEM MEETING TUESDAY 25TH APRIL
PLACE: ERL 237
TIME: 2:30 pm
TITLE: THEOREM PROVING WITH RULES.
SPEAKER: Bill Scherlis
The rule handler in the verifier is described. Its features include
(1) forwards and backwards rules, (2) semantic matching with constraints,
(3) process characterization of rules and matching, (4) case splitting,
(5) rule schemata, and (6) proof search controls.
The rule mechanism interfaces to a large class of propositional
algorithms. The interface specifications are described (they are essentially
those of the Nelson/Oppen satisfiability programs). It presently works
in conjunction with two algorithms: one by Nelson/Oppen and the other by
Steve German. These algorithms work by building up data bases of assertions
and using PUSH and POP to change contexts.
Matching in the data base is described. The matcher makes use of
certain properties of equality and arithmetic, and can be extended in a
uniform way.
This is a preliminary talk.
∂23-Apr-78 1535 CCG new lisp course
I like the idea, but suggest that there be two versions --one for
undergraduates and for for graduate students in csd. I nearly always have
a bimodal distribution of preparation and interest level in cs206. The result
is always a compromise wtihe both modes grumbling.
Actually, the version you suggest is closer to the advanced version.Perhaps
it should have a 200 number and be required for the comp. Then have an easier
undergradute version?
Cordell
∂23-Apr-78 1625 RSMITH at RUTGERS-10 quasar
Date: 23 Apr 1978 (Sunday) 1928-EST
From: RSMITH at RUTGERS-10
Subject: quasar
To: JMC at SU-AI
cc: LEFAIVRE, AMAREL, RSMITH, minsky at MIT-AI
We are going to get a group together. Thanks much for
your comments on the remote control possibilities.
Are there any robotics/hardware people around here that
you might recommend to help?
REgards,
Bob Smith
Pavlidis at Princeton or Max Mathews at Bell Labs Murray Hill might know
someone.
∂23-Apr-78 1804 ELLEN at MIT-MC (V. Ellen Lewis) Your MC password, for TIP use
Date: 23 APR 1978 2105-EST
From: ELLEN at MIT-MC (V. Ellen Lewis)
Subject: Your MC password, for TIP use
To: JMC at MIT-MC
Your password has been set as you requested.
∂24-Apr-78 1010 100 : REM via AMES-TIP#51 IC1
Before I add context to IC1 there are at least three enhancements I want
to install: (1) Allowing non-Baudot conversion, i.e. encoding all 128
ASCII bytes and 2 escape codes (LINNUM and EOF) instead of converting
to Baudot and encoding the non-lower-case 102 ASCII bytes and 6 escape
codes (UPPER LOWER TEMPUPPER CRLF LINNUM and EOF); (2) Allowing bytesize
other than 7 (or even 7) in binary mode where if bytesize doesn't
divide 36 it merely splits bytes across word boundaries instead of
handling the extra bits as LINNUM bits; (3) Allowing a semi-oneway
password for encryption of the binary data as it is crunched, part of
the hash of the keyword would be inserted in the header of the file
so that when uncrunching if a wrong password is given the program
will almost certainly know it and ask for a different password, but
the file doesn't contan enough of the hashed-keyword to allow easy
guessing of the whole encryption key by someone not knowing the original
text-keyword that was typed in.
∂24-Apr-78 1110 JB THESIS.
I'd like to hear your feedback on my report so I can
proceed to a more definitive write-up.
∂24-Apr-78 1119 DON hard one
∂23-Apr-78 1132 JMC
I did the rest; where's your hard one again.
I assume you're referring to cryptograms. The hard one (which, by the way,
I haven't had time to solve for myself, so don't tell me the answer!) is in
CRYP.TMP[1,DON] (a CRYPTO file).
∂24-Apr-78 1134 Levin at SUMEX-AIM reminder
Date: 24 Apr 1978 1102-PST
From: Levin at SUMEX-AIM
Subject: reminder
To: ghg at SAIL, jmc at SAIL
gene and john:
tomorrow there will be 2 mtgs:
1. 1:15 for tenured faculty re: baskett
2. 2:15 for general faculty
both will be in Polya 152...laurie
-------
∂24-Apr-78 1202 Levin at SUMEX-AIM brief biographical info on referees for manna
Date: 24 Apr 1978 1202-PST
From: Levin at SUMEX-AIM
Subject: brief biographical info on referees for manna
To: jmc at SAIL
cc: levin
john:
list of referees:
ashcroft
hansen
burstall
goldberg
hoare
igarashi
nilsson
nivat
pnueli
salomaa
waldinger
yeh
wegbreit
dijkstra
the note from the assoc. dean, Jim Rosse, said the report should
include..."a brief comment on the stature and competence to judge
the candidate of all referees whose judgments have a significant
bearing on the recommendation."
ed said this info should get to the dean soon, otherwise the process
will be delayed...let me know if there is anything that i can do...laurie
-------
∂24-Apr-78 1850 CARL at MIT-AI (Carl Hewitt)
Date: 24 APR 1978 2150-EST
From: CARL at MIT-AI (Carl Hewitt)
To: jmc at SU-AI
CC: CARL at MIT-AI
Jean Sammet asked Barbara Liskov to ask me to please ask you to send
me your draft slides for your talk on the history of LISP.
If you do so then I will look at them but don't do it if you don't
feel like it.
Cheers,
Carl
∂24-Apr-78 2059 100 : REM via AMES-TIP#66 Your comments on my three enhancements to do
Good point. Most important is to push toward a real demo of its
superiority over existing methods, in a limited domain, rather than
to extend its domain to everything; in order to have maximum
likelihood of getting patents and/or copywrights and/or contracts
to develop it for companies, and get rich and famous. I'm now
considering postponing the three enhancements until after full demo.
∂25-Apr-78 1009 RSP
∂24-Apr-78 2156 JMC
Why won't efind FIND MCCARTHY IN FOO[ESS,JMC]?
[The file is not properly sorted. The E directory page shows that the search
should begin on page 6, which is correct. However "MIT" is found on that
page before "Mc..." and the search terminates. As you see, the problem is the
distinction between the ASCII collating sequence and a library sorting
sequence.]
If the search is linear within a page, it might as well continue to the
end of the page. The fundamental optimization of efind is bring fewer
pages into core. It is not necessary to optimize further within the
page as the time taken by search within the page is dominated by
the disk transfer involved in bringing the page into core.
Moreover, since SSORT is the main program that will be used for bring
files into a form for EFIND, EFIND should be compatible with it.
∂25-Apr-78 1146 Levin at SUMEX-AIM mtgs today
Date: 25 Apr 1978 1146-PST
From: Levin at SUMEX-AIM
Subject: mtgs today
To: jmc at SAIL
john:
REMINDER: tenured faculty mtg. today, 1:15, polya 152 (re:Baskett)
general faculty mtg. today, 2:15, polya 152
...ll
-------
∂25-Apr-78 1523 RSP via SU-TIP#1 EFIND
Uh, sorry. I made the erroneous assumption that the 'MIT' paragraphs
were deliberately sorted under 'Massachusetts'. I see now that the problem
has to do with the fact that the file was sorted with case of letters
being significant but was searched with the case not significant
(EFIND EXACTLY Mc IN FOO[ESS,JMC] works). I'll be in tonight to take a
look at this and see what can be done. Thank you for reporting this.
∂25-Apr-78 1916 Levin at SUMEX-AIM senior faculty mtg
Date: 25 Apr 1978 1714-PST
From: Levin at SUMEX-AIM
Subject: senior faculty mtg
To: buchanan, ghg at SAIL, jmc at SAIL, rwf at SAIL, jgh at SAIL
cc: dek at SAIL, ejm at SAIL, tar at SAIL, feigenbaum
notice of meeting
THURSDAY, APRIL 27, 5:30, Polya, Room 152.
-------
∂25-Apr-78 2041 RSP EFIND again
[As this sample from the start of page 6 of FOO[ESS,JMC] shows, the `MIT'
paragraphs are not in the correct position for the file to have been sorted
by the ASCII collating sequence, nor for them to have been so placed by human
hands under `Massachusetts'. If SSORT produced this, then it has a problem.]
*********************************************************
Machan
∂AIL Professor Tibor Machan↓Dept. of Philosophy↓
State University College↓Fredonia, N.Y. 14063∞
(716) 673-3496, 739-1075
MIT 253-1000,253-6765 machine room
MIT-ai 258-7825,7874,8295,8791,selectric 8794,15ch/sec 8795
MIT-ml 258-6744,3,2
Mariana 848-1437
Mark, Hans 965-5111
Martin, Norman U. Texas, assoc. chmn, 512 471-7316 csd, home 453-0714
*********************************************************
[Nevertheless, there is a problem with EFIND if the intention is to use it
to search files which have been sorted with case significant. If you ask
for an inexact search (EXACTLY parameter not used) then the <key string>
is converted to upper case before comparison, as is each line of the
searched file. This occurs both during the search of the E directory page
and of the body of the text. If the file has been sorted with case
significant, then problems such as the one you reported may very well
occur. It would appear that the "correct" solution would be to ask
REM to add a switch to SSORT that would request a sort without case
significance. However, although the case would be insignificant, the
change to SSORT might not be.
As a less desirable alternative, it would be possible for EFIND to broaden
its search range by the kluge of searching from a point before the all
caps version of the key to a point after the all lower case version in
the file. This would require a separate style comparison for hit
detection and for start and end detection (yes, EFIND does check for
the termination point of the search -- which may be one or more pages
after the start). Note that if every paragraph in the file begins with
an upper case letter, then an inexact search for `mc' would have to
search from the first page that might contain a paragraph beginning with
`MC' clear through to the end of the file, as `m' follows `Z' in the
collating sequence.
If the file is SUPPOSED to be sorted in a useful fashion, then it probably
SHOULD be sorted in a useful fashion -- that is, without case significance.
I'll ask REM about the possible change to SSORT and get back to you on
that. As an immediate fix, I suggest adjusting your file with E into
a form sorted without case significance.]
∂25-Apr-78 2238 RSP One more time...
∂25-Apr-78 2137 JMC efind
Now you are arguing ideology rather than simply trying to make a useful
program. If EFIND would search the entire page, it would be insensitive
to such variations. You are right that SSORT didn't produce the file
by itself. I edited the file produced by SSORT in order to make sure
that each page began with a word that wasn't all upper case in the hopes
that then EFIND would find the correct page and search it. When I discovered
that EFIND wouldn't do it anyway, I gave up. It would be better if you
would make EFIND useful rather than argue with REM.
[Oh for heaven's sake. This thing was not my idea you know. Your
original suggestion, as transmitted to me by LES, was for a program
to find paragraphs in a SORTED file as, for example, an address list.
This is what EFIND was designed to do. It appears that what you really
want is for the implicit sorting to be used only in the directory
page search. That can be done. It is a different program. It will
still suffer unless the lines in the directory page are sorted without
case significance. I have no, repeat NO, personal stake in having
this program work one way OR the other OR any way that is useful OR
not useful. I'm agreeable to producing any change that is workable
and is described to me. Please stop trying to assign ulterior
motives to my previous responses.
Now then (touchy aren't I), I gather that you don't mind arranging
for the directory page lines to be reasonably sorted. Am I also correct
in assuming that you want the search limited to one text page
regardless of whether additional hits might be found on the next page?
Should I assume that the text page is not sorted in any particular
order? In other words, exactly what is it you want the program to do?]
∂26-Apr-78 0444 REM via SU-TIP#2 Comparison of CRU2 and IC1
NEW.I[1,REM] 17.3K, using CRU2 and the default trees on [1,REM]
it crunches to 10.4K, using IC1 and no context at all it only does
slightly worse to 10.9K.
∂26-Apr-78 1019 DWW External Employment
To: feigenbaum at SUMEX-AIM
CC: DPB, BS, JMB, JEO, RWF, JLH, JGH, EJM, winograd at PARC-MAXC, ACY
CC: FFY, TAR, SSO, wiederhold at SUMEX-AIM, JMC, DEK, AZS, DAN, DWW
I feel as though there are things which must be said from
the students' standpoint which were missed or not discussed
sufficiently at the meeting yesterday. This is probably my
fault, because I didn't really know what I was doing; I was
shanghaied to fill in for Alfred at the ultimate moment. As such
I can only speak my own convictions, but I have every reason to
believe that they are shared by most of the students.
First of all, I see a genuine danger of muddling what
seem to me to be two very distinct issues. The first issue is
whether or not we should press (e.g.) XEROX to give us something
back for the wonderful intelligent people we are sending them.
As I read the feelings at the meeting, practically everyone is in
favor of this. The second issue is whether or not we press so
hard as to endanger the availability of (e.g.) XEROX as a
research facility for some of our students, both present and
future. Again as I read the feelings at the meeting, most of the
faculty is reluctant to do this, and many are outright opposed.
The straw vote which was taken asked simply whether or
not we should press (e.g.) XEROX to give us something back, and
as such tended to obscure the distinction between these two
issues. Let me propose a scenario to illustrate. As a result of
this vote of confidence, we now go to XEROX and tell them we are
really worried about this 'brain drain' and want to restrict them
to the kind of program which exists at SRI, and that if they
don't agree we will not let future students work at XEROX at all
(short of Honors co-op, ....). If XEROX says that's all right,
everything is cool; if on the other hand they say that they can't
do that, they can't afford the money or the extra time or are
worried about proprietariness or whatever, and that they guess
they don't want our students that badly, then the students have
been shafted without even coming into the discussion.
It's all well and good to say that we are a research
institute, but since we are attached to the university we do have
the additional (moral, if you like) responsibility to train
researchers. If the facilities, projects, or whatever for this
are better for some students at XEROX, I don't see how we can
justify making them unavailable. If there are those who want to have a
research institute which produces research but not researchers,
it seems to me that they are entitled to found one, but it should
not hide under the umbrella of a university, and *SHOULD* make
very clear to prospective employees what the rules of the game are.
Thanks for listening.
-David.
∂26-Apr-78 1028 RSP Getting EFIND to work
Please let me apologize for exploding like that last night. It was
unprofesional. You had every right to be upset that the program didn't
work. I suffered a serious design oversight and didn't catch it in the
testing. In any event, as far as getting something working...
There is more to this problem than meets the eye. Bear with me for a
moment while I examine two cases:
1) If page six of FOO is sorted as it is now, then an EXACT search for
"MIT" will fail. "MIT" from the key will be compared to "Mac" in the
directory line for page six and EFIND will assume that its search should
start on page FIVE! Of course, searching all of page five is of no help.
2) If in fact page six were well sorted, we still would have a problem. An
inexact search for "mac" would fail. "MAC" (upperfied for inexact search)
would be compared to "MIT" in the directory line and the search would,
once again start on page five. As before, searching the whole page isn't
the solution. It doesn't help to leave "mac" unconverted, since the
directory page search would then say that only the last text page could
possibly contain a hit.
The fundamental problem with a file sorted with case significance is
that when I am given an inexact key, I don't have the necessary information
to make resonable assumptions as to how I should upperfy it. There is also
the additional problem of how the file should be manually massaged after
it is sorted. If you have any paragraphs that begin with a lower case
letter, they will sort to the end of the file. If your file is an address
list (like FOO), then these paragraphs are probably typos and can (and
should) be manually moved back to the appropriate page. If your file is
some sort of arbitrary upper-lower case text, the chore could be quite a
nuisance.
There are a couple of ways we could go. The simplest would be to
have a file with 26 text pages, one for each letter of the alphabet. You
select the appropriate page according to the first letter of the key and
then search the selected page without regard to sorting. Of course, if
your "M" entries were numerous, you would be stuck..no splitting it
across two pages.
Or we could go with the current SSORT and make EXACT searches the default
in EFIND. There is a good deal of nuisance value here too, as I think
you'll agree.
Or we could modify SSORT to sort without case significance. EFIND
would then have no problems with inexact searches. For EXACT searches,
I think it suffices to upperfy the key during the start (and termination)
comparisons as if we were doing an inexact search. Within the range so
selected we would use the unmodified key to detect hits. The file should
require no manual mangling after the sort. [Uh, I don't believe I mentioned
that just changing SSORT would not be sufficient -- inexact searches would
work but EXACT searches would fail.]
I may have overlooked another option (my record on this one is nothing
to boast about). Let me know what you would like me to try.
I see the MIT problem, and now I agree that the next thing to do
is to modify SSORT.
∂26-Apr-78 1146 RSMITH at RUTGERS-10 QUASAR
Date: 26 Apr 1978 (Wednesday) 1449-EST
From: RSMITH at RUTGERS-10
Subject: QUASAR
To: jmc at SU-AI
Can you call me regarding QUASAR? We went to a demo
today at Middlesex County College.
(201) 932-3626 office
(201) 238-1650 home
Regards,
Bob Smith
∂26-Apr-78 1531 RSMITH at RUTGERS-10 QUASAR
Date: 26 Apr 1978 (Wednesday) 1832-EST
From: RSMITH at RUTGERS-10
Subject: QUASAR
To: jmc at SU-AI
This is just a quick note to communicate to you what
transpired at the QUASAR demo today; we will be getting
something more detailed together tomorrow.
QUASAR is indeed a transparent hoax. The president
is extremely glib and "visionary" but very short
on technical facts.
The device appears to be capable of very little locomotion.
We discovered--we think--one of the remote control people. I
think I have a picture of him. He had a microphone up his
sleeve, talking into it as Klatu "spoke". I kept close to
him, and he stopped doing it, and the robot stopped talking.
One of our graduate students asked it some standard questions
and it was unable to answer since the man with the microphone
was not using it.
We conjecture that a confederate with a large camera--who
arrived in the truck with the robot and was always within
a 10 foot distance--was operating an RC unit for the
mechanical control.
Many people in the room appeared to be enthused by the robot.
We gave several detailed statements to the press. I am not
sure how this will turn out of course, but I think that the
man from the Home News was quite convinced of the fraud. I
pointed him to the man with the microphone, and he noted that
Klatu stopped speaking after he came under scrutiny.
In a conversation after the meeting, I asked the president several
questions about the capabilities of the robot. He answered
some glibly, some with hedges, and some by refusing to reveal
proprietary information. Many of the answers were absurd and
contradictory.
He made a formal challenge to you--that
he is willing to meet you any time any where to discuss robotics.
He inplied to the press that you were unwilling to make your
charges directly toward him in an open meeting. He made less
direct charges about the "AI community" to the effect that we are
government-sponsored and do not have to show practical results. He
suggested that it was in fact to our advantage to not get practical
results.
There are several questions about what must now be done.
Possibly we must either (1) expose this hoax in a completely
believable and definitive manner, or (2) forget about it. It is
possible that QUASAR is living off of the controversy generated
by attacks by AI people. Though I hope we did a good job today,
it may nevertheless be true that we have played into their hands.
Regards,
Bob Smith
∂26-Apr-78 1538 RSMITH at RUTGERS-10 phone call
Date: 26 Apr 1978 (Wednesday) 1840-EST
From: RSMITH at RUTGERS-10
Subject: phone call
To: jmc at SU-AI
I am at my office, (201) 932-3626.
Regards,
Bob Smith
∂26-Apr-78 1620 Levin at SUMEX-AIM Artificial Intelligence Studies at SRI
Date: 26 Apr 1978 1616-PST
From: Levin at SUMEX-AIM
Subject: Artificial Intelligence Studies at SRI
To: siglunch:
COMPUTER SCIENCE DEPARTMENT
DISTINGUISHED LECTURER MINI-COURSE
ARTIFICIAL INTELLIGENCE STUDIES AT SRI
Lecture 2: Thursday, April 27, 1978, 3:45, Psychology Bldg., rm. 420-041
given by
RICHARD DUDA
Abstract:
PROSPECTOR: A Computer-Based-Consultant for Mineral Exploration
In recent years one of the major applications of artificial
intelligence techniques has been in the area of expert consultation
systems -- programs that contain a considerable body of specialized
knowledge and assist a user through interactive consultation. This
presentation will concern the techniques being used in PROSPECTOR,
the SRI system for assisting geologists in certain kinds of mineral
exploration problems. The topics to be covered include a brief
description of the process of mineral exploration, the MYCIN-like
rule-based representation of judgmental knowledge, the partitioned
semantic network representation of general geologic knowledge, and
he basic cnotrol strategy. Current work on using graphical (map)
data will be described, together with several other topics that
are promising areas for future research.
-------
∂26-Apr-78 1825 DCL
To: GROUP.DIS[VCG,DCL]:
***********************************************************************
JOINT SEMINAR
CONCURRENT SYSTEMS
and
VERIFICATION GROUP
THURSDAY 27TH APRIL
PLACE: ERL 237
TIME: 4:00 pm
TITLE: A simple yet usable concurrent system that is (hopefully) verifiable
SPEAKER: Dick Karp
The structure of a simple batch operating system is outlined. Indications
are given as to how verification might proceed. (Note: DCL will discuss
this in detail next week.) The various types of classes in the system
are examined, and an informal proof that no process waits forever for
a signal is given. Finally, some problems that still need resolving
are discussed.
∂26-Apr-78 2307 Feigenbaum at SUMEX-AIM Re: External Employment
Date: 26 Apr 1978 2254-PST
From: Feigenbaum at SUMEX-AIM
Subject: Re: External Employment
To: DWW at SU-AI
cc: DPB at SU-AI, BS at SU-AI, JMB at SU-AI, JEO at SU-AI,
cc: RWF at SU-AI, JLH at SU-AI, JGH at SU-AI, EJM at SU-AI,
cc: winograd at PARC-MAXC, ACYFFY at SU-AI, TAR at SU-AI,
cc: SSO at SU-AI, wiederhold, JMC at SU-AI, DEK at SU-AI,
cc: AZS at SU-AI, DAN at SU-AI
In response to your message sent 26 Apr 1978 1019-PST
David, thanks for the written expression of your views. To reiterate
what I think the faculty's feeling was as expressed in that vote:
1. We are a research place that trains researchers as part of the process of
doing significant research.
2. We would prefer to train researchers on Stanford research projects.
3. We would be willing to tolerate some "brain drain" if there were
appropriate quid pro quos in the form of honors coop revenue, gifts of
one sort or another, or significant teaching efforts (such as SRI has been
giving us).
I will draft a policy that says this gently (so as not to offend local
industry) but firmly.
Sincerely,
Ed Feigenbaum
-------
∂27-Apr-78 0053 REM via AMES-TIP#20
To: CRUNCH.DIS[1,REM]:
IC1.FAI/DMP[1,REM] now has optional encryption (it was easy).
∂27-Apr-78 0204 LES
To: gls at MIT-MC
CC: JMC, HVA
S-1 Lisp project
This is to confirm that we wish to offer you a temporary position (about
June 1 to Sept. 1) as a system programmer, to develop a Lisp system for
the S-1 Computer. The salary will be based on an hourly rate equivalent
to $1425 per month, based on a 40 hour week.
We will also cover your round trip travel expenses from Boston to here
provided that the sponsor (LLL) approves, as required by our idiotic
contract.
Hope you can come. Let me know when you decide.
∂27-Apr-78 0944 DWW external employment
To: feigenbaum at SUMEX-AIM
CC: DWW, DPB, BS, JMB, JEO, RWF, JLH, JGH, EJM
CC: winograd at PARC-MAXC, ACY, FFY, TAR, SSO
CC: wiederhold at SUMEX-AIM, JMC, DEK, AZS, DAN
Thanks for your response; it's gratifying to be heard.
I still wish the official policy (at least for distribution
within the department) would distinguish between the two issues
I mentioned, but perhaps that's more than I should hope for.
-David.
∂27-Apr-78 1116 WINOGRAD at PARC-MAXC Re: External Employment
Date: 27 APR 1978 1101-PST
From: WINOGRAD at PARC-MAXC
Subject: Re: External Employment
To: Feigenbaum at SUMEX-AIM, DWW at SU-AI
cc: DPB at SU-AI, BS at SU-AI, JMB at SU-AI, JEO at SU-AI,
cc: RWF at SU-AI, JLH at SU-AI, JGH at SU-AI, EJM at SU-AI,
cc: ACYFFY at SU-AI, TAR at SU-AI, SSO at SU-AI,
cc: wiederhold at SUMEX-AIM, JMC at SU-AI, DEK at SU-AI,
cc: AZS at SU-AI, DAN at SU-AI
In response to the message sent 26 Apr 1978 2254-PST from Feigenbaum@SUMEX-AIM
I wanted to comment briefly on the Wall/Feigenbaum exchange.
I basically agree with Ed's statement, but it is carefully imprecise in
some areas which deserve longer-term discussion.
We train researchers "as part of the process of doing research",
and "would prefer to train them on Stanford projects".
There are some priority issues
as to just how large a part that training plays in choosing
what we do, and how strong the preference for supporting
our own research is compared to maximizing the quality
of the learning experience for the students.
I find it a bit ironic that a discussion in which everyone was
wondering why so many students chose not to accept our
admission was followed by one in which it was clearly stated that
the quality of graduate education takes a back seat to the
carrying out of our research projects.
-------
∂27-Apr-78 1249 LLL REMINDER
To: JMC, GHG, JGH, RWF, DEK, EJM, TAR, BS
Senior faculty meeting, TODAY, (4/27), 5:30, Polya, Room 152, re: Baskett.
∂27-Apr-78 2000 JMC*
Make it $240,000 and mail to III.
∂27-Apr-78 2318 JB demonstration.
I will be happy to schedule a demonstration any time at your convenience
next week. Preferably at times when the system isn't too crowded so we
can have faster response.
I have been adding a few features, and will start work on Ramsey's soon.
How about Monday morning 10am?
∂28-Apr-78 0822 TOURETZKY at RUTGERS-10 QUASAR hoax
Date: 28 Apr 1978 (Friday) 1120-EST
From: TOURETZKY at RUTGERS-10
Subject: QUASAR hoax
To: @QUASAR.MAI[3442,245] at RUTGERS-10:
Robots and Snake Oil Salesmen:
The Quasar Controversy
David Touretzky
Rutgers University Department of Computer Science
A group of Rutgers people recently attended a
demonstration of Klatu, the "domestic android" developed by
Quasar Industries, of Rutherford, New Jersey. The speaker
was Anthony Reichelt, head of the company and principle
developer of Klatu. He was accompanied by several
assistants.
Upon arrival, Reichelt's reception was immediately
enthusiastic. His robot captured everyone's attention even
before it was turned on. Its body was a silver truncated
and inverted cone, with a translucent but obviously empty
sphere for a head. The head had eyes and ears attached.
Each eye was a cutout plastic circle, with a small white
light bulb behind a plastic lens at the center. An orange
eyelid was painted across the top of the circle. The ears,
placed 180 degrees apart at about the same level as the
eyes, were similar plastic circles, with slightly larger red
lights. There was no mouth visible; the robot's overall
facial expression could be described as wide-eyed and
innocent.
Attached to the body were two arms, actually metal
pipes covered with large accordioned gray plastic tubing;
the kind you would use to carry away exhaust from an
electric dryer. The arms can rotate a full 360 degrees at
the shoulder. There is also a joint at the elbow that does
not appear to be independently controlled: it bends as the
arm is raised or lowered. At the end of the arm is a white,
mitten-shaped manipulator, which was locked shut for the
entire demonstration. This "hand" appeared to be carved out
of wood. I am not sure whether the two halves were actually
separable or not. The clean white pincers contrasted with
the medium gray of the arms. This tended to give the
impression that the robot was wearing gloves, like Mickey
Mouse.
Reichelt ordered the robot to come to the podium, after
which he reached down below the bottom of its casing and
flipped a switch to "quiet it down." After a suitably
dramatic pause, the robot responded with a loud OUCH,
drawing laughter from the audience and a smile from
Reichelt.
! Klatu can produce a variety of interesting sounds.
First of all there is his voice. When Reichelt asked it a
question, the audio circuit would switch on, emitting a
large amount of random noise. The effect is identical to
that produced by a low quality walkie-talkie when the
channel is empty. Then the robot would speak. It
pronounced each word slowly, with more than a normal pause
between words. The voice was harsh and gravelly, as if
deliberatley distorted by some electronic device. However,
none of the individual phones were faulty. Its pitch and
stress were quite what one would expect from a mechanical
man. At one point I heard a CB operator conversing in the
background; this may have been the fault of the public
address system used for the demonstration.
In addition to speech, the robot displayed several
electronic sound effects. One sounded like a police siren,
another like a machine gun, and the third like a human's
whistling. It sometimes used these to reply to questions,
or to acknowledge the presence of a person it was introduced
to. They were also emitted when the robot was moving its
arms or body in what was supposed to be a comical manner.
These sounds gave the robot a smart-aleck air, and added to
its crowd appeal.
Finally, when asked to dance with a member of the
audience, the robot played some music as an accompaniment.
The sound was disco, with a quality slightly better than
that of the voice, but still noisy.
Reichelt described a number of intriguing applications
of his domestic android. In addition to answering the door
or the telephone, vacuuming the house, checking that
children were in bed at the proper time, and serving (but
not cooking) meals, the robot could be programmed to punish
children who misbehave, or force them to go where it wanted
by wrapping a flexible finger about their wrists. Reichelt
demonstrated this by wrapping a plastic "finger", about nine
inches long, around the wrist of a member of the audience,
and showing that she could not escape while he held the
finger closed. The "production model" android will have
three such fingers on each hand.
Reichelt also described some special application
robots. One is a "paramedic" robot, to aid the handicapped
or retarded in getting about. Another is a "sentry" robot.
This device is reportedly nine feet high, and weighs in at
1000 pounds. It can travel at a top speed of 30 miles per
hour, and make a 90 degree turn (at speed) deviating no more
than 3 degrees from the vertical! Once it detects an
intruder, the sentry robot orders him to remain motionless.
If the intruder disobeys, he is disabled with a sonic
device, and the robot remarks "I told you not to move." The
Quasar factory is supposed to be guarded by two such
!automatons.
Klatu, however, is not so specialized. He was created
simply to make people laugh. At this he is quite
successful. Reichelt notes that people see Klatu as funny
because he resembles a person. I agree with him: the robot
is definitely funny-looking. Its cute appearance and sassy
behavior will quickly put anyone at ease. Reichelt himself
projects a good deal of affection toward both the robot and
the audience. The relationship is quite similar to that of
a ventriloquist and his dummy.
It is obvious to anyone who bothers to think about it
that Klatu is remotely controlled. This was confirmed by
the Carnegie Mellon people who observed a demonstration at a
Pittsburgh department store. The Rutgers group found the
same thing. Terri Nix noticed a man in the back of the room
speaking into a microphone, and according to Bob Smith, his
lips moved in synch with the voice coming from the robot.
One or more additional people were responsible for
controlling the robot's movements and sound effects. An
individual bearing a large camera, who admitted to me that
he worked for Quasar, is suspected of controlling the sound
effects. This man took very few pictures, although he was
constantly fiddling with the controls of his camera, as well
as its extra handgrip.
No one, including Reichelt, would admit that the robot
was remotely controlled. The man with the microphone, when
confronted by Bob Smith, claimed to be communicating with a
friend on the other side of the room in an effort find out
how the robot was operated. At some point in the
demonstration he had become aware that he was being watched,
and would shake his head negatively while looking at
Reichelt to indicate that the robot couldn't talk right
then. When Bob Smith began sticking close to him, Pentax in
hand, he left the room, and the robot lost its voice for the
rest of the show. When I spoke with him later, he told me
he worked for a hang gliding magazine, and had stopped in at
the demonstration because he happened to be in the area. He
did seem to know something about hang gliding, but professed
to know nothing at all about how the robot might work.
While Bob Smith was busy shadowing the man with the
microphone, I asked Reichelt if the robot could tell jokes.
He had me come up to the podium, and told me to ask it a
question instead. "What are you made of" I asked.
Unfortunately, the robot couldn't answer, since its
spokesman had been driven from the room. Instead, it
waited, as if expecting to speak. When nothing came out, it
began to roll around, wave its arms, and produce sound
effects. "That's no answer" I told Reichelt. The robot
rolled around some more. It wasn't controlled very well.
Sometimes it would get too close to people before it
!stopped, and they would instinctively back away. As the
robot approached me (in reverse), I held my ground, and it
rolled a little too far, so that the bottom of its body came
to rest on my shoe. The temptation was too great to resist:
I leaned toward the mike and announced "It's standing on my
foot!" This brought laughter from the audience, but no
reaction from Klatu. "Would it like to get off?" I asked.
After about a 5 second delay, the robot rolled forward a few
feet. Reichelt invited me to come back after the
demonstration to ask my question again, and went back to his
speech.
We did talk to one other individual suspected of
controlling the robot. Like the man with the microphone, he
claimed to be there by chance. Supposedly he was interested
in attending Middlesex County College, where the
demonstration was held, and had come down to take a look at
the school. He had a Quasar patch sewn onto the bottom of
his pant leg. Later, one of the other Quasar people said he
thought they sold such patches at some of their
demonstrations.
After the demonstration, the robot was wheeled out to a
black Chevy van with a hang gliding emblem on the back which
read "Vol Libre" (fly free.) The head and arms were removed,
as was the cone-shaped covering on the body. What remained
was mostly covered by purple netting, but the lower section
was visible. The robot runs on three wheels; the front one
is for steering. We saw a few wires and a large
transformer, but there definitely wasn't room for the 147
subsystems Reichelt claimed it contained.
Technically, the robot's construction could be
improved. At one point it gave a little girl a ride on its
arm, after which Reichelt had to walk up and straighten the
head, which had gotten knocked out of kilter because it was
barely attached to the body. After the demonstration people
were lifting it up to see what lay below (not much was
visible.) The robot's movements were generally jerky,
although it could move in a straight line fairly well at a
speed Reichelt claimed was 1 foot per second. Its
controllers needed some practice, however. When Riechelt
told the robot to dust off a little boy (after inserting the
handle of a dust mop into a hole drilled into one of the
hands), Klatu had a lot of trouble lining himself up with
the boy and getting to the right distance. Perhaps the
controllers were afraid of hitting the child in the face
with the mop, because the robot was holding it at the level
of his head.
Since most of the people present (with the exception of
the group from Rutgers) were interested in theatrics rather
than technical quality, the presentation was an unqualified
success. Reichelt is a skilled showman, and his mix of
!comedy and science fiction is highly entertaining. While
his insistence that the robot is entirely self-contained may
annoy some members of the AI community, I would rather shrug
it off as "part of the show" than take it as a challenge to
AI research.
Reichelt did have some comments to make about AI. He
stated that a "small group of scientists on the east and
west coasts, with a few in the midwest", were claiming that
what he had done was impossible. He answered this charge by
pointing to Klatu and saying "OK, this isn't here. It
doesn't exist." He also claimed that research people are
funded to look for answers, while he, being a businessman,
only gets paid when he delivers results.
I don't see any real animosity in these remarks; it's
simply part of the hype that makes the show as good as it
is. Reichelt freely admitted that his robots do not have
artificial intellegence, because, he said, they do not think
for themselves. Afterwards, when pressed, he also admitted
that the eyes were only light/dark sensors, and that the
elaborate spoken commands he had given were only for show,
because the robot's behavior is supposedly triggered by
keywords.
Apparently Quasar's profits come from demonstrating
Klatu for a fee. The claim that they will market a domestic
android in less than two years, at a price of $3995 each, is
merely a way of generating audience interest. Since Quasar
will not accept orders, sell stock, or admit investors, the
operation is perfectly legal.
It is, however, unfortunate that the news media
cooperated in this venture by reporting everything Reichelt
has said as absolute truth. Since there is a danger that
people who haven't seen the robot will believe the Quasar
press releases, it is important that the true nature of the
Quasar product -- entertainment -- be made known. Bob Smith
likens the demonstration to that of a snake oil salesman,
with the accompanying dangers of (a) being led away from the
people who can really help, and (b) investing money in
magical elixirs that turn out to be worthless. Reichelt's
business card reads "Anthony Reichelt, Robot Master. Have
Robot, Will Travel."
The strong reaction from the AI community has led to a
more cautious coverage in the press. But even after Bob
Smith and Rick LeFaivre managed to convince some of the
local reporters that Klatu was fake, the newspaper articles
that appeared the next day were biased towards, not against,
Quasar. Apparently, objectivity is too much to ask of a
reporter with a "public interest" story.
∂28-Apr-78 1355 CET
To: JB, JMC
April 24, 1978
Juan Bulnes
JB@SAIL
Dear Juan:
At the Gray Tuesday meeting in February John McCarthy reported that you
are "doing fine."
Since you are getting close to finishing, let me remind you of the important
University deadlines which are approaching:
May 15 - Last day for filing Ph.D. Dissertation for June conferral
June 8 - Last day for oral examination in Spring Quarter
July 10-August 11 - Oral examinations in Summer Quarter
August 11 - Last day for filing Ph.D. dissertation for September conferral
You must submit a form to schedule orals. This form must be submitted to the
Department Secretary three weeks before the oral's date. At that time you
should have a draft of the important portions of your dissertation available
for the members of the oral's committee. The oral's committee is composed
of (at least) five Academic Council members. Normal procedure is that you
choose three of these. The CSD Graduate Study Committee chooses one "at
random" from our department. The person is usually chosen from outside your
field of specialization. The University Graduate Studies Office chooses a
Chairman for the committee who is not in any department already represented
on the committee.
I suggest that you formulate a detailed schedule for meeting the necessary
deadlines. I hope to see you finish this summer.
Sincerely,
Denny Brown
Associate Chairman
DPB:cet
∂28-Apr-78 1439 CET
To: PMF, JMC
Aril 28, 1978
Paul Farmwald
PMF@SAIL
Dear Mike:
John McCarthy reported that you are working hard on S-1 while you are on
leave. He suggested that you will need to focus your efforts when you begin
serious work on a dissertation. I see a potential problem developing
sine you:
1. Passed the Analysis of Algorithms Qual,
2. Have an advisor in AI and Formal Reasoning,
3. Have been working on systems design most recently.
It is current Department policy that students pass a Qual in the area of the
dissertation. One student, for example, passed the AI Qual and later was
required to take the AA Qual.
Please give some thought to this problem. You should begin by choosing an
appropriate dissertation advisor in whichever area you decide upon.
Sincerely,
Denny Brown
Associate Chairman
DPB:cet
P.S. We still must update your leave of absence form. It was originally
filed as a one-quarter leave. Talk to Carolyn to take care of this.
∂28-Apr-78 1525 Horning at PARC-MAXC Visit by Prof. F. L. Bauer
Date: 28 Apr 1978 3:19 pm (Friday)
From: Horning at PARC-MAXC
Subject: Visit by Prof. F. L. Bauer
To: Owicki at SU-AI, Knuth at SU-AI, Floyd at SU-AI, McCarthy at SU-AI
cc: Horning
I have received the following letter from Fritz Bauer:
Between the Atlanta Conference on Software Engineering, May 10-12, 1978, and
the Los Angeles Conference on the History of Programming Languages, June
1-3, 1978, I shall be in California....
I plan to stay after Atlanta a few days at Alston Householder's home, otherwise I
am completely free and could arrange my visit as it is convenient for you.
Apart from seeing friends at Stanford, I shall have to meet Susan Owicki, too,
and would also like to see Zilles. ... Maybe you can check with Don Knuth, Bob
Floyd or John McCarthy about their wishes.
My contact address after May 10 will be
c/o A. S. Householder
6235 Tapia Drive
Malibu, Calif. 90265
Phone (213) 457-3547
I would like to discuss ... questions of recent development in programming
languages and techniques of program development.
F. L. Bauer
---------------------------------------
I will be writing him shortly. I have several other visitors who are working
with roughly the same constraints, and I imagine that you do, too. If you have
particular preferences about the timing of his visit to Stanford, please let me
know--you might also wish to contact him directly.
Jim H.
∂28-Apr-78 1707 PAT
I couldn't find manoug.le1 anywhere - patte
∂28-Apr-78 1847 GLS,GJS at MIT-AI History of LISP Paper
Date: 28 APR 1978 2147-EST
From: GLS,GJS at MIT-AI
Sent-by: GLS at MIT-AI
Subject: History of LISP Paper
To: JMC at SU-AI
We got a copy of your "History of LISP" paper, and noticed
a remark that D.M.R.Park pointed out that LABEL was logically
unnecessary, because LAMBDA could be used. Is this actually
true of the LISP at that time, with its dynamically scoped
variables? Or did it depend in some way on the indefinite
evaluation of the function position? We would be interested
in knowing the precise technique, if you have it at hand.
If you look at my 1960 paper "Recursive functions of symbolic
expressions ...", you will see what I knew at the time.
I don't believe there was an implementation or even that one was
worked out.
∂28-Apr-78 1905 GLS at MIT-AI (Guy L. Steele, Jr.)
Date: 28 APR 1978 2205-EST
From: GLS at MIT-AI (Guy L. Steele, Jr.)
To: jmc at SU-AI
thank you.
∂29-Apr-78 1800 BCM xoroshiye izvectiya
ya cdal ekzamen po sistemam.
izdanie knig po politicheskim shutkam sostoit uz dvux tomov.
eto stoit $6.00 + $.50 (dlya pochti). tom na odnom yazike stoit $3.50.
knigi nazivayutcya <<Forbidden Laughter (Soviet Underground Jokes)>>
avtor: Emil Draitser
bot adress izdatel'stva:
Almanac Press
P.O. Box 480264
L. A., CA 90048
∂29-Apr-78 2145 LLW The Solar Surrey and Sun Day
To: JMC
CC: LLW
As you must have expected, John, I'm in at least basic (and mostly
detailed) agreement with all the major points you made in your Sun Day
manifesto. I thoroughly agree with the appropriateness of your taking a
public stance on the matter--if I weren't embroiled in so many
controversies already, I would pitch in myself, particularly since I also
suspect the anti-nuclear motives of the Sun Crowd.
While I speak out of the depths of nearly total ignorance on the
particular matter, I suggest that you may be weakening your powerful
general agruments by the attack on the Solar Surrey, which seems to me to
be a totally insignificant aspect of the 'sun power now' push. I pedal by
this contraption several times per day, and it's rather ludicrous on its
face. It sits in the shade much of the morning and afternoon, because of
the carelessness of the people who are in charge of it, and I've never
seen it (or heard of its being) in use. It carries a single 12 volt car
battery (in view, at least) and has the technical sophistication of a
venerable golf cart. It's a laughingstock of the LLL technological
community, and no one seems at all certain how we got mixed up in this
obviously misbegotten JPL scheme.
It seems to me that using the finite space your have in your manifesto
(and the limited attention span of your audience) to draw attention to the
two-order-of-magnitude cost gap that must be spanned before solar
photovoltaic power can possibly become economically competitive would be
preferable to kicking such a pathetic little mistake as the Solar Surrey.
In general, though, more power (and/or energy) to you!
Lowell
∂30-Apr-78 2041 JB DEMONSTRATION.
Monday (tomorrow) morning at 10 AM is fine.
(Sorry that my answer comes so late. Friday I was
moving, so I just received your message today.)
∂01-May-78 0000 JMC*
Make reservation for HOPL - see Lisp history file.
∂01-May-78 0845 CET
To: HPM, JMC
April 24, 1978
Hans Moravec
HPM@SAIL
Dear Hans:
At the Gray Tuesday Meeting in February John McCarthy reported that you are
making good progress and should finish this year.
Since you are getting close to finishing, let me remind you of the important
University deadlines which are approaching:
May 15 - Last day for filing Ph.D. Dissertation for June conferral
June 8 - Last day for oral examination in Spring Quarter
July 10-August 11 - Oral examinations in Summer Quarter
August 11 - Last day for filing Ph.D. dissertation for September conferral
You must submit a form to schedule orals. This form must be submitted to the
Department Secretary three weeks before the oral's date. At that time you
should have a draft of the important portions of your dissertation available
for the members of the oral's committee. The oral's committee is composed
of (at least) five Academic Council members. Normal procedure is that you
choose three of these. The CSD Graduate Study Committee chooses one "at
random" from our department. The person is usually chosen from outside your
field of specialization. The University Graduate Studies Office chooses a
Committee Chairman, who is not in any department already represented
on the committee.
I suggest that you formulate a detailed schedule for meeting the necessary
deadlines. I hope to see you finish this summer.
Sincerely,
Denny Brown
Associate Chairman
DPB:cet
P.S. We are still awaiting a note from John McCarthy or Tom Binfod to
clear the conditional on your Qualifying Examination.
∂01-May-78 0846 CET
To: PMF, JMC
Aril 28, 1978
Paul Farmwald
PMF@SAIL
Dear Mike:
John McCarthy reported that you are working hard on S-1 while you are on
leave. He suggested that you will need to focus your efforts when you begin
serious work on a dissertation. I see a potential problem developing
sine you:
1. Passed the Analysis of Algorithms Qual,
2. Have an advisor in AI and Formal Reasoning,
3. Have been working on systems design most recently.
It is current Department policy that students pass a Qual in the area of the
dissertation. One student, for example, passed the AI Qual and later was
required to take the AA Qual.
Please give some thought to this problem. You should begin by choosing an
appropriate dissertation advisor in whichever area you decide upon.
Sincerely,
Denny Brown
Associate Chairman
DPB:cet
P.S. We still must update your leave of absence form. It was originally
filed as a one-quarter leave. Talk to Carolyn to take care of this.
∂01-May-78 0857 CET
To: REP, JMC
Results of Gray Tuesday
April 27, 1978
Mr. Richard Pattis
REP@SAIL
Dear Rich:
At the Gray Tuesday meeting on 14 February the faculty evaluated the
progress of each of its Ph.D. students. You are apparently doing well
although you don't seem to be in close contact with your advisor. We
encourage you to talk with your advisor on a regular basis. You may
choose a new advisor at any time.
The next step is to pass a Qualifying examination. The Qualifying exami-
nation must be passed during the first 3 years as a Ph.D. student.
Keep up the good work.
Sincerely,
Denny Brown
Associate Chairman
/cet
∂01-May-78 1042 CET
To: DEW, JMC
Results of Gray Tuesday
April 27, 1978
Mr. David Wilkins
DEW@SAIL
Dear David:
At the Gray Tuesday meeting on 14 February the faculty evaluated the
progress of each of its Ph.D. students. You were judged to be making good
progress.
Congratulations! Keep up the good work.
Sincerely,
Denny Brown
Associate Chairman
/cet
∂01-May-78 1043 CET
To: REF, JMC
Results of Gray Tuesday
April 27, 1978
Mr. Robert Filman
REF@SAIL
Dear Bob:
At the Gray Tuesday meeting on 14 February the faculty evaluated the
progress of each of its Ph.D. students. You were judged to be making good
progress.
Congratulations! Keep up the good work.
Sincerely,
Denny Brown
Associate Chairman
/cet
∂01-May-78 1207 DPB Binford as Principal Dissertation Advisor
To: TOB.DIS[1,DPB]:
At the faculty meeting in April it was decided that we would
deal with "non-academic-council-members as principal dissertation advisors"
on a case-by-case basis. The mechanism decided upon was:
1.) The faculty member who otherwise would supervise asks the Chairman (EAF)
to request approval from the Graduate Program Office.
2.) If the Chairman agrees with the proposal on its merits, appropriate
documentation is assembled. (This will include at least a vita, and a
statement about the inability of the regular CS faculty to supervise the
particular work. Details of the documentation to be determined by the
Graduate Program Office.)
3.) The request is presented to the Graduate Dean for approval.
Three students, Arnold (RDA), Gennery (DBG), and Goldman (ARG), are
currently waiting for a decision on their requests to have Tom Binford
serve as official advisor. Each has submitted the Reading Committee
form with Binford as PDA. I hope we can come to a quick resolution
of these cases. First move is JMC's.
-Denny
P.S. SAIL distribution list is TOB.DIS[1,dpb].
Is the following message to Ed Feigenbaum factually correct. Are there
details that should be added?
Ed:
This is a proposal that Tom Binford be asked to serve as official
thesis adviser to Richard Arnold, Don Gennery and Ron Goldman.
Their theses are in the area of computer vision in which Tom's
competence has increased as mine has declined. They are all working
on projects in which Tom is the leader and effective principal
investigator. As you remember, we are hoping to regularize the
situation by getting Tom appointed Adjunct Professor in either our
department or Mechanical Engineering in connection with the proposed
Institute of Automation.
∂01-May-78 1552 DCL
To: GROUP.DIS[VCG,DCL]:
***********************************************************************
JOINT SEMINAR
CONCURRENT SYSTEMS
and
VERIFICATION GROUP
TUESDAY 2ND MAY
PLACE: ERL 237
TIME: 2:30 pm
TITLE: A weak theory of transfer processes
SPEAKER: David Luckham
Transfer processes are concurrent processes that read
data from input channels and write the results of their computations
on output channels; input and output channels are always distinct,
and the computations are very simple. We present a weak theory in which
only a few of the required properties of systems of transfer processes
can be proved. The theory proposes an axiomatic semantics for Cycle
statements and Processes.
It is hoped that this theory will permit proof of
some of the obvious requirements of RAKs operating system.
∂01-May-78 2312 MRC QUASAR and the censorship of MsgGroup
To: MsgGroup at USC-ISI
CC: MRC, JMC, LES
That is an interesting point. Prudence would counsel us to
take the self-censorship approach that Stef and Dave have
proposed. I do not feel, however, that prudence in this case is
warranted.
It is most unfortunate that people cannot speak as
individuals, or that we, as scientists, are unable to communicate
with fellow scientists freely about a matter which concerns us.
In QUASAR's case, we are referring to what certain members of the
computer science communitity feel is a hoax being perpretrated
upon the public. The charge is indeed a serious one, and if true
it means that the public is being subjected to a deception which
besides making money for QUASAR is discrediting the work of the
computer science community. Suppose Senator Proxmire were to
demand a cutoff for ARPA-sponsored AI research on the grounds
that QUASAR's "developments" have made AI research unnecessary?
I agree a lawsuit is no joking matter. However, restraining
the freedom to question the veracity of reports of alleged
developments in any field is preventing us from performing our
duties as scientists. Are we to allow any charlatan who can say
"lawsuit" and "lawyers" to run roughshod over scientific inquiry
and freedom? I am aware of one ARPAnet site which is right now
being literally blackmailed by threats of a lawsuit regarding the
rights to the source of their proprietary software (the claim is
that since it was developed partially with ARPA funding, it is
"public domain" and whomever wants a copy for commercial purposes
can have one). They submit to this extortion, not because it is
reasonable, but because there are no "legal precedents" and don't
want to take the risk, however unlikely, of losing.
I propose that it should be officially noted that the
members of MsgGroup are speaking as individuals in a candid
fashion to the other members of the group, and are speaking
neither for their organizations, or ARPA, or MsgGroup or any
other individual in MsgGroup.
This is the only way I can see that can possibly work. I
think that censoring MsgGroup because of QUASAR would be a
heinous crime against scientific freedom. We might as well be in
the USSR or the PRC if we must kowtow to charlatans and frauds
like QUASAR. There. I've said it. If QUASAR feels they've been
irrepairably damaged or libeled by that, let them sue me for the
epsilon assets I possess. If they do, I will dedicate the rest
of my life to the utter ruin of QUASAR and anybody associated
with them.
-- Mark
[PS. Excuse the spelling. It's been a long day...]
I have received a copy of MRC's message without knowing what
preceded it. However, I can say that neither Minsky nor Brian Reid
nor I have been deterred by speculation that Quasar might sue us.
We have stated our opinion that they are fraudulent to any newsmen
that asked (I have talked to more than 20), issued a press release,
and written a joint letter to the Justice Department and the Postal
Inspector. I think someone seems to be frightened of his shadow,
because I know of know case in which a scientist has been sued for
expressing his opinion that something is fraudulent. When Reichelt
gave his spiel at some community college in New Jersey he denounced
me by name according to Bob Smith, but did not threaten suit. According
to Smith, as with the three other groups of eyewitnesses, the fraud
is transparent enough so that Reichelt has no chance of successfully
suing anyone. Moreover, it has never been the custom of carnival
snake oil salesmen to sue their critics. So have no fear.
∂02-May-78 0029 STEFFERUD at USC-ISI Re: QUASAR and the censorship of MsgGroup
Date: 2 MAY 1978 0020-PDT
Sender: STEFFERUD at USC-ISI
Subject: Re: QUASAR and the censorship of MsgGroup
From: STEFFERUD at USC-ISI
To: MRC at SU-AI
Cc: Farber at KL, JMC at SU-AI, LES at SU-AI, Stefferud
Message-ID: <[USC-ISI] 2-MAY-78 00:20:01.STEFFERUD>
In-Reply-To: Your message of MAY 1, 1978
Hi Mark -
Your missive gives me trouble, of the very kind I asked for help
with. Sorry bout that, but you are making the situation much
more difficult when I need help in sorting out the issues.
Clearly I cannot stop you from mailing anything you want to
anybody you wish to. But, I cannot mail things for you that I
find expose me to legal actions I am not interested in being
exposed to, or which I cannot directly support. I trust you can
see the difference clearly. I am distinguishing between your
right to say what you want and my right not to publish it, so to
speak.
Thus, I will not forward your peice to which this is the reply,
knowing that you know how to get the mailing list and do it
yourself. I will not and cannot stop you. What I will not do,
and can avoid, is to become a "cosigner of sorts to your willing
effort to incur a lawsuit.
From all I have read, I agree that the Quasar Robot appears to
be a fake and that in fact, if they are dishonest, they could
actually commit real criminal fraud. However, I have no
evidence of dishonesty or of criminal fraud in this case.
I would submit that many people are capable of such fraud with
the same ease that the quasar folk could, but you are not
interested in the lot of them. The distinction is quite
important, and I am asking that we disuss that issue before
charging off after quasar with no holds barred, or with no
further questions of propriety asked.
Cheers - Stef
∂02-May-78 0111 JB MEETING
O.K., not tomorrow. Do you want to re-schedule? I'm working on RAMSEY now.
∂02-May-78 0312 MRC
To: Stefferud at USC-ISI
CC: Farber at SRI-KL, JMC, LES
You have my permission to send my message without the last
paragraph, which is the only one that can be construed as
libeling QUASAR (note that the rest carefully avoids it).
I wish it could be sent with the last paragraph...oh well.
Is that alright? -- Mark
∂02-May-78 1027 STEFFERUD at USC-ISI
Date: 2 MAY 1978 1027-PDT
Sender: STEFFERUD at USC-ISI
From: STEFFERUD at USC-ISI
To: MRC at SU-AI
Cc: Stefferud, Farber at SRI-KL, JMC at SU-AI, LES at SU-AI
Message-ID: <[USC-ISI] 2-MAY-78 10:27:02.STEFFERUD>
In-Reply-To: Your message of MAY 2, 1978
I see you are beginning to see my problem. Your polemic really
∂02-May-78 1039 Nii at SUMEX-AIM Siglunch 5/3
Date: 2 May 1978 1039-PDT
From: Nii at SUMEX-AIM
Subject: Siglunch 5/3
To: siglunch:
Due to various talks at other places, there will be no Siglunch this week.
Penny
-------
∂02-May-78 1041 JBR Music compiler
To: RWW
CC: JMC, LES
Whereas I believe you have a perfect right to show up the musicians and prove
how wonderful and elegant LISP is for everything, I don't think it is reasonable
for your group to be using up any prime computer time to do so. The load average
is now 10 partially do to AMR running the LISP music compiler. I strongly suggest
that you keep this an extra-curricular project. The music group is doing just
fine monopolizing all of the system's resources without any help from the FOL
group. If you must burn up computer cycles, why don't you prove some mathematics?
∂02-May-78 1124 Nii at SUMEX-AIM Siglunch Notices
Date: 2 May 1978 1124-PDT
From: Nii at SUMEX-AIM
Subject: Siglunch Notices
To: Siglunch:
Our Siglunch distribution list has become exceedingly long, and we would like
to shorten it. If you are still interested in being notified of Siglunch
meetings, please let me know by Friday, 5/5.
No reply will be assumed to mean that your name can be deleted.
Penny
-------
I remain interested in SIGLUNCH.
∂02-May-78 1145 Rindfleisch at SUMEX-AIM NO HACKER'S LUNCH THIS MONTH
Date: 2 May 1978 1135-PDT
From: Rindfleisch at SUMEX-AIM
Subject: NO HACKER'S LUNCH THIS MONTH
To: siglunch:
-------
∂02-May-78 1157 FARBER at USC-ISI
Date: 2 MAY 1978 1155-PDT
Sender: FARBER at USC-ISI
From: FARBER at USC-ISI
To: JMC at SU-AI
Cc: MsgGroup, MRC at SU-AI, LES at SU-AI
Message-ID: <[USC-ISI] 2-MAY-78 11:55:08.FARBER>
In-Reply-To: Your message of MAY 2, 1978
John,
I have no fear of being sued by someone who does not like me to
say the truth about them. However, we are using a public vehicle
called the ARPANET . We therebye expose ARPA, DOD and our future
access and use of that network to certain dangers when we use
that vehicle for potential libelous material ( and I mean that in
a legal way). I suggest that we reserve our interactions
relative to Quesar on the net to the reporting of uninterpreted
facts. That way we serve the purpose of informing people without
the danger of cutting our own throat.
Dave
∂02-May-78 1212 MRC lossage with Quasar
To: Farber at USC-ISI, Stefferud at USC-ISI
CC: JMC
Well, I think I have made my point in my opinions of Quasar, and that
I feel the merits or non-merits of Quasar should be discussed as
openly without fear as, say, whether or not the earth is round. If
there is any reason not to discuss Quasar, it should be for the same
reason why there aren't any discussions about whether or not the
earth is round; there is seemingly universal agreement within the
ARPA community about Quasar, making discussing it at least moot and
at worst boring (which at times it does get).
Anyway, if you object to the last paragraph (I realized fully when I
wrote it that it was anything BUT self-censoring irt Quasar), I would
rather see the rest of the message distributed (as a "revision") than
sticking to the last paragraph and not going out. I don't have an
online copy of the message any more; would you edit out the last
paragraph and send it out? I carefully worded the rest to attempt
NOT to say anything which could be construed as libeling Quasar
(except in possibly the vaguest sense).
-- Mark
∂02-May-78 1825 EJM additional accounts
To: JMC
CC: LES
As per our discussion last week I would like to request
authorization of SAIL accounts for some of my associates.
Specifically I would like the following to have accounts:
Danny Beaudry, research associate;Jacob Savir, post-doc;
David Lu, RA;Peter Fu,RA;Peter Thompson,RA;Marion Blount,RA;
and Bezhad Khodad,RA. I don't expect any
significant computing to be done at SAIL by any of them but
I would like them to have access to mail and files. If file
space is an obstacle I would be willing to share my
allocation with them.
∂02-May-78 2306 DCL Group Allocations
To: JMC
CC: LES, JC
John, I heartily congratulate you on this proposal and urge that
a group allocation scheme be implemented ASAP.
In the face of recent load averages of 6 at 4 am, 8-10 at 8 am, 10-15
at 10 pm, I have found my group beginning to quit work entirely both
on system development and experiments. We still hope to have a verifier
ready for distribution as an introduction to our next ARPA proposal.
Group allocation is probably the only way we can accomplish this.
∂02-May-78 2335 STEFFERUD at USC-ISI
Date: 2 MAY 1978 2330-PDT
Sender: STEFFERUD at USC-ISI
From: STEFFERUD at USC-ISI
To: MRC at SU-AI
Cc: Stefferud, Farber at SRI-KL, JMC at SU-AI, LES at SU-AI
Message-ID: <[USC-ISI] 2-MAY-78 23:30:14.STEFFERUD>
In-Reply-To: Your message of MAY 2, 1978
HI Mark -
Unfortunately my copy was also truncated. Perhaps it was not
meant to be sent???
What I said was that I would prefer that you send it to the
MsgGroup list yourself, even with the last paragrph removed.
There is no prohibition against your using the list. My
problems stem from the psuedo publishing, public broadcasting
nature of my situation. Your use of the list is not the same.
If you are not willing to send directly, why should I
redistribute it?
I will return a copy to you to edit as you would prefer.
OK? Stef
PS: Please note that I am not censoring, nor do I appreciate
claims that I am. I merely raised the issue for consideration,
and you in return pressed me directly with a severe test. My
reaction is quite natural, and very clear.
Do it your self my friend.
Best - Stef
∂03-May-78 0134 Gaschnig at SRI-KL Re: MSGGROUP# 678 Comment on Quasar Discussion Question
Date: 3 May 1978 0115-PDT
From: Gaschnig at SRI-KL
Subject: Re: MSGGROUP# 678 Comment on Quasar Discussion Question
To: JMC at SU-AI
cc: GASCHNIG
In response to your message sent 2 May 1978 0310-PDT
Here, here! (but the general issue remains valid, in conceivable circumstances
not concerning Quasar)
-------
∂03-May-78 0146 MRC
To: JMC, LES
∂03-May-78 0104 RAH via UTAH-TIP#27 Dialnet
Hiya,
You don't know me (I was one of the Evergreen State College
students from Olympia, WA who came to visit the lab
in late Jan., if that helps)
anyway...
I was refered to you by BH who mentioned that there might be
a chance of a job at SAIL: this summer working in conjunction with
you and JMC on the DIALNET project.
Ever since discovering the Arpanet in Jan. I have been facinated
by the possabilities inherent in public networks and would like to
become involved in their development. Hence, the DIALNET project
is particularly interesting to me.
I would be very interested in knowing if there IS a job opening
and also in any information you might have on DIALNET's progress
, philosophy, etc.
Thanks
Rick
mrc - I dunno, is there? I think we have enough people now. Maybe
if he's a pdp11 wizard or something. I don't know anything about him.
∂03-May-78 0547 MRC Dialnet programming documentation
To: Dialnet
CC: DGR, GFF, BH
The file:
DLNSER.DOC[DCP,MRC]
at ARPAnet host SU-AI contains the specification for user-mode
programming for Dialnet on the SU-AI WAITS time-sharing system.
It describes all the UUO's and magic modes of I/O. Code already
exists for most of the features of all the UUO's documented in
this file.
Anybody who is attempting to write Dialnet software for WAITS
MUST understand the contents of the file, and everybody should
read it. This file will be updated from time to time, so anybody
writing code should always look at the latest version, or better
yet, talk to me personally.
-- Mark
∂03-May-78 0706 100 : DAVID SHIELDS via NYU DEC10 ACCOUNT
Dear Professor McCartyyλλhy:
I am on the staff at NYU. When Jack Schwartz was here last weekeλ< i asked him to look into the possibility_↔↔↔↔
∂03-May-78 0920 Tajnai at SUMEX-AIM ( Forwarded Mail )
Date: 3 May 1978 0921-PDT
From: Tajnai at SUMEX-AIM
Subject: ( Forwarded Mail )
To: JMC at SAIL
cc: DPB at SAIL, Tajnai
1 30 Mar To: AMR at SAIL C.S. Comprehensive
2 3 May To: AMR at SAIL Student Status
1 -- ************************
Date: 30 Mar 1978 1014-PST
From: Tajnai
Subject: C.S. Comprehensive
To: AMR at SAIL
cc: DPB at SAIL, Tajnai
Andrew, in checking records, I noticed that you are a second year
Ph.D. student, and you have not passed the C.S. Comprehensive.
It is required that you pass it in the first two years of your
candidacy. It will be given again in May. Notices will be posted
on the AI and Sumex bulletin boards, plus physical bulletin boards.
I will also make a mental note to remind you when it is first announced.
Carolyn
-------
2 -- ************************
Date: 3 May 1978 0913-PDT
From: Tajnai
Subject: Student Status
To: AMR at SAIL
cc: DPB at SAIL, Tajnai
Andrew, Denny told me that you are not planning to return to Stanford
autumn quarter 1978-79. Would you please send me a message to that
effect to put in your file.
Good luck,
Carolyn
-------
John, Andrew Robinson is your advisee. I wanted to make sure that you
were aware of the situation. Carolyn
-------
∂03-May-78 0946 JLH Pascal compiler
To: PASCAL.DIS[1,JLH]:
The current Pascal compiler is very fast (1 pass).
Since the compiler is fast, for programs on the order of 300-500
lines the loader takes longer than the compiler.
This would a make an incore compiler a real advantage.
It does however, require a significant amount of work.
(As does a microbatch feature).
∂03-May-78 1043 HVA Vents and Cold Air - your office
I've reported problem to Plant Services - will keep trying to rectify.
∂03-May-78 1324 MRC Comments?
To: JMC, LES
∂03-May-78 1200 DCROCKER at USC-ISI Politics vs. Legalities
Date: 3 MAY 1978 1050-PDT
Sender: DCROCKER at USC-ISI
Subject: Politics vs. Legalities
From: DCROCKER at USC-ISI
To: [ISI]<MsgGroup>Mailing.List;153:
Message-ID: <[USC-ISI] 3-MAY-78 10:50:36.DCROCKER>
Mark: Please don't forward me any more messages like the latest
from DCROCKER. The world is full of pompous young bureaucrats
and would-be bureaucrats,
expert at thinking of reasons for not doing something, and it
is best to ignore them unless you really need access to some
resource they control.
In the early days of the ArpaNet, I once investigated the
possibilities of obtaining funding to establish a preliminary
"Community Resources Information" network, using SRI's NLS system
to organize textual data about groups offering assorted community
services and using the ArpaNet to link geographically separated
communities. The project was to start small and allow graceful
evolution; participants would merely need to gain access to the
ArpaNet.
My investigations were cut short when it was pointed out to me
that Arpa affilliation with such an effort (it was, after all,
their network) was a prime candidate for some news sensationalist
to begin shouting that "The Defense Departments Supports Abortion
Acitivies".
It was not important then, or now, whether such publicity is
accurate; it IS important that such publicity tends to have
disastrous effects to which there is no recourse. Faced with
such problems, instituional decision-makers tend to be very
conservative; justice and truth tend to be irrelevant. The
solution to this sort of problem usually is to avoid puting the
decision-maker in the position of having to make a choice in
which you can lose.
For example, erroneous publicity might develop, claiming that
various subversives at sites around the country were using
Department of Defense communications systems for the purpose of
suppressing free enterprise. A consequent of such adverse
publicity might be an edict that all ArpaNet mail had to pertain
directly to the activities for which the participant was funded,
and had better be justifiable as such in the future.
The point is not that we should not discuss such things, but
rather that we should discuss them with the understanding that
our statements, through distribution lists such as MsgGroup, are
in the public domain and that statements made within such a group
are usually attributed to the group and not merely to the
individual making the comments.
Dave.
∂03-May-78 1534 PAT
∂20-Apr-78 1016 CET
Memo to Professors McCarthy and Binford, Hans Moravec
copy to Denny Brown
Subject: Conditional on Qualifying Exam.
Ref: Letter to Mr. Moravec from Prof. Floyd daed 7/1/75
"This letter is to inform you that your AI Qual.Exam. committee has been
unable to determine a grade for you based on your oral presentation on
June 10, 1975.
It is the recommendation of this committee that you continue this exam
during the summer quarter 1974-75. Specifically, the committee found that
you must prepare a survey of the "state of the science" with broad coverage,
but also with specific emphasis on perception research.
It is up to you to contact the memers of your committee to set up a new time
and place. They ae: Prof. Feigenbaum, L. Quam and R. Duda or T. Binford."
Hans said he has completed this work. I need a statement in his folder
confirming this. Please advise.
Patte, will you please help me on this. I have talkd to Hans and Tom
to no avail. The records should be updated. Thanks, Carolyn
JMC - what should be done about this??? - patte
∂03-May-78 1546 DCL Group Allocation
To: JMC, LES, JC
I understand that JC is going to police the music group to use only 20%
dubloons per month, and if he is unsuccessful the JMC will enact an
administrative procedure based on monthly allocations and LES's Bureaucracy
program. I am very much in favor of the latter step being taken immediately.
If the JC gentleman's agreement is to be given a chance first, I strongly
urge a bound of 15%. 20% is too little too late.
∂03-May-78 1541 LCP Pascal for CS10x
To: PASCAL.DIS[1,JLH]:
The original Hamburg implementation produced core image files (with extensions
.LOW and .SHR, as they were called in those days - what is now .EXE), skipping
the loading phase. This compiler was also smaller and faster. Unfortunately,
due to the greater flexibility of .REL files, this compiler was converted to
produce .REL files instead, and evolved into the compiler we have now. A copy
of the old compiler is still available, but it is not supported (and therefore
buggy), does not provide the SIN, COS etc. functions or work with the interactive
debugger (due to the lack of a loading phase) and does not conform to the latest
Pascal standard (e.g. for the program heading). If someone wants to do a lot of
work...
Larry Paulson
∂03-May-78 1613 DCL
To: GROUP.DIS[VCG,DCL]:
***********************************************************************
JOINT SEMINAR
CONCURRENT SYSTEMS
and
VERIFICATION GROUP
THURSDAY 4TH MAY
PLACE: ERL 237
TIME: 4:00 pm
TITLE: A weak theory of transfer processes
Part 2
SPEAKER: David Luckham
Continuing:
How the theory can be used to prove some properties of processes and
systems of processes.
OLD ABSTRACT
Transfer processes are concurrent processes that read
data from input channels and write the results of their computations
on output channels; input and output channels are always distinct,
and the computations are very simple. We present a weak theory in which
only a few of the required properties of systems of transfer processes
can be proved. The theory proposes an axiomatic semantics for Cycle
statements and Processes.
It is hoped that this theory will permit proof of
some of the obvious requirements of RAKs operating system.
∂03-May-78 1646 JBR
squish.e[ess,jmc] tells you how to compress your mail file
∂03-May-78 2048 DCL Testing your Group allocation scheme
John, I think your group allocation scheme should be tested. The actual
cutoffs do not need to be made. You can have Hersche keep the statistics
and a list of who would have been cut off when, and compare the results
with our actual world. It would be a nice experiment and might yield
some interesting results. Maybe the test should not be generally known
inorder to yield the best results.
Perhaps unfortunately, perhaps not, my interest in optimal administration
is too limited for that.
∂04-May-78 0241 STEFFERUD at USC-ISI Re: Your msg of 4 May 1978 0048-pst
Date: 4 MAY 1978 0233-PDT
Sender: STEFFERUD at USC-ISI
Subject: Re: Your msg of 4 May 1978 0048-pst
Subject: [STEF at SRI-KA: SIDE DISCUSSION RE QUASAR & DEC & MSGGROUP...]
From: STEFFERUD at USC-ISI
To: JMC at SU-AI
Cc: stefferud
Message-ID: <[USC-ISI] 4-MAY-78 02:33:38.STEFFERUD>
Thank you John for your appology. I was really having trouble
correlating what you said with what I know about you.
The following is offered to let you see that I was really about
to do something as foolish as you did. I will not send this out
as intended.
Among other things, I am not a coward in any sense of the word.
FRom your reactions, I was beginning to believe that it was a
crime to question the proriety and ask for discussion of the
issues. I know that what I did was right, and I really wonder
why you guys acted so threatened? Clearly a reduction of the
polemic volume on QUASAR in MSGGROUPP can have no effect on you
all unless something is very wrong someplace.
Best - Stef
Begin forwarded message
Mail from SRI-KA rcvd at 4-MAY-78 0215-PDT
Date: 4 May 1978 0215-PDT
From: STEF at SRI-KA
To: MsgGroup at ISI
Subject: MSGGROUP# 691 SIDE DISCUSSION RE QUASAR & DEC & MSGGROUP
[PBARAN at USC-ISI: DEC Pitch]
[STEF at SRI-KA: Re: DEC Pitch]
[PBARAN at USC-ISI: Re: DEC Pitch]
[STEF at SRI-KA: Re: DEC Pitch]
Message-ID: <[SRI-KA] 4-May-78 02:15:09.STEF>
Sender: STEF at SRI-KA
Text:
Begin forwarded messages
Mail from USC-ISI rcvd at 2-May-78 2134-PDT
Date: 2 MAY 1978 2133-PDT
From: PBARAN at USC-ISI
To: stef at SRI-KL
Subject: DEC Pitch
Redistributed-To: Stefferud at ISI
Redistributed-By: STEF
Redistributed-Date: 2 May 1978
Text:
The recent msggrp discussions about the
propriety of using govt/ARPA resources
to discuss the QUASAR thing look
like small potatoes when compared to
the recent sales pitch from THUREK@DEC MARLBORO to an immense mailing list.
Dave Caulkins
(mailbox PBARAN@ISI)
-------
--------------------
Mail from SRI-KA rcvd at 2-May-78 2254-PDT
Date: 2 May 1978 2254-PDT
From: STEF at SRI-KA
To: PBARAN at USC-ISI
Cc: stef, Farber at SRI-KL
Subject: Re: DEC Pitch
In-Reply-To: Your message of May 2, 1978
Message-ID: <[SRI-KA] 2-May-78 22:54:15.STEF>
Sender: STEF at SRI-KA
Redistributed-To: stefferud at ISI
Redistributed-By: STEF
Redistributed-Date: 2 May 1978
Text:
Hi Dave -
I can imagine that there are a bunch of very questionable things going
on that make MsgGroup look like a bunch of the finest guys around.
My problems are more subtle. The Quasar discussion had been getting
more and more polemic in nature, and way off the subject of MsgGroup.
I wanted to stop the polemics if I am a psuedo publisher, and I wanted
to get the issues of such "public discussions" on the table for
autopsy. Nothing like having a real corpse if you want to have an
autopsy, I always say.
The responses have been quite interesting. JMC has virtually called
me a coward, and the whole thing is quite hot. I intend to sit an
watch for a bit to see who really jumps out with good ideas. Some of
the stuff is quite thoughtful.
By the way, I thought after sending your note to me out to MsgGroup
that maybe you had not meant it for redistribution. I should have
asked. Did I goof? Your use of this box reinforces my analysis. Not
that I think it was so terrible, but I will ask in future cases of
that kind, just to be sure.
Thaks for your comments - I find them helpful.
Cheers - Stef
--------------------
Mail from USC-ISI rcvd at 4-May-78 0135-PDT
Date: 3 MAY 1978 2235-PDT
From: PBARAN at USC-ISI
To: STEF at SRI-KA
Cc: PBARAN
Subject: Re: DEC Pitch
Text:
In response to your message sent 2 May 1978 2254-PDT
I think you play an important role as
the msggrp 'electronic publisher'. I
direct things to you with full permission
to redistribute, extract, edit, or
ignore.
It is a continual surprise to see what
becomes a hot issue; first return-
receipt msgs and now the freedom to
call the QUASAR robot a f---d.
Keep up the good work,
Dave C
-------
--------------------
Mail from SRI-KA rcvd at 4-May-78 0158-PDT
Date: 4 May 1978 0158-PDT
From: STEF at SRI-KA
To: PBARAN at USC-ISI
Cc: STEF
Subject: Re: DEC Pitch
In-Reply-To: Your message of May 3, 1978
Message-ID: <[SRI-KA] 4-May-78 01:58:48.STEF>
Sender: STEF at SRI-KA
Text:
YEH!
Now what I would like is the freedom to call JMC a narrow minded
polemicst! but you know about sacred cows! Hmmmm/ I wonder if JMC
would like to have mentioned that in fact it is very difficult to deal
with sacred cows such as he when they make such fierce noises?
Cheers - Stef
--------------------
End forwarded messages
--------------------
End forwarded message
∂04-May-78 0749 100 : David Shields via NYU Mangled 3 May message
Dear Professor McCarthy*λ: I think you may have received a partial message from me yesterdayλλλλλλλλλλλλλλλλλλλλλλλ____∪
I think you may have received a partial message
from me yesterday. Please accept my apologies, as I am nnew
to Arpannet mail. Jack Schwartz or Edith Deak will will be in
touch with you shortly. I am seekinng a small 'guest' accounnt
onn a DEC10 to assist inn bootstrap of language LITTLE to DED10λλλc10.
I need to runn small MACRO programs, etc.
David Shields
Courant Institute o Mathematics
New York University
∂04-May-78 1145 Bmoore at SRI-KL AI and Philosophy business
Date: 4 May 1978 1145-PDT
From: Bmoore at SRI-KL
Subject: AI and Philosophy business
To: jmc at SAIL
I ran into Dan Dennett in Cambridge a couple of weeks ago. He said that
he and you and Minsky and Pylyshyn were definitely going to be at the
Center a year from this fall. Do you have any information about whether
they intend to invite anyone else? By the way, my thesis committee thought
that my thesis needed more work on the presentation, so it will be a while
longer before I am officially done.
Bob
-------
They are still pondering your name. The degree certainly must be granted
before you can come to CASBS, but maybe they can't even decide until the
thesis is accepted.
∂04-May-78 1249 Barr at SUMEX-AIM CSD Picnic
Date: 4 May 1978 1234-PDT
From: Barr at SUMEX-AIM
Subject: CSD Picnic
To: siglunch:
Please check out the CSD BBD for the announcement of the Annual Department
Picnic -- Friday May 12th in Huddart Park. Sign up soon with Laurie Levin!
-------
∂04-May-78 1255 STT 224 lecture reminder
Just a reminder that you are scheduled to speak on Formal Methods for
Representing Knowledge this Tuesday, May 9. The class meets 11:00-12:15
in 111 Polya; there are about 50 students; it is not televised this year.
I presume you'll want an overhead projector and will have one out.
Students have been told to read "Some Philosophical Problems from the Standpoint
of Artificial Intelligence" (except the Discussion of Literature), which is a
course handout, and also to read the introduction to predicate calculus in
Nilsson's book if they aren't familiar with p.c. There are bound to be some
people who won't have done the reading, so I think it would be good to give
English interpretations of p.c. formulae as you go.
Ed will introduce you and provide continuity with the rest of the course, please
allow about 5 minutes for this.
The Panel on the History of AI, consisting of Ed, Arthur Samuel and yourself,
is two weeks from today, May 18.
The Panel on Representation has been cancelled (I don't know specifically why.)
Thank you very much for your participation.
--- Steve Tappel (CS/Psych 224 TA)