perm filename OUTGO.MSG[1,JMC]4 blob
sn#613623 filedate 1981-09-24 generic text, type C, neo UTF8
COMMENT ⊗ VALID 00002 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00002 00002
C00003 ENDMK
C⊗;
∂01-Apr-81 1424 JMC
To: FFL
Please send Georgia Navarro at SRI a copy of my long biography.
∂01-Apr-81 1425 JMC
To: navarro at SRI-AI
846 Lathrop Dr.
Stanford CA 94305
857-0672 home, 497-4430 Stanford
I worked for Information International at this rate several years
ago, but I haven't been doing consulting work recently.
You can FTP BIOJMC[PAT,JMC] from SAIL, but I'll send you a paper copy.
∂01-Apr-81 2115 JMC
To: RPG
I have to read more of your thesis to tell.
∂02-Apr-81 0045 JMC
To: rosenschein at SRI-KL, nilsson at SRI-KL
Unless I hear otherwise, I'll come next Thursday at 10.
∂02-Apr-81 1906 JMC Tang
To: FFL
Mr. Tang will leave here June 30. However, we should leave open
the date of his ticket since he may visit a few other places before
he goes back to China. There is no harm in finding out whether
the ticket would be cheaper with some specific dates or advanced
purchase, however.
∂02-Apr-81 2200 JMC more arpa prose
To: JK
I have edited your text slightly for style, but it's fine for describing
what you have done. Now I need half a page or a page on what you plan
for the next two years.
∂02-Apr-81 2205 JMC
To: JK
Blank lines between paragraphs and two spaces after periods help TEX or PUB.
∂02-Apr-81 2243 JMC arpa2.doc[doc,jk]
To: JK
Well, I think it needs a bit more. I put the "Ketonen plans ..." in a separate
final paragraph, but the sentence isn't liftable by Engelmore, because it
refers to "these techniques". If you're tired, perhaps you can give it
some more thought tomorrow.
∂02-Apr-81 2308 JMC
To: JK
You were right, and it looks ok now.
∂03-Apr-81 0122 JMC
To: FFL
If you haven't sent Pournelle report to Levinthal yet, please hold it.
∂03-Apr-81 2138 JMC
To: hprintz at BBNE
I didn't continue with the public files, partly from lack of feedback.
∂04-Apr-81 2130 JMC
To: minsky at MIT-ML
thanks
∂04-Apr-81 2146 JMC
To: LLW
Please phone me 857-0672. A moderately urgent idea.
∂04-Apr-81 2204 JMC
To: pourne at MIT-MC
If you get this by 1:30 please phone me at 415 857-0672.
∂06-Apr-81 0119 JMC arpa proposal
To: FFL
Please get me a xerox of the last ARPA proposal and also find out
if the file is in the computer somewhere. If Les is around, you can
ask him.
∂06-Apr-81 0135 JMC arpa proposal
To: LES
CC: FFL
Do you know what file previous ARPA proposal might be?
∂06-Apr-81 1356 JMC
To: ACY
Notice was sent to obsolete list. You aren't on it this time. Sorry.
∂06-Apr-81 1738 JMC
To: FFL
Please retrieve the files that correspond to pp. 1-31 of the 1979 proposal.
∂06-Apr-81 1743 JMC
To: TOB
∂06-Apr-81 1740 CSD.ULLMAN at SU-SCORE binford case
Date: 6 Apr 1981 1737-PST
From: CSD.ULLMAN at SU-SCORE
Subject: binford case
To: jmc at SU-AI
It went through the A&P committee to the provost!
-------
Tom: The A&P committee is the largest single hazard in appointments.
It isn't sure, but it's very probable now barring some effect of
prospective changes in high-level personnel.
∂06-Apr-81 2132 JMC
To: CLT
PUB yourself a copy of conjec[w78,jmc].
∂07-Apr-81 0142 JMC
To: LGC
CC: RPG
I think I can be in at 11:30, and if so, I'll join you.
∂07-Apr-81 1224 JMC travel agent
To: CLT
Call Dina Bolla Travel 329-0950 and ask for Franklin Hersch.
∂08-Apr-81 2011 JMC
CC: DPB at SU-AI
I cannot take King as a student as our interests don't sufficiently overlap.
∂08-Apr-81 2012 JMC
To: FFL
We will hold the meeting Monday anyway. Golub has agreed.
∂08-Apr-81 2223 JMC
CC: csd.dbrown at SU-SCORE
I cannot take King as a student as our interests don't sufficiently overlap.
∂09-Apr-81 1502 JMC
To: stan at SRI-AI
ANSWERS TO STAN ROSENSCHEIN'S QUESTIONS
1. Predicate calculus (presumably with functions) is ok for
the communication language. The LISPish syntax given in my memo
is still desirable since the purpose is machine-machine communication.
2. Q: What soes it mean to be talking about the same thing? Is this a semantic (e.g. model the
theoretic) notion?
A: Yes, it's semantic, assuming you are referring to
the buyer and seller talking about the same thing. The problem of
qualifying a reference enough so that they are talking about
the same thing is central in human business communication, and much
of the back-and-forth is about that. Notice that they negotiators
often have only a tenuous knowledge of the objects they are
dickering about. If SRI buys a D0 from xerox, this will be done with
an incomplete knowledge of exactly what constitutes a D0, and
the D0 may be somewhat different from the one on display. In general,
however, buying and selling is discussed (usually) as though
the objects were natural kinds.
3. Q: What type of conventions could/must be observed for two P-C
speaking machiees to know they are talking about the same thing?
A: This cannot be assured by conventions, if I understand
what is meant. After some dialog about (say) what color pencils,
both sides think they are referring to the same thing. Often it
can then be tied down by referring to a model number or stock number.
Perhaps one can say that dialog often involves approximating
some continuously variable aspect of the world by a discrete model.
4. Q: Is there a "universal protocol" for shipping cognitively-oriented
(belief and desire) self-descriptions among communicating machines?
A: I'm not sure I understand the question. It seems at first glance
that shipping such concepts is not much more problematical than
referring to pencils.
5. Q: Is an initial shared "subtheory" requred? HOw much of a
description of the world (as opposed to a theory of cognitive
agents) need be in such a shared subtheory?
A: It seems to me that a theory of the world is more required
than any theory of cognitive agents. Consider a company
controller dickering with an IBM salesman about the purchase of
a computer. The salesman says that a power conditioning unit is
required in addition to the CPU, etc. Both can be vague about what
it actually is. The controller's question is "What else will cost
me money?". The ability to introduce new entities is required for
a full ability to negotiate. In order to make a valid contract,
model numbers are often sufficient, provided they refer either implicitly
or explicitly to the common practices of the industry.
6. Q: How much proliferation of speech-act types is really required?
For instance, do we need to distinguish "imperative force" of an
utterance from "causing the hearer to believe the speaker desires
that he take some action" (i.e., a special case of informing", which
would be handled by a single, general-purpose "inform" protocol?
A: I think some performatives will be required, and should be set off
syntactically in order to give legal force to the transactions.
Thus a company can commit itself to pay for what is ordered in a
prescribed manner by its purchasing computer program. The
alternative is that it be a tort to lie about one's intentions.
This is too murky to treat generally. Thus if a computer
says, "If 300 gross of pencils are delivered to me by January 1, I
intend to print a check for $1000 made out to your company
and mail it", this seems harder to treat legally in a general
way than "I hereby offer (on authority of xyz) to buy 300 gross
of pencils for $1000 for delivery by January 1".
7. Q: Can we come up with an interesting scenario to actually
implement? Can we do this quickly?
A: In an ideal world, SRI would devise the CBCL and get paid
by its users, perhaps without ever writing a program using the language.
I'm not sure what is meant by a scenario, and can think of
two kinds. First two programs written here communicate in CBCL
playing some kind of Monopoly game. Second, SRI writes a program
for getting reports to people that keeps track of SRI's own reports,
asks similar program over the ARPAnet about reports of
other labs, answers inquiries and accepts orders. People
Programs written elsewhere communicate with the SRI program in CBCL.
∂09-Apr-81 1733 JMC
To: mrc at SU-SCORE
Is there a way in TOPS-20 to make a file FTPable without a password?
∂10-Apr-81 0024 JMC
To: foo at PARC-MAXC
Is there a user called foo?
∂10-Apr-81 0042 JMC Benjamin Cohen thesis
To: cohen at PARC-MAXC
If you are Benjamin Cohen or if you can send me his Arpanet co-ordinates,
I'll mail comments on the thesis.
∂10-Apr-81 1103 JMC
To: SQU at SU-AI
Good luck in the Math Dept., and I look forward to further discussions.
∂10-Apr-81 2004 JMC
To: JD
The file with the old proposal hasn't been restored yet. Perhaps tonight.
∂10-Apr-81 2153 JMC
To: cohen at PARC-MAXC
Comments on "Understanding Natural Kinds: Part I" by Benjamin Cohen draft
of March 1981
1. I find myself more in agreement with what you actually say in the paper
than what your lecture seemed to be saying (as far as you were allowed to
get) and with what the abstract promises. In particular your footnote
interprets the efforts at non-monotonic reasoning as an attempt to treat
typicality withing logical framework. You don't discuss the prospects for
success of such attempts, but what you say doesn't directly argue against
prospects for such success.
2. As a part of a philosophy dissertation, the paper makes a
presupposition common to philosophers. Namely, it presupposes that
counterexamples kill a formalism dead. AI has to be more modest and take
a positive view of formalisms, asking what can be done with it rather than
immediately looking for something it can't do and letting that kill it.
Unfortunately, much AI work, especially AI dissertations, tend to sweep
under the rug everything a formalism won't do.
In the present case, a proper treatment of the "formal semantics" paradigm
would identify the class of problems for which it is likely to succeed in
formalizing some common sense knowledge well enough to be used.
3. Your point that Montague semantics doesn't even begin to express the
real world facts important for understanding sentences is well taken. I
hadn't realized how little interest philosophers had in such questions
till I attended Barwise's lectures on perception, and he eventually told
me that no specific features of vision were to be treated.
4. Bob Moore is right that definability isn't a very live issue. The
"formal semantics" approach will content itself with axiomatizing concepts
without demanding that they be eliminable by definitions.
Of course, ever since I started working on non-monotonic reasoning about
1975, I have agreed that axiomatization itself isn't enough to express our
common sense knowledge without non-monotonic rules of reasoning. Even in
"Some Philosophical Problems from the Standpoint of Artificial
Intelligence" which appeared in 1969, Pat Hayes and I recognized that
there were serious difficulties in trying to get correct axiomatizations
of common sense. We called the difficulties the qualification problem.
The current non-monotonic reasoning formalisms are probably still
inadequate.
5. The problem of compositionality has to be redefined to become
meaningful, because it is always possible to give any system a
compositional semantics if one is willing to use suitable function domains
for the domains of meanings of expressions. Thus suppose you have some
rule for assigning meanings to sentences as a whole. Then I can define
the meaning of an arbitrary segment of a sentence as a map from the
Cartesian product of possible preceding segments with the space of
possible following segments to meanings for the sentences.
Compositionality can be made a meaningful restriction of semantic rules
only if we restrict the domain that can serve as meanings. Thus modal
logic is non-compositional if wffs must have truth values as meanings but
becomes compositional if we use suitable functions from possible worlds as
meanings.
6. From the AI point of view the reformulation on page 2 of the problem of
natural kinds as a linguistic problem is dubious. Natural kinds like
biological species are a fact about the world and the ability of men and
machines to acquire information about it. These facts have linguistic
consequences, but they aren't basically linguistic facts.
I don't think that natural kinds occur only in nature, and moreover I
think the concept is a relative one. A small child treats almost all
words as denoting natural kinds, i.e. as having meanings that aren't
merely matters of definition and which depend on facts that he doesn't
know. Discovering that some words have extensions whose boundaries aren't
definite comes at an age of perhaps ten.
The word "hill" should be compared to the word "horse". No-one supposes
that there is something that science will tell us about which eminences
are to be counted as hills. Nor will science tell us which animals are to
be regarded as domestic.
It doesn't seem to me that the notion of "natural kind" should be used to
revive (Aristotelian or Platonic ?) essences.
7. Grumbles and misprints
Using Chang and Keisler as a reference muddles what is important about
Tarskian semantics for philosophy with what is of purely mathematical
interest. Isn't the little bit in Rogers' "Mathematical Logic and
Formalized Theories" enough for your purposes.
Corrected spellings: wierd → weird, curousity → curiosity, a-typical →
atypical, , proto-typical → prototypical, quadraped → quadriped.
8. I hope you will sometime switch to putting your main efforts into
making a formalism you consider adequate. My experience is that one
cannot win methodological arguments against formalisms without having a
competitor and examples that the competitor does better.
∂10-Apr-81 2329 JMC
To: FFL
pereir.1
∂11-Apr-81 1103 JMC
To: JD
The old proposal is FR[s81,jmc].
∂11-Apr-81 1103 JMC
To: LES
Thanks.
∂11-Apr-81 2033 JMC
To: steveh at MIT-MC
Suggest you correct your error giving 2 min 33 sec as result of doubling
1.28 sec.
∂12-Apr-81 0102 JMC paper needed
To: FFL
Please get me a xerox of R. Kowalski, Predicate logic as a programming
language, Proc. IFIP 1974
∂12-Apr-81 1139 JMC
To: CLT
How about 2pm for walk, or ... ?
∂13-Apr-81 0950 JMC proposal
To: JD
It looks like an earlier proposal or an early draft has been restored.
Perhaps we had best mark up a paper copy and have Fran type it in.
I suggest you print fr[s81,jmc], and if Les is around ask him what is
the right file. Otherwise, we can have Fran type the stuff in again.
∂13-Apr-81 1125 JMC
To: LGC, RPG
Yes, let's discuss advice taker at 4:30.
∂13-Apr-81 1444 JMC correction of correction
To: cohen at PARC-MAXC
I corrected "quadraped" to "quadriped". Actually it's "quadruped".
Perhaps something substantive later.
∂13-Apr-81 1617 JMC
To: RWW
Call Anil Jain 202 357-7345.
∂13-Apr-81 1624 JMC
To: JD
No problem about 2 month unpaid leave June and July.
∂13-Apr-81 2047 JMC
To: FFL
Friday is possible. Thursday is not.
∂13-Apr-81 1457 FFL Thesis Committee meeting for Vic Scheinman
To: JMC, FFL
Bernard Roth asks if you can meet on Thursday, Apr. 23, at 2 p.m. with
the thesis committee. If not, can you meet at any other time on Thursday.
If not, can you meet on Friday. He would appreciate a reply as soon as
possible.
∂14-Apr-81 0044 JMC REM
To: ellen at MIT-MC
I know there is a temptation to "do something about REM", but I'm inclined
to suggest resisting it unless there is more of an acute problem for other
people than you have indicated. The last time REM gave suicidal indications
I talked to him on the phone quite a lot, and I have talked to him twice
since. My impression is that REM is not as desperate as he makes out.
I think some expression of displeasure may be called for if it hasn't been
tried already, i.e. a direct request by recipients of messages with unwanted
confidences that he refrain in the future. I think there should be no conerted
campaign to do it, but whoever is most annoyed should do it. Of course, this
may have already been tried. Complete success shouldn't be expected - merely
a reduction in the noise level.
∂14-Apr-81 0054 JMC sharing Dick Gabriel
To: LLW
I would like to support him to work on an Advice Taker, but I expect to have
money enough only for half of his time, and anyway he wants to put much of
his time into LISP. Is there any possibility that S-1 could support him
half time, say through the contract with Stanford to work on LISP?
∂14-Apr-81 0134 JMC
To: ellen at MIT-MC
That's a fine birthday cake, and I'll not worry my head trying to figure
out how REM will take it. I don't see that people being disturbed by
fingering REM is quite reason enough to take action even to replace the
plan by something else. I too find REM's appeals disturbing - comparable
to the continual guilt-mongering begging of the listener supported
radio and TV stations. The remedy is the same in either case - switch
to another channel.
I'm curious about Betsy, however. I don't know if I've met her although
she's a user of SCORE, but REM's last flap was about her. I supposed that
she was an innocent user on whom REM had become fixated, but I gather
from your last message that she is also somewhat of a nuisance. In
what way?
∂14-Apr-81 0156 JMC curiosity
To: ellen at MIT-MC
Thanks. My curiosity is satisfied. I may talk further to REM about how
he is making the fulfillment of his goals unlikely, but I expect more
education than success from the effort. Unless you request it, I won't
delete your messages. I see nothing in them to be ashamed of or likely
to have a harmful effect beyond someone possibly taking it upon himself
to explain to one or both of us how we are mistaken about something.
∂14-Apr-81 1648 JMC visit next fall
To: rms at MIT-AI
I'm budgeting at the moment. Do you want to come again for one
or two months? I said I would pay travel expenses, but I know
I can't afford salary. Do you remember what your travel reimbursement
arrangements were with M.I.T. and roughly what it cost them?
∂14-Apr-81 1649 JMC Gabriel
To: LLW
He will probably accept our offer which will be $27,000 or $30,000 for
12 months depending on Stanford politics. Can you pay half of that with
usual overheads?
∂15-Apr-81 1703 JMC
To: csd.golub at SU-SCORE
got your test message
∂15-Apr-81 1720 JMC parser
To: JK
For reasons I will be glad to enumerate, it would be very helpful if
your parser would allow operators comprising more than one operator
symbol, e.g. ∧' or ∧∧. All it needs to do is to consider a string
of operator symbols as designating a compound operator, there being
no necessary semantic relation between ∧ and ∧', etc. This would
also permit expressing ≤ as =< by people with impoverished character
sets.
∂15-Apr-81 2159 JMC
To: CLT
I forgot that I will be at SRI tomorrow. I'll be at Jacks shortly past 5.
∂15-Apr-81 2203 JMC
To: CLT
Unfortunately, I forgot I'm at SRI. Can you make it 5:15 or after?
∂15-Apr-81 2228 JMC
To: CLT
I've checked. I'll cancel my lunch with Elliott Bloom if necessary to
make this meeting. I prefer tomorrow if JD can do it, because it's sooner.
∂15-Apr-81 2334 JMC
To: CLT
1:30 is ok
∂16-Apr-81 0006 JMC
To: FFL
We have to send NSF a new budget totalling $75,000 for my NSF.
∂16-Apr-81 0032 JMC
To: CLT
Done although the due date is listed as April 28.
∂16-Apr-81 0110 JMC
To: JK
I admit I hadn't thought of this last, and I suppose that any of the
reasonable solutions is acceptable.
(1) it is interpreted as a single symbol so that ∧ ¬p must be so written.
(2) it is interpreted the same as ∧ ¬.
(3) it is interpreted as ∧¬ if this has been defined otherwise as ∧ ¬.
∂16-Apr-81 0939 JMC
To: JK
By the way, likewise with combinations like ∃'.
∂16-Apr-81 0940 JMC
To: FFL
Please tell Bloom's sec'y to remind him to send Dugan lists.
∂16-Apr-81 2328 JMC
To: rem at MIT-MC
Good to hear you're better.
∂17-Apr-81 0058 JMC McDermott paper
To: bmoore at SRI-KL
I don't believe the theorem he states on page 14 and purports to prove
on page 17. A counterexample involves the fact that his axioms don't
exclude the possibility of an interval of states with the same date within
a chronicle. Such an interval can have missing limit points, e.g. can
have the order type of the real numbers without the point 0. An
infinite sequence of days and nights having the limit point 0 is possible
now without violating Adams' axiom, because the state corresponding
to 0 doesn't exist. Have you an opinion?
∂17-Apr-81 0956 JMC
To: FFL
yale.1
∂18-Apr-81 1434 JMC ANALYST
To: JD
It seems to me that the time is ripe to actually do a version of the ANALYST
proposed in our old proposal. I have discussed this with Dick Gabriel and
Lew Creary, and they agree. I would like to solicit your participation in
the project and, if you agree, would like to mention it in the proposal.
ANALYS[S81,JMC] contains a preliminary draft of a section of the proposal,
but it is also a preliminary plan for actually carrying it out. Please
phone when you get this. We may want to get together tomorrow (Sunday)
afternoon as I'll be away from Monday morning till Wednesday evening and
will be at SRI Thursday during the day.
∂18-Apr-81 1616 JMC
To: LGC, RPG, JD, CLT
How about 3pm Sunday for getting together and finishing it off.
∂18-Apr-81 1747 JMC
To: LGC, RPG
Any time you two can agree on is ok with me.
∂18-Apr-81 2207 JMC
To: LLW
Do we need to talk about Dick Gabriel?
∂18-Apr-81 2320 JMC Monday
To: darden at SUMEX-AIM
Monday I arrive at Baltimore-Washington airport around 10pm.
My meeting Tuesday is at the National Bureau of Standards. If
these places are close to your place and you have a spare bed,
it would be nice. Otherwise, how about Tuesday evening. I had
to stay for only two days of a three day meeting. I hope you
had a good trip to Boston.
∂18-Apr-81 2326 JMC
To: CLT
The descriptions of other people's work are fine.
∂19-Apr-81 1447 JMC uses of home computer
To: pourne at MIT-MC
Not counting writing or communication, what uses do you get from your
home computer? Consider this as a request for information and not as
a rhetorical question.
It is clear that word-processing by computer is worth the money, and so
is electronic mail and access to data bases when enough cheap enough ones
are available. However, a lot of the proposed uses, such as keeping track
of the pantry fail because they aren't worth the input-output. Others like
closing the window when it rains require standardized sensors and effectors
to be of use to most people. The question of how we get from here to
such applications requires more discussion than I have time for now.
∂19-Apr-81 1602 JMC
To: darden at SUMEX-AIM
That will all be lovely. My flight is UA108 which arrives at BWI
at 1035pm. My breakfast tastes are entirely omnivorous. I too
look forward to it.
∂19-Apr-81 1605 JMC
To: CLT
My accomplishments are in ACCOMP[S81,JMC].
∂19-Apr-81 1733 JMC intro
To: CLT
intro[s81,jmc] contains a draft of the sections preceding ANALYST. Please
add a description of the organization of the proposal to the first page
of it.
∂19-Apr-81 1739 JMC
To: JDH
More specifically, Friday at 3:30 would be good.
∂19-Apr-81 1750 JMC
To: CLT
ARPA[s81,jmc] is the section on my own plans.
∂20-Apr-81 0056 JMC
To: RWW
(1) Anil Jain. (2) Can you support Joe Weening this summer?
∂22-Apr-81 2233 JMC tomorrow
To: konolige at SRI-KL
I plan to come to SRI tomorrow assuming you will be available some
time during the day.
∂22-Apr-81 2248 JMC paper copy of our part of proposal
To: FFL
Please xgp fr81.xgp and also a copy of our (formal reasoning) budget.
I think I'll phone Engelmore and see if he wants a copy.
∂22-Apr-81 2249 JMC verification
To: cerf at USC-ISI
I would like to continue our conversation about getting some support
for our verification related work. I'll be going to France for two
months (visiting logic programming center in Marseille), so would it
be possible to talk between now and Wednesday. I cculd also send a
letter outlining what I have in mind.
∂23-Apr-81 0932 JMC
To: konolige at SRI-AI
If you will be further along then, next Tuesday might be better.
∂23-Apr-81 0934 JMC
To: konolige at SRI-KL
What's your phone number at SRI?
∂23-Apr-81 2136 JMC
To: daul at OFFICE-2
Read ARPAnet directory.
∂23-Apr-81 2158 JMC
To: darden at SUMEX-AIM
Many thanks for picking and putting me up. A propos of a discussion we
had when you were at Stanford about early work in genetics, Judson on
p. 205 quotes Brenner as saying about Morgan and others "They did what
they were able to do, which was genetics". He points out that most of
them started wanting to study embryology and development.
∂24-Apr-81 0028 JMC
To: LLW
I received the offer today. My main problem is with the dates.
I won't be back from France until early in July, and I'll need to spend
some time at Stanford for a while - probably I should be there the month
of July. My best time would be September. One alternative is to put
this on the form. The other is to formally accept for the latest date,
but then put it off, which I would do only to keep the process going.
I have a question concerning pay, which has been proposed at precisely
my academic rate. Is how LL generally pays, or is it characteristic of
this particular summer program? It isn't an important concern for this
summer.
∂24-Apr-81 0038 JMC
To: daul at OFFICE-2
Inspecting a few random pages gave 19 out of 99. The formula
sigma = sqrt(n*p*q) suggests a standard deviation of 4. Several
of the women were known to me as secretaries. Of course, the
directory is heavily loaded towards officials and their administrative
helpers.
∂24-Apr-81 0108 JMC
To: LLW
OK, I've signed it, but we'll see how far $12 per diem goes.
∂24-Apr-81 0150 JMC
To: LLW
I'll talk to Tom. I've just done something for him, so ...
∂24-Apr-81 1222 JMC
To: ROD
It's all true, so do as you propose.
∂25-Apr-81 1313 JMC
To: BYY
The main reference on my use of situations is:
%3McCarthy, John and P.J. Hayes (1969)%1: "Some Philosophical Problems from
the Standpoint of Artificial Intelligence", in D. Michie (ed), %2Machine
Intelligence 4%1, American Elsevier, New York, NY.
∂26-Apr-81 1926 JMC 8th day
To: darden at SUMEX-AIM
It's indeed excellent, although I found the parts where I knew
the outcome (as in the existence of mRNA and tRNA) more readable
than the parts for which I didn't (repressors, etc.). It seems
to me that the discoveries were mainly technique-driven - when
the techniques became available the discoveries were made. It also
seems that the Club was trying to be too clever. The fancy ideas
for codes were no more intrinsically plausible than the ultimate
3 base code with redundancy. The one experiment besides that of
Meselson and Stahl that might have been done earlier seems to
be Nirenberg's, although it isn't clear what techniques he used
that might have been newly developed. It occurs to me to ask whether
Jacob and Monod when they observed bacterial "mating" might have
tried to inject some DNA or RNA themselves, e.g. polyA or polyU.
Would that have worked? It could conceivably have been done with
a single bacterium using the radioactive tracer techniques used
by the Berkeley chemists who were discovering new elements and
detecting the decay of single atoms. The idea was to guess the
chemistry and purify the substance blind and they put the hypothetical
substance in a counter and detect the decay. If you guessed that
polyA would produce a certain protein polymer, then it would be
necessary to go through electrophoresis and chromatography to separate out
that and then put the bit of paper in a particle counter.
See you after we return.
John
∂27-Apr-81 0902 JMC
To: darden at SUMEX-AIM
leaving for Marseille Thursday and returning around July 1
∂27-Apr-81 1204 JMC
To: CLT
Franklin Hersch 329-0950 needs your passport number by 4:30pm.
∂27-Apr-81 1208 JMC
To: CLT
I found it and called it in. It's B1225574.
∂27-Apr-81 2142 JMC
To: FFL
Please send a copy of the Shapiro paper to Suppes.
∂28-Apr-81 0041 JMC
To: DCL, MAS
I think she has had it for some days. At least Fran isn't asking me
for anything now.
∂28-Apr-81 2323 JMC
To: RWW
∂28-Apr-81 1202 DGCOM at USC-ISIC FOL
Date: 28 Apr 1981 1202-PDT
From: DGCOM at USC-ISIC
Subject: FOL
To: jmc at SU-AI
cc: dgcom
Prof. McCarthy,
We discussed in Washington the possibility of my examining FOL.
If you could send me a manual, open an account, etc. I would
appreciate it. More generally, I am interested in the logical
and epistemological issues in AI. I've spent the last ten years
studying philosophy. I am now preparing to reenter the world of
research and am looking for fruitful things to work on. I've spent
the last eight months working in program verification in order to get
aquainted with computer culture. I'd be interested in any reprints,
advice, etc. that you could pass on.
Richard Platek
-------
∂29-Apr-81 1852 JMC proposal
To: engelmore at USC-ISI
We have made a major modification to the content of what I told you
would be the proposal. Namely, Dick Gabriel and Lew Creary think they
can make a start on implementing the intelligence ANALYST that has
been the focus of the theoretical work. It will be a kind of
Advice Taker. We are terribly squeezed by the budget limitation you
gave us, and I have started to discuss with Vint Cerf the possiblity
that some of our work on mathematical theory of computation may be
appropriate for additional support from the program verification
program.
∂29-Apr-81 1944 JMC
To: FFL
Please put dedije.1 into TEX and print it for me to sign.
∂29-Apr-81 2155 JMC
To: dgcom at USC-ISIC
CC: FFL at SU-AI, JK at SU-AI, RWW at SU-AI
∂28-Apr-81 1202 DGCOM at USC-ISIC FOL
Date: 28 Apr 1981 1202-PDT
From: DGCOM at USC-ISIC
Subject: FOL
To: jmc at SU-AI
cc: dgcom
Prof. McCarthy,
We discussed in Washington the possibility of my examining FOL.
If you could send me a manual, open an account, etc. I would
appreciate it. More generally, I am interested in the logical
and epistemological issues in AI. I've spent the last ten years
studying philosophy. I am now preparing to reenter the world of
research and am looking for fruitful things to work on. I've spent
the last eight months working in program verification in order to get
aquainted with computer culture. I'd be interested in any reprints,
advice, etc. that you could pass on.
Richard Platek
-------
I will have information about both Weyhrauch's FOL and Ketonen's EKL
sent, but please send a U.S. Mail address to my secretary Fran Larson,
FFL@SU-AI.
∂29-Apr-81 2157 JMC
To: FFL
Please send Platek when he supplies an address copies of my AI and
philosophy papers.
∂29-Apr-81 2159 JMC
To: JK
∂25-Apr-81 1112 Susan L. Gerhart <GERHART at USC-ISIF> M. Davis paper
Date: 25 Apr 1981 1105-PST
From: Susan L. Gerhart <GERHART at USC-ISIF>
Subject: M. Davis paper
To: mccarthy at SU-SCORE
cc: gerhart at USC-ISIF
I'd like to follow up on the paper you mentioned at the Verkshop.
If you just have the title our library can chase it down but a copy
of the paper would be appreciated, if it's available.
I have a student down here from Berkeley who's interested in the structure
of proofs for a possible thesis area.
Thanks.
Susan Gerhart
-------
She wants the paper about obvious deductions. I think you have a copy.
∂30-Apr-81 1607 JMC →14367 (1-Jul-81)
To: "#___JMC.PLN[2,2]"
I will be at the University of Aix-Marseilles to June 15 and
back here July 1 approximately. Address:
c/o Prof. A. Colmerauer
Groupe d'Intelligence Artificiel
Universite d'Aix-Marseille
70, Route Leon-Lachamp
13 - Marseille (9 e)
FRANCE
telephone: France + 91 41 32 48
∂30-Apr-81 1634 JMC verification
To: cerf at USC-ISI
I hope to continue discussion of possible ARPA support for our
work in the mathematical theory of program verification after I
return to Stanford around July 1. The numbers I mentioned to you
would support all my group's work in this area. A smaller amount
$50K to $100K is needed just to keep it alive in addition to what
we are getting in Engelmore's program.
∂20-May-81 1439 JMC
To: carl at MIT-AI
∂14-May-81 1318 Randall Davis <KRD at MIT-AI>
Date: 14 May 1981 15:43-EDT
From: Randall Davis <KRD at MIT-AI>
To: konolige at SRI-AI
cc: KRD at MIT-AI, nilsson at SRI-AI, jmc at SU-AI
Kurt:
Read over your TN232 recently and liked it very much (tho I don't
claim to be much of a logic hacker at all). It appears to put down
a very nice foundation for the reasoning about beliefs and actions stuff.
A couple of minor typos that you've probably heard about:
-- on page 27 you say PO1 several times and I suspect you mean LT1.
-- page 35, middle paragraph, after the ref., you seem to have been
trying to make up your mind about how many examples had been done.
A comment about the wise men problem: it seems to involve not only
reasoning about belief, but about capabilities. The difference in
speed of reasoning of the agents is both necessary and closely bounded.
To see this, imagine that the differences in intelligence are in fact
very large. Then consider #1's standard reasoning:
"if mine were black, then #2 would see a black and a
white, and he could say
'if mine were black, then #3 would see two
blacks and he would have responded by now'."
BUT: if #1 is VERY dumb, then #2 can't make that inference. And (much more
plausibly), if #3 is significantly faster than #2, then he can't say "since #2
hasn't responded, mine can't be black, or #2 would have solved his (simpler)
problem already."
The problem itself is made plausible by the fact that each level of
hypothesization reduces the problem complexity (eg, the nth guy is left with
a trivial observation and inference), but the time issue is a tricky one.
In effect, there are hidden assumptions that each agent believes that each
other agent is only a little bit dumber, and hence given the hypothesized
simpler problem to solve, would have solved that simpler problem already.
You get around this in your formalization because you poll the men. But in
principle at least, the problem is insoluble without assumptions about speed of
reasoning. The first guy could be SO dumb that even when seeing 2 black spots
it takes him arbitrarily long to make the obvious inference [maybe he's running
a resolution theorem prover]; my claim gets more plausible as we move up each
level of embedding [the second guy's problem is a little harder, the third more
difficult still, and so forth if there were more people involved.]
Agree, or have I overlooked something?
cheers
Randy
∂20-May-81 1442 JMC
To: engelmore at USC-ISI
I'm in France till July 1.
∂20-May-81 1447 JMC
To: feldman at SUMEX-AIM
Please make sure your application is primarily considered by Jeff Ullman
and Ed Feigenbaum. In any case, I'll be gone till July.
∂20-May-81 1450 JMC
To: FFL
There is a letter on my desk at the left from Donald Perlis asking about a job
in the department. Please write him that I welcome a look at his thesis, but
he should make sure that his application is considered by Ullman and Feigenbaum.
∂20-May-81 1534 JMC →14367 (1-Jul-81) via MIT-AI
To: "#___JMC.PLN[2,2]"
Address till June 19 arrival
John McCarthy
Groupe d'Intelligence Artificielle
Faculte de Sciences Luminy
70 route Leon Lachamp
13288 Marseille Cedex 2
FRANCE
till June 25, ask FFL
Back at Stanford around July 1
∂24-Jun-81 0258 JMC via MIT-AI space and communications
To: OTA at SU-AI, pourne at MIT-MC
I have been abroad for two months, and I am reading
this from Sweden over a kludgy connection. I will be back July 1
and I'll deal with both matters then.
∂24-Jun-81 0308 JMC via MIT-AI Perry use of terminal for now
To: BYY
I'm away till July 1, and I don't know about terminal requirements
within our project. It's ok
for now, but I'll review it when I
come back. I'm eager to to encourage
philosopher use of our machine for
communication reasons, but ultimately John should buy his own
terminal.
∂26-Jun-81 0214 JMC via MIT-AI machado
To: wiederhold at SUMEX-AIM
I'll be back July 1 and prefer to talk to people before
scheduling a meeting, but I'll be at Stanford for all of July.
ges before
I return.
∂26-Jun-81 0224 JMC via MIT-AI
To: mtruro at MIT-AI
thiss is jmc in Sweden. Query KEN at ai for details later.
∂02-Jul-81 1105 JMC
To: RWW
Thanks. Yes, she's back too.
∂02-Jul-81 1107 JMC
To: JRA
July 10 is fine. Is "Proving correctness of LISP programs" a suitable
topic? If so, that's it.
∂02-Jul-81 1116 JMC
To: ken at MIT-AI
Your June 29 note is a bit sketch, because I have forgotten about Omega, but
I get the general idea. I'll send you a note soon about map coloring strategies
as control for the logic of a simple PROLOG program. Also Pereira and Porto
tinkered with my eval, and there are now "official versions" with and without
cut although the version without (by Porto) has some constructs that may
be somewhat impure. More on this later. Regards to Sten-Ake.
∂02-Jul-81 1121 JMC
To: RPG at SU-AI
I'm back. See if the attached note from Betty Scott presents any opportunities.
We can certainly lend the money out of my unrestricted funds. If Engelmore
has already been in touch with you, let me know what was decided or propoed.
∂29-Jun-81 1420 Betty Scott <CSD.BSCOTT at SU-SCORE> ARPA Supplement for LISP Project
Date: 29 Jun 1981 1412-PDT
From: Betty Scott <CSD.BSCOTT at SU-SCORE>
Subject: ARPA Supplement for LISP Project
To: JMC at SU-AI
cc: CSD.BScott at SU-SCORE
Bob Engelmore called me a week or so ago to say that ARPA has no more money
for this year, so it will not be possible to supplement your contract for
work this summer for the LISP Timing Evaluation Project. He did say,
however, that he would have no objection to funding the project after 10/1,
so that if you could (or want to ) do the work this summer, and be funded
"after the fact" for "evaluation of the results," he thought this would be
possible. It's just that there is no more money available to be awarded
in this fiscal year.
Betty
-------
∂02-Jul-81 1127 JMC
To: GIO at SU-AI, FFL at SU-AI
We'll do our show-and-tell as scheduled July 28, 8:30-noon.
∂02-Jul-81 1139 JMC
To: LLW
I have known Bill Gosper for more than ten years, and consider him
the one of the two or three most inventive and dedicated computer mathematicians
around. Indeed it seems that his dedication to always working on
ever new mathematical applications of computers has hindered his pursuing
a conventional academic career. I have no hesitation in recommending him
without qualification.
P.S. to LLW. I have just returned and will call you soon unless you call
me first. If you need a more elaborate version of the Gosper reference,
I will do it.
∂02-Jul-81 1728 JMC
To: gabriel at CMU-20C
There is to be a show-and-tell for the new ARPA AI program manager
on July 28. Can you come for that and add weight to the Advice Taker
implementation plan? As for France, there were several interesting
things including getting introduced to PROLOG and some work on
non-monotonic reasoning. I have become convinced that there is much
that is interesting in logic programming, and I have brought back
PROLOG and LOGLISP systems to play with.
∂02-Jul-81 1905 JMC
To: JD
I guess you should send me a letter of resignation.
∂03-Jul-81 1201 JMC your paper
To: JD
It's Ottawa. I was naturally curious about your reaction to circumscription.
Your reference the paper, but I couldn't find any discussion.
∂03-Jul-81 1309 JMC
To: REG
If needed, Ernie Sibert 315 423-4442 will advise on LOGLISP.
∂03-Jul-81 1314 JMC
To: FFL
ROBINS.2[LET,JMC]
∂03-Jul-81 1322 JMC
To: JD
I found your letter of resignation.
∂03-Jul-81 1323 JMC inquiries about LISP
To: JRA
I often get inquiries about LISPs for various machines. If there is any
way you think you can make money out of it, I would be glad to refer them
to you. I send one as a sample, but you needn't answer it.
∂03-Jul-81 1547 JMC
To: LGC, CG, CLT, JK, RWW
ARPA AI wallah here July 28 for show-and-tell.
∂04-Jul-81 1342 JMC
To: FWH
Is Burstall still around?
∂04-Jul-81 1347 JMC
To: FFL
Chris plans to support himself from other sources for the other half time.
∂04-Jul-81 1355 JMC
To: OTA
If you are still editing my paper, I'll be glad to look at what you have.
∂04-Jul-81 1356 JMC
To: pourne at MIT-MC
Sorry to have missed you in Palo Alto. Just got back from France.
∂05-Jul-81 1232 JMC symposium
To: FFL
Please call Prof. Ralph Hester or secretary 7-4183 to say that I
accept their invitation to take part in the International Symposium
on Order and Disorder.
∂05-Jul-81 1536 JMC Chronicle
To: csl.bkr at SU-SCORE
I believe you are mistaken in your correction. There is a law passed
in the early 70s which permits newspapers to merge for production
purposes under conditions that require them to remain distinct
editorially. The purpose was to reduce the trend to one-newspaper
cities caused by increasing costs of production. The Examiner and
the Chronicle share production and the Sunday paper under the protection
from antitrust of this law. No-one to my knowledge has claimed that
the Chronicle has become a Hearst newspaper in editorial content.
∂05-Jul-81 1634 JMC
To: JPM
Club of Rome
I was surprised at the following remark and especially at the
remark that the study is unchallenged by modellers. The most thorough
refutation is that of the University of Sussex group.
"The Club of Rome study is basically valid. No one with any knowledge
of world modelling disputes this. Many details can be altered, but
basically if we continue to industrialize, expand population, have no
radical breakthroughs in technology, and continue to exploit the
present (and reasonably expected) resource base, the world (i.e. 80%
of the population) will die in 100 to 300 years. That probably would ..."
I find two major faults.
1. The one dimensional model assumes a fixed fraction of
wealth created is reinvested in more factories, etc. This drives
the exponential and the results is the explosion predicted.
However, this completely ignores factors that saturate demand.
Suppose someone in 1875 had commented that the post civil war cattle
boom if continued at the same exponential rate he had observed since
1865 would require every American to eat a cow a day by 1950 and
formed an organization to reduce growth. A more trivial example:
Let me warn you that if you continue sitting at your terminal reading
this you will eventually fall out of your chair from exhaustion,
thirst and starvation.
In short, before one organizes to stop a trend, one has
to be sure it won't stop by itself. The Club of Rome study, because
of its extreme aggregation, neglects the fact that each increment
in production must be justified by demand for the product in
comparison with more leisure.
Remember that the disaster predicted by the Club of Rome
is not the population explosion but a production explosion.
2. The aggregation of all countries together conceals the
fact that some countries have current population problems and
others don't. Bangladesh, India and China have population problems,
and the U.S. doesn't now and won't for at least 100 years and most
likely for several hundred years. A globalist prefers to consider
the problem on a world wide scale, but this neglects the sovereignty
of nations. The U.S. cannot solve Bangladesh's population problem,
and won't even seriously try.
The globalist may prefer a world government, but this would
be a disaster at present, because any one political unit is too
likely to be swept by some ideology into disastrous policies. Political
diversity is necessary until there has been a great advance in
political science and a great advance in its acceptance by the
political process. The only thing that limits persistence in
disastrous policies in many countries is the observation that
other countries with different policies aren't suffering the
disaster. The Chinese Communists are even saying "Learn from
Taiwan?" although out of the other side of their mouths they
are demanding that we let them suppress the embarassing example
of different policies leading to prosperity.
While I favor spending about three times what we do now
on space exploration, I don't believe we can't survive without
it. The breeder reactor can provide sufficient energy for
hundreds of millions of years, and low grade ores can provide
minerals.
My main reasons for wanting more space exploration are:
1. Esthetic. Mankind expanding into space seems beautiful
to me. Also I like great technology for its own sake.
2. I would like to go in space myself. I am somewhat bitter
that environmental and anti-technological sentiment has made it quite
unlikely that I will be able to experience space travel - even a
vacation in a low orbital hotel. I still have some hopes for this
though.
3. Space colonization would disperse the human race sufficiently
so that nuclear war would be less likely to wipe out mankind.
4. Our world would benefit from having a new frontier where
individuals, groups and societies could develop to suit themselves.
5. Eventually, we'll explore the galaxy and beyond.
∂06-Jul-81 1333 JMC
To: pourne at MIT-MC
I don't know enough about it to have an opinion.
∂06-Jul-81 1339 JMC
To: burstall at PARC-MAXC
How about lunch tomorrow or soon. I just got back from two months in
Europe.
∂06-Jul-81 1445 JMC
To: burstall at PARC-MAXC
How about Monday the 13th at the Little Hsi Nan on University and Emerson
at noon? It is the generally favored Chinese place.
∂06-Jul-81 1844 JMC
To: admin.mrc at SU-SCORE
Please phone 7-4430.
∂07-Jul-81 1210 JMC
To: FFL
I'll be here tomorrow to meet the Chinese girl.
∂07-Jul-81 1333 JMC
To: CG
We need to talk about your support for the summer.
∂08-Jul-81 1252 JMC
To: icl.redford at SU-SCORE
Congratulations on your point 3 about preferring not to get up to
fix the windmill.
∂08-Jul-81 1256 JMC
To: CLT
The blind robot puzzle is due to Donald Michie. As I remember it, the robot
was initially not supposed to be blind, and it was to be a more typical
AI problem. It's just that Michie or someone else noticed that it could
be solved even by a blind robot. My suggestion is either to call or
write Michie (possibly at U. Illinois or Edinburgh), ask Harry
Barrow at Fairchild (who probably was in Edinburgh at that time), or
look in my files or the library for a Michie memo.
∂08-Jul-81 1607 JMC
To: csd.tajnai at SU-SCORE
No. I have several papers to write before then, and I haven't
scheduled the time.
∂08-Jul-81 1728 JMC
To: konolige at SRI-KL
9:30 will be fine. See you then.
∂10-Jul-81 1814 JMC
To: FFL
I have informed Davis.
∂11-Jul-81 1426 JMC AI and philosophy meeting
To: CLT
It is now called the "Information and the Mind/Brain Conference" and
will be held Aug 21-23 at U.B.C. in Vancouver. In charge is Steven
Savitt, U. of B.C., Department of Philosophy. His telephone number
is 604-228-2511, and there is a postal strike in Canada.
∂11-Jul-81 1510 JMC
To: FFL
I need to change Bosack's L to A.
∂13-Jul-81 0014 JMC Beckman error
To: REM
I think Beckmann has slipped in omitting California water power, especially
since I believe he has mentioned it in the past. I believe it amounts
to 25 percent in the PG&E area, presumably less in Southern California.
For the "California Scientists' Statement on Nuclear Energy", I tried
to find out how much of our electric power comes from oil. 57 percent
is the figure that sticks in my mind. No coal is burned in California,
but some coal generated power comes from Four Corners.
∂13-Jul-81 1352 JMC
To: bledsoe at SRI-KL
Two numbers m and n are chosen such that 2 lesseq m lesseq n lesseq 99.
Mr. P is told the product, and Mr. S is told the sum. The following
dialog occurs:
Mr. P - I don't know the numbers.
Mr. S - I knew you didn't know them. I don't know them either.
Mr. P - Now I know them.
Mr. S - Now I know them.
What are the numbers?
∂13-Jul-81 1422 JMC
To: JMM
I would like to see you this afternoon or 3pm tomorrow.
∂13-Jul-81 1816 JMC
To: ROY
Thanks for fixing the Imlac.
∂13-Jul-81 2346 JMC
To: pourne at MIT-MC
Thanks for the copy of "The Survival of Freedom".
∂14-Jul-81 1459 JMC vertical hold on my Datadisc
To: DCO at SU-AI
Besides my Datamedia problem, my Datadisc has just lost its vertical hold.
∂14-Jul-81 1513 JMC
To: oppen at PARC-MAXC
Previous message was to Don Coates and typed on moving screen.
∂14-Jul-81 1704 JMC Garland publishing
To: FFL
I want to recommend that Garland Publishing include Chris's thesis
in their series of outstanding computer science theses. For this
I would like to know the full name and address of the man I wrote
to before. It would have been between 1977 and 1979. If you can't
find it, you could phone and find the name of the boss.
∂14-Jul-81 2245 JMC
To: energy at MIT-MC
I think Beckman is right in deriding the idea that utilities should
fund conservation measures. Redford remarks that adding new capacity
is expensive, but it is no more expensive than it has been in the
past. The difference is that regulation drags out construction time
and high interest rates make that extremely expensive. Moreover, and
this is most important, the regulators don't permit amortizing the
real dollars that a plant cost but only the nominal dollars, and we
are in a period of high inflation. For this reason utility stocks
and bonds are not a good investment, and the utilities are constrained
to keep demand down.
A utility like Consolidated Edison that charges ten cents a kilowatt
hour (I think it's reached that), because it imports oil makes a profit,
while a utility like Commonwealth Edison which uses nuclear energy and
charges five cents a kilowatt hour loses money. I don't remember the
exact numbers.
We are paying dearly for this false economy. In the first place
there are oil imports for generating electricity. More important than
that, after we displace the use of oil for generating electricity by
nuclear energy (or less favorably coal), we can start using electricity
to displace other uses of oil and gas such as heating. If we ever develop
good electric cars or hydrogen powered cars, we will need the generating
capacity to charge the batteries or electrolyze the water to produce
hydrogen. (Maybe we can use nuclear energy to split water thermally).
Other countries less rich than the U.S. like France and Taiwan
have no problem in raising the capital for new electric plants. The
utility problem is strictly caused by the populist and environmentalist
regulators.
∂15-Jul-81 0041 JMC
To: csd.malik at SU-SCORE
The weekend will be ok. Call me when you return at home if necessary.
∂15-Jul-81 0100 JMC
To: LLW
When's a good time to phone?
∂15-Jul-81 1755 JMC
To: FFL
Yes, please. I like salmon.
∂15-Jul-81 1758 JMC
To: nilsson at SRI-AI
Yes, without exclusive rights, it's fine as I remember it.
∂16-Jul-81 1149 JMC
To: FFL
Remind me to look for old films for Sosna.
∂16-Jul-81 1600 JMC
To: FFL
CC: TOB
Fran: Please make a letter for my signature to whoever is in charge of
faculty club membership.
∂16-Jul-81 1442 TOB
John
I applied to the faculty club and was turned down. Would you sign
a letter in support of my membership?
Tom
Dear
I write to support Dr. Binford's petition for membership in the faculty club.
Dr. Binford, who has been at Stanford since 1970, is the leader of the Computer
Science Department's large research program in
computer vision and industrial automation. He has been recommended
by the Computer Science Department for promotion to the rank of Adjunct
Professor. He is in every way the kind of person who should be a member
of the Faculty Club.
Moreover, his use
of faculty club facilities would benefit our joint research program.
Sincerely,
∂17-Jul-81 1458 JMC
To: BYY
Any time till 5:30.
∂17-Jul-81 1702 JMC calling Woodward
To: RWF
He is a patent lawyer contacted by Stanford about password protection
for equipment. I talked with him and he thinks we need a bit more
specification. I'm reminding you that you agreed to take the next step
in getting things specified. I suggest you call Hank Woodward 326-0747
and meet with him.
∂18-Jul-81 1432 JMC
To: oaf at MIT-MC
No. I didn't study any Japanese till after my first visit in 1969.
∂18-Jul-81 1712 JMC
To: PJH
Where are you now? Why don't you make a new plan?
∂19-Jul-81 2117 JMC bug in prolog
To: ken at MIT-AI
In the prolog I got from you not doesn't seem to work correctly - always
giving the result NO. In particular, not(fail) and not(not(true)) give
this result. Defining nott in accordance with the definition of not in the
1978 User's Guide gives a predicate with the expected properties. Is
it broken chez vous also?
∂19-Jul-81 2136 JMC
To: LGC
5pm is ok.
∂20-Jul-81 1035 JMC
To: TW at SU-AI, DPB at SU-AI, feigenbaum at SUMEX-AIM,
buchanan at SUMEX-AIM, genesereth at SUMEX-AIM,
lenat at SUMEX-AIM
In principle I agree with having as few as possible items on the reading
list and therefore go along with the proposal. Might it be possible to
reduce the it a bit more?
∂20-Jul-81 1043 JMC
To: JRA
I was afraid things would turn out as you say they are.
∂20-Jul-81 1514 JMC return
To: darden at SUMEX-AIM
I'm back now from France. Are you planning to be in the Stanford area
or at IJCAI this summer?
I don't know when I'll next be in the Washington area.
∂21-Jul-81 0006 JMC
To: LGC
Well, this is more comprehensible to me than the paper, so it must be the
terms in which philosophical controversies are couched that baffle me. Did
Cartwright have any reaction - or any of the other local philosophers?
∂21-Jul-81 0009 JMC
To: rms at MIT-AI
OK, but I may ask a similar favor some time.
1. Say again what file it is.
2. Are you coming out this Fall, and if so, can I choose the time?
Probably after October 1 would be best.
3. Are there any prolog hackers at M.I.T.?
∂21-Jul-81 1111 JMC
To: pourne at MIT-MC
How would you like an article advocating one way missions to the moon
and possibly Mars? These aren't suicide missions, because we are
committed to resupply the astronauts, but they are committed to stay
there for a long time - perhaps even the rest of their lives. The point
is that the rocketry required for one way missions plus resupply is
vastly less than that required to return people, and people who are
there permanently can do vastly more exploration. The risks are no
greater than those endured by nineteenth century explorers. The trouble
is the fear of being blamed by the media - so nothing must go wrong.
We need to overcome this fear.
Another possibility is some version of my article on the feasibility
of interstellar travel based on present science.
∂21-Jul-81 1128 JMC
To: human-nets at MIT-AI
The article about Data General's Eagle contributes one amusing fact.
The rush project described was needed, because Data General didn't
realize that 16 bit addresses were becoming obsolete until D.E.C.
came out with VAX. The trade press has been trumpeting the decreasing
cost of memory since the early seventies. Machine designers have
always reserved too few bits for address. I remember arguing with
the designers of Illiac 2 in 1958 that 8192 words was not enough
memory for a number cruncher. The trouble is that the engineers
can find lots of "interesting" uses for the bits in the instruction
word, and using them for mere address is dull. Curiously, while IBM
is generally reactionary, since the IBM 704 came out in 1955, they
have usually (but not always) done better than their competition in
providing large addresses. However, the 704 would have had an 18 bit
address instead of 15 but for 5 cute instructions.
∂21-Jul-81 1450 JMC
To: paul.rosenbloom at CMU-10A
There is no reference. I invented it around 1957 or 1958 and named
it the killer heuristic at that time. It was first used by Paul
Abrahams in a two-move-mate program based on my legal move routines.
Abrahams verified that it did indeed reduce the search required to
do two-move-mate problems, but I'm pretty sure he didn't publish
either. I can't recall if it was used in the kalah programs written
at M.I.T. and Stanford which were described in early Stanford AI
memos.
∂22-Jul-81 0136 JMC
To: pourne at MIT-MC
I remember your citing Frosch as having said that it would
be impossible to return to the moon in the 1980s. Of course, a giant
space program could do it, but there may be a way to do it for a cost
the Reagan administration would pay. The question is whether the
Shuttle could land enough payload on the moon for a one way mission.
Naturally, supply rockets could land first and there would be
additional supply rockets afterwards. Do you know or do you have
a reference to the payload the Shuttle could land on the moon?
While I was thinking about including this in my article, it might
be worthwhile to consider launching the trial balloon in a better
way. Perhaps even the Citizens' Space Council might risk its
prestige by suggesting it first. Or maybe it would be better
for them to support it rather than initiate it. Anyway apart from
the politics, I will work up a draft, but it will be light on
details.
∂22-Jul-81 0357 JMC
To: konolige at SRI-AI
How about 3pm today?
∂22-Jul-81 0738 JMC
To: JRA
No, but I suppose there must be people at SRI in the Computer Science Dept.
∂22-Jul-81 1620 JMC visit
To: stan at SRI-AI, nilsson at SRI-AI, konolige at SRI-AI
Unless you especially want me to, I won't come tomorrow or until the
beginning of August. Nils mentioned starting a new person on CBCL,
so I'll come when and if I could help initiate him. Otherwise, it
may be better to wait till after deduction month.
∂23-Jul-81 0001 JMC
To: FFL
lib.lst[1,jmc] has been renamed LIBRAR[1,JMC].
∂23-Jul-81 0001 JMC
To: ME
Thanks for restoring those files
∂23-Jul-81 1119 JMC
To: konolige at SRI-AI
That's right. They only did the arithmetic.
∂23-Jul-81 1128 JMC
To: FFL
Yes, I'll go to lunch on Monday.
∂23-Jul-81 1350 JMC
To: hearn at RAND-AI
I have Stoyan twice in the Soviet Union and had a considerable correspondence
with him when he was writing about the history of Lisp. I guess I could
match your $500 contribution, but I don't have a job for him. Can you
MAIL me his current address and phone - if you have it. I presume he
is in West Germany now.
∂23-Jul-81 1555 JMC
To: hearn at RAND-AI
I will be at Stanford on the 17th but must leave for Boston on the
18th. Where is Stoyan now?
∂23-Jul-81 1616 JMC
To: hearn at RAND-AI
Let's see the titles, although there may not be many people around at the
time. As you may know, it requires special effort to get a visa that
permits payment of travel expenses and even more effort to be able to
pay an honorarium. I am counting on you to get that disentangled.
What is his actual address and phone number?
∂23-Jul-81 1651 JMC
To: hearn at RAND-AI
Stanford has lately been enforcing the rule against it.
∂23-Jul-81 1923 JMC
To: hearn at RAND-AI
No doubt RAND knows how to do it, but I understand that if he has a letter
of invitation that offers to pay expenses or an honorarium, he can get the
kind of visa that allows him to accept it. Ask your local bureaucrat.
∂23-Jul-81 2129 JMC
To: hearn at RAND-AI
Ok, but what are Stoyan's address and phone.
∂23-Jul-81 2143 JMC
To: hearn at RAND-AI
Thanks for the address and phone.
Are you acquainted with the Charles Babbage Foundation, which is concerned
with the history of computing? It occurs to me that they might be interested
in supporting Stoyan's research in the history of LISP if he wants to
continue it. Do you know his job situation in Germany? Does he have
prospects there and does he want to stay there. By the way, the money
bags on IJCAI is Don Walker at SRI. If you don't know him, I can try
for Stoyan, although I already arranged for a subsidy for someone else
with him. What has been your contacts with him?
∂24-Jul-81 1422 JMC
To: wertz at SRI-KL
If it's you playing HAUNT logged in as PG, give good reason or stop.
∂24-Jul-81 2232 JMC Stoyan books
To: PP
If you still have them, I'd like them back. Stoyan is in West Germany,
and Tony Hearn of RAND is trying to arrange a visit to the U.S. and to
IJCAI. I want to take another look at what he wrote about the history
of Lisp.
∂25-Jul-81 1342 JMC
To: human-nets at MIT-AI
I cannot accept the Mother Jones article as a clerk's view of office
automation without further verification. The problem is that Mother
Jones has a prejudice that new technology introduced in a capitalist
society is more likely than not to be harmful to workers. Therefore,
they encourage writers who look for facts tending to confirm this view
and solicit statements from workers confirming it and to ignore improvements
in working conditions resulting from the new technology. I scanned
the Mother Jones article, but I forget whether it mentions the benefits
of not having typewriters clacking away in an office. I assume that
Barbara Garson is a writer not a clerk.
I have not observed mass clerical work, but I have observed the effect
of the use of text editors (like E and EMACS) and text formatters (like
PUB and TEX and SCRIBE) on secretaries and typists. The effect has been
to differentiate skills. Those who have learned to be at home with
computers and use them in a sophisticated way have a premium skill.
We are allowed to pay them something extra, but not enough. Namely,
they are often hired away from us by companies with less rigid personnel
departments than Stanford University.
Productivity measurement has two sides to it. On the one hand it lends
itself to all kinds of petty tyranny. On the other hand it increases
productivity which makes the company less resistant to wage increases.
A company will take a long strike if the consequences of giving in to
the demands are to put it from a profitable position to a loss, while
it will resist much less if it can do so without reducing its profits
much and can hope to make up the immediate cut in profits by an increase
in volume. Also a productive company will bid higher for skills it
wants and will be more likely to match offers from the outside.
∂25-Jul-81 1439 JMC
To: KDO
326-6065
∂25-Jul-81 1730 JMC
To: TOB
What more would be required for the system to run on the 4331?
∂27-Jul-81 1805 JMC
To: RPG
I don't anticipate that Doyle will be there - 252 I guess.
∂27-Jul-81 1816 JMC
To: hearn at RAND-AI
I think it can be changed and I suspect it must.
∂27-Jul-81 1827 JMC
∂27-Jul-81 1805 JRP forwarding barwise's mail
should barwise's (byy) forwarding address be put on the
list? it is, barwise.uwisc@udel. I would doso if
i knew how. j.perry
∂27-Jul-81 1827 JMC
To: ME
∂27-Jul-81 1805 JRP forwarding barwise's mail
should barwise's (byy) forwarding address be put on the
list? it is, barwise.uwisc@udel. I would doso if
i knew how. j.perry
∂28-Jul-81 1437 JMC
To: PJH
I'm back at Stanford, I will be at IJCAI and also at the logic programming
meeting on the Queen Mary just before. See you.
∂29-Jul-81 1545 JMC visit
To: bledsoe at SRI-AI
Would it be convenient for me to visit you tomorrow AM with Kurt and
exchange views, either in parallel or in sequence about the S and P
and related problems? Friday would also be feasible. I understand
this is your last week at SRI.
∂29-Jul-81 1623 JMC
To: csd.brown at SU-SCORE
I met Fan in Peking and would be glad to see him again. I don't know about
half a day.
∂29-Jul-81 1750 JMC
To: csd.dbrown at SU-SCORE
I met Fan in Peking and would be glad to see him again. I don't know about
half a day.
∂30-Jul-81 1110 JMC
To: schauble.multics at MIT-MULTICS, human-nets at MIT-AI
It would seem that the problem described by Schauble is not with C but
with the 8086. If you must use 20 bit pointers on a machine with
16 bit arithmetic, there will be slowness regardless of language.
Presumably microprocessors with longer words weren't considered to
be available when Bell made its choice.
∂02-Aug-81 1329 JMC
To: energy at MIT-MC
Is Kerns's reference to the actual Edward Teller or some canonical Edward
Teller whose position on all issues is extreme by definition? Edward
Teller's recent book "Energy from Heaven and Earth" takes positions that I
would characterize as intermediate between Kerns's and Pournelle's. If
the real Teller is meant, what positions are regarded as extreme?
∂03-Aug-81 1353 JMC New technical reports
To: admin.librar at SU-SCORE
1. The list of new technical reports looks very useful but also formidable.
How many months does that list cover?
When referring to a file at SAIL it is helpful to keep the reference all
on one line as in NEWTRS[LIB,DOC]. That way the E command XPO allows the
user to switch to that file without retyping its name. You wrote NEWTRS
[LIB,DOC] so the command didn't work till I changed it.
Anyway I think the new system will be very helpful.
∂03-Aug-81 1421 JMC
To: admin.library at SU-SCORE
1. The list of new technical reports looks very useful but also formidable.
How many months does that list cover?
2. When referring to a file at SAIL it is helpful to keep the reference all
on one line as in NEWTRS[LIB,DOC]. That way the E command XPO allows the
user to switch to that file without retyping its name. You wrote NEWTRS
[LIB,DOC] so the command didn't work till I changed it.
Anyway I think the new system will be very helpful.
∂04-Aug-81 0138 JMC Meeting with Martin Marietta that may be worth a grant
To: TOB
Are you available tomorrow at 1 or 2 for this purpose?
∂04-Aug-81 0143 JMC
To: TOB
OK, my office
∂04-Aug-81 0144 JMC
To: pourne at MIT-MC
Certainly. I'll miss Sunday, that's all.
∂04-Aug-81 0145 JMC
To: JMM
CC: JK
Please check promptly whether the new EKL does append properly.
∂04-Aug-81 1048 JMC
To: REM
I don't recall whether it even mentions solar power satellites; it certainly
doesn't emphasize it.
∂04-Aug-81 1256 JMC
To: csd.genesereth at SU-SCORE
Surely you will be criticized for not calling it MS.
∂04-Aug-81 1539 JMC tomorrow
To: LLW
I can come at 10 or at 2 your convenience. Besides seeing the S-1 Project,
I would like to talk with you about a variety of topics.
∂04-Aug-81 2340 JMC
To: LLW
I'll come at 2pm unless I hear otherwise from you.
∂05-Aug-81 0058 JMC
To: ROY
The imlac vertical deflection has conked out.
∂05-Aug-81 0141 JMC
To: ROY
It recovered after a rest.
∂05-Aug-81 1941 JMC
To: csd.tajnai at SU-SCORE
CC: PMF at SU-AI
Yes he did give a seminar and it did meet the requirement.
∂05-Aug-81 1944 JMC
To: JMM
Let's talk about it, but in the meantime, you should prepare versions doing
it both ways. My intuition is that one should be able to have the best of
both worlds - get both from one proof and make the simplification rules do
the verifying.
∂06-Aug-81 1040 JMC
To: CET at SU-AI, DDY at SU-AI, RPG at SU-AI, PAM at SU-AI,
ROD at SU-AI, FFL at SU-AI
Since the issue is important to some people, let me reverse my decision.
We'll continue AI memo numbers for at least another year. Fran will
assign them.
∂06-Aug-81 1350 JMC
To: JMM
It might be best to do the proof of termination first and then use this
to validate a reinterpretation or slight transformation of the function
definition so that it is assumed to be total.
∂06-Aug-81 1355 JMC
To: JMM, JK, CLT, FGA
Meeting Friday at 1pm about program property proving for CS206.
∂06-Aug-81 1614 JMC adjunct
To: TOB
It seems to have fallen into a crack between Ullman and Golub, but
Golub now promises to find out its state and pursue it.
∂08-Aug-81 1046 JMC meeting
To: darden at SUMEX-AIM
Will you be in Boston Tuesday evening August 18, and would you like to
get together then? I will come in that day for a meeting on the following
day and go to L.A. Wednesday evening.
∂08-Aug-81 1436 JMC
To: FFL
dreyfu.1
∂08-Aug-81 1556 JMC
To: CLT
∂08-Aug-81 0254 JMM associativity of append
To: JMC, JK
CC: JMM
The proof of assoc of apppend using the approach suggested by Carolyn
is ready - prf3.ppr[ekl,jmm]. It is 55 lines of which the first 30 lines
are general declarations and axioms etc and 25 lines are specific to
append .This proof did not use the more powerful commands like DECSIMP
and can be tightened to about 15 lines( I think !)
-- Jitendra
Please look at his proof. Even 15 lines looks like too much, and I
think you can easily make suggestions either for improving the proof
or EKL.
∂08-Aug-81 1649 JMC
To: CLT
OK, though I think you'd find what JMM has done transparent.
∂08-Aug-81 1702 JMC jmm
To: CLT
He is shy enough already and probably easily discouraged even though quite
smart. He'll soon learn that some regard his efforts as irrelevant and
will stop offering ideas.
∂08-Aug-81 1751 JMC
To: JK
I now recall the meeting was in Tampere. Is there such a place?
∂08-Aug-81 1757 JMC
To: rguest at UTAH-20, hearn at RAND-AI
To Herbert Stoyan:
What dates do you have in mind for Stanford visit? I'll be gone from
August 18 to August 28. John Allen will probably be accessible. Whom
do you wish to visit at Xerox?
∂08-Aug-81 1813 JMC
To: cbf at MIT-MC
Password is 848-0898.
∂09-Aug-81 1437 JMC
To: JMM
The proof seems to be all one line.
∂10-Aug-81 0120 JMC advice request on Chinese inquiry
To: engelmore at SU-SCORE
Included in a letter from Ma Xiwen, who visited here from Peking
University, was the following:
"Recently, I'm trying to set up an AIL in the Beijing Institute
of Computer Science (different from Prof. Tang's Institute). I
want to know if I can get some softwares as FOL or PASCAL verifier
(if we have MACLISP or SAIL or PASCAL or even a computer which is
consistent with PDP-10); I would like to hear any suggestion on it."
I have no idea how he proposes to get a computer "consistent with
PDP-10", and I have no opinion on the desirability of giving him
the programs he is interested in, except to say that
Ma is a basic researcher by background and that the interests
expressed elsewhere in the letter are in basic research topics.
Is anyone in DARPA likely to be interested in this inquiry? Is
there a definite policy or are cases decided individually?
I don't suppose there is any great rush to answer the inquiry. When
I next write, which probably won't be very soon, I am inclined to say
that I am making inquiries about the possibility of giving him FOL
but that the other programs belong to other organizations, which I
would name.
∂10-Aug-81 1122 JMC
To: csd.engelmore at SU-SCORE
Included in a letter from Ma Xiwen, who visited here from Peking
University, was the following:
"Recently, I'm trying to set up an AIL in the Beijing Institute
of Computer Science (different from Prof. Tang's Institute). I
want to know if I can get some softwares as FOL or PASCAL verifier
(if we have MACLISP or SAIL or PASCAL or even a computer which is
consistent with PDP-10); I would like to hear any suggestion on it."
I have no idea how he proposes to get a computer "consistent with
PDP-10", and I have no opinion on the desirability of giving him
the programs he is interested in, except to say that
Ma is a basic researcher by background and that the interests
expressed elsewhere in the letter are in basic research topics.
Is anyone in DARPA likely to be interested in this inquiry? Is
there a definite policy or are cases decided individually?
I don't suppose there is any great rush to answer the inquiry. When
I next write, which probably won't be very soon, I am inclined to say
that I am making inquiries about the possibility of giving him FOL
but that the other programs belong to other organizations, which I
would name.
∂10-Aug-81 1523 JMC
To: FFL
stich.1[let,jmc].
∂10-Aug-81 1839 JMC
To: FFL
Please check that my reservations are all on existent flights.
∂11-Aug-81 1208 JMC
To: rguest at UTAH-20
To: Herbert Stoyan
1. Friedman and Wise are at the University of Indiana in Bloomington,
Friedman, Daniel 812 337-4885
2. I will try to round up some people interested in the history of LISP.
3. Do you need help with travel from the airport, hotel, etc?
4. Where are you at the moment? How can I telephone you?
5. My telephone numbers are: home 415 857-0672, office 415 497-4430.
∂11-Aug-81 1538 JMC
To: darden at SUMEX-AIM
Have a good camping trip. Don't get bitten by too many mosquitoes.
∂11-Aug-81 1732 JMC seminar Friday
To: FFL
Please find a room, say 252, for a seminar.
Special seminar: logic programming
Speaker: Alain Colmerauer, University of Marseilles, France (Professor
Colmerauer is the inventor of PROLOG).
Time: 3pm, Friday, August 14
∂11-Aug-81 1745 JMC
To: FFL
Should Cuthbert Hurd come looking for some books, they're on my file cabinet.
∂12-Aug-81 1514 JMC Poole on file server
To: REG at SU-AI, VRP at SU-AI
1. F5 with 256K and ethernet 55-65K.
2. 20K per 670 megabyte disk
3. 30-35K per 125 in/sec 6250bpi tape drive. He'll look into "streaming mode"
drive.
4. delivery could be 4 months from now
later will be 75 days delivery
5. substituting F4 (about 2060 speed), raises basic cost to 125 to 130K
(F4 with megaword, usual peripherals is 300K)
6. will look into bigger disks, claims all RP drives are CDC. He can get
them but sometimes with longer deliveries.
Therefore if a system with 4 gigabytes, the F4, one 6250 bit drive and
256K main memory will do the job, our cost would be $220K. We might
need more memory, and we might want more disk to start. I would prefer
larger disks, and Poole will look into it, but I think this constitutes
an existence proof, so we should proceed promptly to get a decision to
build a file system.
∂12-Aug-81 1520 JMC
To: GIO at SU-AI
∂12-Aug-81 1514 JMC Poole on file server
To: REG at SU-AI, VRP at SU-AI
1. F5 with 256K and ethernet 55-65K.
2. 20K per 670 megabyte disk
3. 30-35K per 125 in/sec 6250bpi tape drive. He'll look into "streaming mode"
drive.
4. delivery could be 4 months from now
later will be 75 days delivery
5. substituting F4 (about 2060 speed), raises basic cost to 125 to 130K
(F4 with megaword, usual peripherals is 300K)
6. will look into bigger disks, claims all RP drives are CDC. He can get
them but sometimes with longer deliveries.
Therefore if a system with 4 gigabytes, the F4, one 6250 bit drive and
256K main memory will do the job, our cost would be $220K. We might
need more memory, and we might want more disk to start. I would prefer
larger disks, and Poole will look into it, but I think this constitutes
an existence proof, so we should proceed promptly to get a decision to
build a file system.
∂12-Aug-81 1527 JMC
To: FFL
I forgot to ask you to post announcements of Colmerauer seminar.
∂12-Aug-81 1654 JMC
To: GIO at SU-AI
My proposal is for Ralph Gorin to "take on the hardware". I believe we can
pay for it with Computer Facility funds, and I would prefer to proceed
promptly. From my own point of view, which regards purges and the removal
of old reports from the file system as the primary problem to be overcome,
the file computer would be a big improvement at this price even if all
transfers to it were done by explicit FTP by the user.
Brian Reid has software at SCORE and perhaps on the VAXen that permits
a file computer to be treated like a user disk pack. Something similar
can be done at SAIL. No one has automatic migration yet. Your plan
to store information rather than files (if I understand it) would require
further adaptations. The other schemes mentioned are all compatible with
starting with FTP and maintaining that capability while more advanced
facilities are implemented. I believe that you would also have to work
in that mode. There needs to be a meeting to sort out the various ideas
for file service and see which can be implemented.
My understanding is that the IBM 3380 is going to be substantially delayed,
and there is no present sign of cheap imitations.
∂12-Aug-81 1714 JMC
To: csl.bkr at SU-SCORE
∂12-Aug-81 1520 JMC
To: GIO at SU-AI
∂12-Aug-81 1514 JMC Poole on file server
To: REG at SU-AI, VRP at SU-AI
1. F5 with 256K and ethernet 55-65K.
2. 20K per 670 megabyte disk
3. 30-35K per 125 in/sec 6250bpi tape drive. He'll look into "streaming mode"
drive.
4. delivery could be 4 months from now
later will be 75 days delivery
5. substituting F4 (about 2060 speed), raises basic cost to 125 to 130K
(F4 with megaword, usual peripherals is 300K)
6. will look into bigger disks, claims all RP drives are CDC. He can get
them but sometimes with longer deliveries.
Therefore if a system with 4 gigabytes, the F4, one 6250 bit drive and
256K main memory will do the job, our cost would be $220K. We might
need more memory, and we might want more disk to start. I would prefer
larger disks, and Poole will look into it, but I think this constitutes
an existence proof, so we should proceed promptly to get a decision to
build a file system.
∂12-Aug-81 2322 JMC regular expressions formed from rewrite rules
To: JK, JMM, CLT, FGA, RWW
Some useful compound rewrite rules can be expressed by regular expressions
in simpler rules.
1. If r1,...,rk are rules, r1+ ... + rk means apply some ri.
2. r1. ... .rk means apply the ri successively, rk first.
3. r* means apply the rule r as long as possible.
The usual way of combining rules corresponds to the regular
expression (r1+ ... +rk)*. If we want to apply some rule such
as the definition of a LISP function only once, we may want
combinations like (r1+r2)*.recdef.(r1+r2)* which uses the rule
recdef exactly once and the others as much as possible.
So far this seems to me to be only a notational convenience rather
than a relation between the theory of regular expressions and
the theory of rewrite rules. Other operators on rules may also
be wanted. For example, as Jitendra also suggested, we may want
to regard using an equation in the opposite direction as a formal
inverse. Other formal inverses can be written, but probably require
some problem solving capability to be used, but the system can simply
regard an inverse as inapplicable if it is incapable of solving the
problem presented.
We may also want to be able to express control over what parts of
an expression are to be rewritten. Thus in proving the associativity
of append we want to attack only the first *.
Needless to say, we would need to be able to write such definitions as
r4 ← (r1+r2)*.r3.(r1+r2)*
and then be able to use r4. We may even want to parametrize such definitions
so that we can readily use a rule which consists mainly of general lisp
definitions but takes the definition of a function under study as a
parameter. Thus we might write
defsimp(fndef) ← R1.fndef.R2
and then use defsimp(appenddef) when proving facts about append.
We can discuss this today at our 11am meeting.
∂13-Aug-81 1333 JMC
To: sharon.burks at CMU-10A
If Nov. 17 or 19 is feasible, I could combine it with another trip to the
East Coast. Otherwise November 4 is best. The title will be "Non-monotonic
reasoning and ambiguity tolerance".
∂13-Aug-81 1416 JMC
To: JK
I tried to declare car as a prefix so that the number of parentheses
in formulas would be reduced. It didn't work. The declaration was
(DECL (CAR CDR) |GROUND→GROUND| CONSTANT nil prefix 1010)
but (assume |a = car x|) and (assume |a = (car x)) both generated
the message "expected another op before X". Allowing one variable
functions as prefix operators greatly reduces the parenthesis level
and increases the readability of expressions.
∂13-Aug-81 1433 JMC
To: JK
(∀e phi |λv.(v=copy(v))| 19) gets ;UNBOUND ARGUMENT MUST BE PROPER LIST
∂13-Aug-81 1437 JMC
To: JK
The above occurs after dskin of lisp.ax[e81,jmc].
∂13-Aug-81 1659 JMC
To: konolige at SRI-AI
Was it you wanted Sato thesis which is now available.
∂13-Aug-81 1720 JMC
To: konolige at SRI-AI
ok
∂13-Aug-81 2158 JMC
To: JK
Thanks. Perhaps RPG can get something done about the compiler bug.
∂14-Aug-81 1404 JMC
To: bobrow at PARC-MAXC
I agree with all your proposals and plan to be at the meeting.
∂14-Aug-81 1445 JMC
To: JK
Now that I have gotten a bit used to it, I find EKL very pleasant.
∂15-Aug-81 1333 JMC
To: REM
Everyone has been zapped including me. It is being fixed.
∂16-Aug-81 1304 JMC
To: hearn at RAND-AI
What is Stoyan's flight?
∂16-Aug-81 1748 JMC
To: CDR
Are you interested in Paella dinner with me and Colmerauer, etc. 6:30?
∂16-Aug-81 1805 JMC
To: CDR
Too late. I had to get someone else.
∂16-Aug-81 2015 JMC
To: hearn at RAND-AI
I lost the paper, but I found the flight number and will meet him.
∂16-Aug-81 2332 JMC
To: hearn at RAND-AI
Gottim. What's the current financial arrangement?
∂17-Aug-81 0205 JMC se2
To: csd.genesereth at SU-SCORE
I forgot to ask you if you were interested in joining Stanford SE2 (Scientists
and engineers for secure energy or coming to a meeting. We meet rather
rarely, and the next will be late September or early October. I am
chairman, and Professor Elliott Bloom of SLAC is secretary. If you
like, I will ask him to put you on the list to be notified of the next
meeting.
∂17-Aug-81 1156 JMC
To: sharon.burks at CMU-10A
My other commitment in the East is on the 18th, but I could perhaps
come a day early.
∂17-Aug-81 1201 JMC
To: csd.tajnai at SU-SCORE
1:15-2:30 for CS258 is fine.
∂17-Aug-81 1202 JMC
To: csd.tajnai at SU-SCORE
That's what I thought you meant.
∂17-Aug-81 1306 JMC
To: "@SERVER.LIS[E81,JMC]" at SU-AI
I did send my message to Brian Reid and had a long conversation with him.
I think we agree about enough things. In my view it is essential to
subordinate the goal of pursuing research interests with that of relieving
the file crunch and give the latter priority for CCF funds. Of course, if
we can't rais facility money and do get a research contract, then the
reverse priority might be required.
However, I think that so long as we implement FTP to the file server
promptly, we can pursue Brian's goal of allowing page-at-a-time reference
to files which is clearly much better. Moreover, Brian is far enough
advanced in the latter and sufficiently sympathetic to the former that I
think there will be no problem. Gio Wiederhold also has research goals in
this area, but because they involve databases and not merely files, it
isn't so clear that they fit into the file server project itself, though
it might be feasible to put some hooks into the file server for his
benefit.
With regard to the conversation with Poole, I think we need one more
thing than was mentioned in my message. While various disks may be
most cost-effective at present, most likely disks based on the IBM
3380 technology will be most cost-effective fairly soon, and others
beyond that later. Therefore, it is very desirable that the file
server have some reserve in the speed of file transfer that it allows.
I will ask Poole to add this question to those he has promised to
look into.
I believe that it is possible to promptly develop a plan that will meet
the various needs. The big issue is whether it is to be financed by
the CCF borrowing from Stanford and adding the amortization and expenses
to the rate base or whether a proposal to ARPA or NSF is required. I
encountered Bill Massy (Stanford V.P. for finance) at Tressidder and
he is agreeable in principle to Stanford financing it. I favor self-
financing, because getting gov't money will take a relatively long time.
∂17-Aug-81 1339 JMC
To: minsky at MIT-AI
Can I stay at your house Tuesday night before III board meeting.
∂17-Aug-81 1857 JMC directory for Herbert Stoyan
To: csd.hill at SU-SCORE
I have created 1,hxs for a guest Herbert Stoyan. It can be killed
after Aug. 21.
That's right only for 2 days
∂18-Aug-81 0304 JMC transparencies
To: FFL
We need to stock the kind of transparency material on which one writes
with a special pen and also to stock the pens. See you in two weeks.
∂18-Aug-81 1611 JMC via MIT-MC definitions via eval
To: CLT
I worked on compiling function definitions on the
and the arguments inhabiting the initial a-list. Mutual
recursion offers little problem apparently, and LABEL is
unneeded. A lot of simplification is required to get the
recursion equation, and I have some new ideas on how to
control simplification. Maybe more later.
∂18-Aug-81 1636 JMC
To: CLT
If the simplifications are individually legal, no metatheorems
need be proved to justify a control strategy for using them.
The results will be correct, and the only question is whether
they are what the user wants.
∂19-Aug-81 0835 JMC via MIT-AI progress on apply
To: CLT
You may find it interesting to look at apply.ax[e81,jmc] which contains and
incomplete sketch of my present approach. I don't expect another chance
at the ARPAnet before I get back, but I'll phone you soon whether there is anything
more to say or not.
Love,
∂28-Aug-81 2155 JMC
To: csd.crangle at SU-SCORE
Yes, I have seen it recently, and I'll send you a message when I've given it
to Fran for you.
∂28-Aug-81 2158 JMC
To: FFL
Please call Dugan about this for me.
∂21-Aug-81 0915 FFL Call from E. Bloom
To: JMC, FFL
Panofsky has been accosted by intervenors. He needs facts on Diablo..
1. How many hearings have been held to determine earthquake proofness
of the plant by the NRC?
2. Final statement of the last hearings with respect to the safety of the plant.
E. Bloom asks that you prepare packet and get to Panofsky as quickly as possible.
He will need the material when in gets back from vacation in 2 weeks. Bloom is
now going away.
∂28-Aug-81 2159 JMC
To: FFL
OK, I'll check with Dugan.
∂28-Aug-81 2204 JMC
To: RPG at SU-AI
I was on a trip; I hear bad things were proposed and good things neglected.
∂29-Aug-81 1353 JMC suggestion for line editor
To: ME
How about a delayed <cr>? It would result in a <cr> when the program
received the line, but the line wouldn't be transmitted until a real <cr>
was typed. It is often convenient to prepare several lines of input to a
program at once, keeping the ability to edit them and using only one line
of screen display. If it is difficult to think of an available character,
there are several possibilities. One is to have a command (includable in
OPTION.TXT) that will set a character, and another is to allow the user to
define an <esc><number><cr> that would produce the delayed <cr>.
∂29-Aug-81 1511 JMC mailing list and file server
To: VRP at SU-AI, csl.bkr at SU-SCORE
Please put me on the list. I thought I was a member of the future
hardware committee. You are using the wrong numbers about the
non-research file server. The Foonly quote is $220K, not $500K, and the
number of disks is six not three. The storage amount is four gigabytes.
I just received a letter from Gordon Bell in response to my grumble about
RP20 unavailability on VAX, and it merely re-iterates the unavailability,
expresssing regret. The early $500K estimater referred to VAX, and I fear
it still applies. Is the convenience of 32 bits worth that much money?
I fear that an estimate of 6 million for hardware (60 full professor years
including overhead) is much more than we're likely to get for hardware
from DARPA and more than we would allocate to hardware were the money our
own. It is high priority to get some feeling for what we're actually
likely to get so that we don't buy early items because of quick
availability or ease of specification and then find ourselves with
important needs unmeetable.
I will send an expanded version of this message to the EQUIP list as soon
as someone tells me how to do this from SAIL.
∂29-Aug-81 1540 JMC prolog
To: stan at SRI-AI, admin.mrc at SU-SCORE
I received David Warren's permission to update my PROLOG from the SRI
version. Whom do I get Mark Crispin to talk to about what files to FTP
and related setup information? I see from a letter from David that
the current version is 3.3.
∂30-Aug-81 1415 JMC
To: pourne at MIT-MC
CC: JPM at SU-AI
There is no JPTM at SCORE. The only person who might have made
"an attractive offer" to L-5 that comes close to your description
is Jim McGrath, who is known as JPM. He is away to the summer, but
I'll send him a message asking if it's he.
∂30-Aug-81 1657 JMC
To: admin.library at SU-SCORE
Do you want my issues of Avtomatika i Telemekanika?
∂31-Aug-81 0205 JMC wanted person
To: FFL
Someone came around a few weeks ago looking for a Russian emigre who
had given a seminar on brain models. The speaker was Victor Eliashberg
who works at Varian, but now I've forgotten who was looking for him.
∂31-Aug-81 1418 JMC
To: admin.library at SU-SCORE
For Harry Llull:
I have no special opinion about who will read the journal. They send it to
me free (in Russian), but I haven't found anything appropriate to my
interests. If any Stanford library is interested I can forward them.
Otherwise I'll write them to stop sending it. I'll send you an issue.
∂01-Sep-81 0107 JMC
To: FFL
Please put BINFOR.RE2[LET,JMC] in Scribe.
∂01-Sep-81 1337 JMC
To: csd.golub at SU-SCORE
thanks
∂01-Sep-81 1341 JMC
To: FFL
Please send Stoyan time.his[s80,jmc].
∂01-Sep-81 1446 JMC
To: RPG
Please inspect EQUIP.MEM[E80,JMC] before I send it to EQUIP.
∂01-Sep-81 1446 JMC
To: RPG
Are you here now?
∂01-Sep-81 1601 JMC
To: RPG
∂29-Aug-81 1425 REG
∂21-Aug-81 1251 CSD.PRATT@SU-SCORE via Ethernet
Date: 21 Aug 1981 12:43:46-PDT
From: CSD.PRATT at SU-SCORE
Mail-from: ARPANET host SU-SCORE rcvd at 21-Aug-81 1248-PDT
Date: 21 Aug 1981 1243-PDT
From: Vaughan Pratt <CSD.PRATT at SU-SCORE at SUMEX-AIM>
Subject: mailing list
To: equip at SU-SHASTA
I've created a mailing list, equip@shasta, consisting of the seven people
who attended this morning's meeting. I'm sending this message from
Score to establish that you can mail to equip@shasta from Score. I'll
check whether this can be done from Sail.
Vaughan
-------
∂21-Aug-81 1822 reid@Shasta via Ethernet
Date: 21 Aug 1981 18:19:01-PDT
From: reid at Shasta
To: equip
Subject: reformatted table
I Scribe-ified Jeff's message so I could read the table (it came through
without tabs in my copy of the message). Here's the reformatted words:
Culminating an effort begun by Ralph Gorin and a committee from CSD to
probe the needs of the department (and by ``department'' I include the people
in CSL as well), the following sketch of departmental computer and equipment
needs over the next three years has emerged. These items represent our best
estimate of realistic needs; they do not constitute a ``wish list.''
However, it is unlikely that we can find the money to purchase all items on
the list without considerable fundraising effort on our part.
ITEM 1982 1983 1984 TOTAL
Dec 2080: 1@1.2M $1,200 $1,200
Vax 11/780: 4@400K $400 $800 400 1,600
Dolphins: 15@60K 600 300 900
Large-scale File Store 500 500
SUN Stations: 300@5K 500 500 500 1,500
Ethertips: 20@3K 30 30 60
Ethernet 25 25 50
Terminals 100@1K 50 50 100
Laser Printers: 7@30K 90 60 60 210
Total $2,195 $1,765 $2,160 $6,120
Certain items, such as the file store, terminals, SUN workstations, ethernet
cable, and printers are for general consumption of the department. Other items
have been identified with particular research activities. In particular, the
intent is that the Dolphins would be primarily for the use of HPP, as would two
of the Vaxen. Another Vax would be for systems activities, and the fourth would
be available for activities such as TEX, database research, and general
departmental needs. These Vaxen are in addition to proposed purchases for the
Numerical Analysis group and the Robotics group, that will take place in early
1982.
The purchase of a 2080 in 1984 is problematical, depending on our apparent
needs for large-scale time-sharing services once the smaller scale items such
as Dolphins and Vaxen are available to the community of users.
At the present time, we have $150K available from Stouffer for the purchase
of SUN terminals. The $5K price of the SUN's is an estimated average,
reflecting ultimate economies of scale; initial purchases will be higher.
The file store will very likely be part of a McCarthy-Reid research project
into the utilization of such systems. All other items are intended to be purely
service machines.
∂22-Aug-81 1413 pratt@Shasta via Ethernet
Date: 22 Aug 1981 14:08:14-PDT
From: pratt at Shasta
To: equip
Subject: Sun pricing
I think the $5k figure for Suns should be raised to $7k until we know more
certainly that they can be obtained for $5k, which at present is wishful
thinking. The $7k figure is already very low, representing
approximately twice parts cost. The retail price would be in the range $10k
to $15k depending on the company involved (companies vary between 3 and 5
times parts cost for a profitable retail figure, a ratio that depends both on
the organization of the company and the details of the product). A reduction
to $5k would be possible only with a combination of a benevolent manufacturer
and high volume.
∂22-Aug-81 1453 pratt@Shasta via Ethernet
Date: 22 Aug 1981 14:49:42-PDT
From: pratt at Shasta
To: equip
Subject: file store
(Metafact: a log for this mailing list is accumulating in
[shasta]/usr/spool/mail/equip - its first message is Jeff's with Brian's tabs.)
In addition to the $500k for the research project on an intelligent file
server, I would like to see a budget item for a non-intelligent non-research
file server.
Item
10 localnet file servers at $10,000 each for each year. (Hence 30 file
servers after three years.)
Planned construction: 100 megabyte 8" Winchester (first year, larger for
subsequent years), 68K processor board, Ethernet interface, disk controller.
By the time we have a disk controller ready we should be able to get all
these parts for $7000 plus or minus $1000. The small volume permits these
items to be constructed in house, whence the rather low overhead.
My present thinking is to put two on each floor's local net per year,
including the CSL floor in ERL. These file servers would contain files
accessed within the order of a month or two. Files would migrate from file
servers to other local nets:
(a) to be dumped to tape for backup (incremental only, but with each file
being dumped at least twice on separate tapes, on day of creation and two days
later for security)
(b) for archival of old files to free space on the local file servers
(c) on demand by users on other local nets.
∂22-Aug-81 1512 ullman@Shasta via Ethernet
Date: 22 Aug 1981 15:08:53-PDT
From: ullman at Shasta
To: equip, pratt
Subject: Salaries
Our general strategy has been to charge salaries for maintainance of
equipment to CSD-CF and figure the charges into computer costs to be
paid by grants and contracts. While we have not done so in the past,
a faculty member's time spent working for CSD-CF on SUPPORT (not
development) could be factored in. Undoubtedly the total staff cost
for CSD-CF is going to go way up. If a lot of the equipment is paid for
by ARPA, depreciation costs could be down in compensation, so we don't
necessarily overspend our collective computing budgets.
∂22-Aug-81 1512 pratt@Shasta via Ethernet
Date: 22 Aug 1981 15:03:18-PDT
From: pratt at Shasta
To: equip
Subject: Salaries
We need to estimate salary requirements to maintain all this equipment.
We need at least a couple of technicians for repair work on Suns, and there
should also be someone in charge of Vaxes and Dolphins, if not two people.
In addition I seem these days to be making the coordination of production of
equipment a full time activity, and therefore an appropriate percentage of my
Stanford salary should come from funds allocated for equipment-related
salaries. (The NSF is down to 10% during the academic year for us
theoreticians these days.)
∂23-Aug-81 0133 reid@Shasta via Ethernet
Date: 23 Aug 1981 01:23:52-PDT
From: reid at Shasta
To: equip
Subject: file server
$500K was John McCarthy's figure for a non-research file server
consisting of a Foonly F5, 3 big disks, and a fast magnetic tape
subsystem. I am very dubious of budgeting for file servers
built out of 68000's unless we budget a full-time programmer
to make them work reliably.
∂23-Aug-81 1207 GABRIEL at CMU-20C Hardware
Date: 23 Aug 1981 1458-EDT
From: GABRIEL at CMU-20C
Subject: Hardware
To: reg at SU-AI
I notice that there was a hardware meeting to ask, again, the question
of what the dept needs. Since I was on "that" committee before,
I would like to make a few comments related to the LISP computing
needs of the community, of which HPP is a part.
First, the Common Lisp dialect will run on the following machines:
VAX-n, S-1, PERQ (extended) and LISP Machine. The Lisp Machine
version will be existent firt, then the PERQ, then S-1, then VAX.
I think that if we are going to buy Dolphins (to cover HPP needs only,
which questionable coverage for a dying language) then I think we
need Lisp Machines too in order to give the dept Common Lisp coverage.
In case you don't know what Common Lisp it, it is the merger of S-1
Lisp, VAX NIL, SPICELISP, and Lisp Machine Lisp (Alphalisp), a
descendent of MacLisp. This dialect will surely dominate the next
10 years Lisp usage, and may, in the future, incorporate InterLisp.
I would recommend that the purchase of Dolphins be reduced and
some number of Symbolics Lisp Machines substitute. If HPP wants
to buy them, that's fine, but I don't see the department's need
for Dolphins to the exclusion of LISP Machines, given the existence
of COmmon Lisp.
-rpg-
-------
∂23-Aug-81 1216 pratt@Shasta via Ethernet
Date: 23 Aug 1981 12:11:25-PDT
From: pratt at Shasta
To: equip, reid
Subject: file server
From reid Sun Aug 23 01:23:12 1981
To: equip
Subject: file server
$500K was John McCarthy's figure for a non-research file server
consisting of a Foonly F5, 3 big disks, and a fast magnetic tape
subsystem. I am very dubious of budgeting for file servers
built out of 68000's unless we budget a full-time programmer
to make them work reliably.
1. The Sun project already has some very good programmers. If they can't
produce file server software it will be the first time I've seen something
lick them. More generally the whole issue of software for the Sun should be
treated by this committee as internal to the operation of the Sun project.
The committee should not feel obliged to find funding for Sun software
separately from finding funding for the Sun project as a whole. (Since the
purpose of the Sun project is to develop equipment for Stanford, it seems to
me that an equipment committee should give some thought to contributing to
the funding of the development aspects of the project.)
2. Sun software is written in C. This substantially eases the task of
developing and maintaining systems software, as well as contributing to its
reliability.
3. A set of decentralized file servers is inherently more reliable than a
single centralized file server in two ways: a malfunction of one file server
affects only a fraction of the users, and redundant copies of files may be
maintained on separate machines as opposed to merely separate cylinders or
separate disks. Thus I predict that it will not require a giant software
effort to get decentralized file servers up to an acceptable level of
reliability.
4. The present Ethernet organization will continue to work only for as long
as it remains lightly loaded. We are firmly committed to moving to a
localnet organization of the Ethernet as soon as possible. Local file
servers are an essential component of this reorganization; neither gateways
nor the CSD Ethernet spine will support the load that a centralized file server
would impose.
5. I am not proposing Sun file servers instead of a centralized file server
but in addition to one. The former will have a longer lead time than the
latter - I know for certain that the Sun project does not presently have the
resources to produce the necessary hardware before February at the very
earliest. Instead I envisage that an initial centralized file server will
gradually be phased into a more archive-oriented file server as soon as
localnet file service appears and starts being used. (You were wondering
perhaps where I was going to put the month-old files I said would migrate off
localnet file servers.)
On the other hand I do not endorse the F5, primarily for software reasons - I
predict that it will be a software maintenance headache in the long run. My
objections would be less vehement of if we could run most of the same software
on the F5 as on Vaxes and 68000's. A major reason for our having a large
amount of 68000 software is that it was ported with little effort from Vax and
PDP-11 software. Moreover we maintain increasingly more Vax and Sun software
as a unit; changes are made to one source, and tested on both machines. This
helps us substantially in getting along with the present level of software
personnel. Thus I would strongly recommend that any file server we acquire
be able to run all software that runs on both the Vax and the Sun.
(Inevitably there will be machine dependencies, but these are routinely taken
care of with compile-time conditionals in the case of Vax and Sun machine
dependencies, and the same will apply to other machines.) For the F5 to fit
this requirement it would have to support C - to be exact the version of C
running under Berkeley Vax Unix, which is what we have recently moved Suns to
for maximum portability.
My preference for a centralized file server processor would be a 32-bit
machine with speed and pricing on the same scale as the F5 and offering C.
The above is not meant as an unconditional endorsement of C as a programming
language, but rather is based on the needs of our present software
development and maintenance environment. C has more failings than I like to
think about.
6. This is not the sort of item that would bring the budget to its knees,
for the first year we're only talking about 20% of the proposed cost of a
centralized file server. (My prices are based on estimates of 100 megabyte
Winchester pricing for 1982, today it would be more than double. Price per
bit on small Winchesters is falling very rapidly.)
∂23-Aug-81 1217 GABRIEL at CMU-20C Common Lisp Blurb
Date: 23 Aug 1981 1508-EDT
From: GABRIEL at CMU-20C
Subject: Common Lisp Blurb
To: reg at SU-AI
Here is the description of Common Lisp to be given out at IJCAI:
@make [text]
The following is a DRAFT of a blurb to be distributed at the IJCAI.
A common subset of Lisp, called "Common Lisp", is being defined by the
designers of several new Lisp dialects, to provide portability among the
different dialects.
Several dialects of Lisp descended from Maclisp are in various stages of
development:
@begin [itemize]
Spice Lisp is being developed as a part of the SPICE personal computing
project at Carnegie-Mellon University (C.M.U.). The first implementation of
Spice Lisp, for an extended version of the PERQ computer from Three Rivers
Computer Corporation, is currently under way.
NIL (New Implementation of Lisp) is being developed for the S-1 computer
at Lawrence Livermore National Laboratory (L.L.N.L.) and for the Digital
Equipment Corporation VAX at the Massachusetts Institute of Technology
(M.I.T.). It is partially operational on both machines.
Alphalisp, on the M.I.T. Lisp Machine, has been in use for several
years.
@end [itemize]
All of these dialects are descendants of MacLisp; a great deal of
improvement, modernization, and development has occurred in each of them.
Until recently, development of each dialect was done by the maintainers
of that dialect, without much interaction with the maintainers of the
other dialects. Observers in the Artificial Intelligence community
rightly pointed out that this was leading to a group of incompatible
languages.
Representatives for each dialect are working on the definition of "Common
Lisp". Common Lisp is a powerful and modern Lisp dialect of which SPICE
Lisp, NIL, and Alphalisp are supersets. By writing a program within
Common Lisp, the programmer is assured that it will work in any of these
dialects. The effort is being led by Guy L. Steele Jr. and Scott Fahlman
of C.M.U.; Richard P. Gabriel of the Stanford Artificial Intelligence
Laboratory and L.L.N.L.; Jon L White of M.I.T.; Richard Greenblatt of
L.M.I., Inc., and M.I.T.; and David Moon and Daniel Weinreb of Symbolics,
Inc., and M.I.T.
Common Lisp will be be a complete and usable language in its own right,
and it will be well-supported by all of the above-mentioned
dialects; the participants are all committed to providing and maintaining
support for the entire subset. Common Lisp will also be very stable:
additions to the definition will only be made if there is general
agreement among all participants that such an addition should be made,
and incompatible changes will be avoided. New Lisp language features
that are still considered to be under development will not be added to
Common Lisp; only stable and well-tested features will be added.
A preliminary specification of Common Lisp is expected to be available
by about January 1982 from the Department of Computer Science at
Carnegie-Mellon University.
I hope to return within 2 weeks to Stanford, and would be happy
to discuss how these Lisp developments might affect Stanford's
policy towards hardware acquisitions.
-rpg-
-------
∂23-Aug-81 1537 reid@Shasta via Ethernet
Date: 23 Aug 1981 15:33:13-PDT
From: reid at Shasta
To: equip, pratt
Subject: file servers
I find a bit of an inconsistency in the following two (excerpted) messages:
From pratt Sat Aug 22 14:49:09 1981
To: equip
Subject: file store
In addition to the $500k for the research project on an
intelligent file server, I would like to see a budget item for a
non-intelligent non-research file server.
From pratt Sun Aug 23 12:09:42 1981
To: equip, reid
Subject: file server
3. A set of decentralized file servers is inherently more
reliable than a single centralized file server in two ways: a
malfunction of one file server affects only a fraction of the
users, and redundant copies of files may be maintained on
separate machines as opposed to merely separate cylinders or
separate disks. Thus I predict that it will not require a giant
software effort to get decentralized file servers up to an
acceptable level of reliability.
To my way of looking at things, the construction of something that nobody
has ever successfully done before is a research project. It is not enough to
know how to do something. I know how to build the world's finest document
preparation system; I know how to build fantastic distributed operating
systems and even more fantastic computer message systems. But I haven't done
any of these things yet, and neither has anybody else, and therefore I call
them research projects. Nobody has ever succeeded in building a reliable
fast distributed file system out of any brand of computer, and therefore by
my definition of a research project the 68000-based systems that you are
talking about is a research project.
The only file systems that have ever been constructed by anyone anywhere
that actually work, that are fast, that are reliable, and that are
inexpensive, are centralized file systems. I therefore assert that your
statements of desiring a non-research non-intelligent file server and of
desiring a distributed file system are mutually contradictory.
Naturally I support the distributed file service; naturally I am opposed to
having any kind of software built around a dinosaur of a machine like the
Foonly. Naturally I support your claim that the only sane way to program a
file server is in C on a 32-bit machine. I am only jumping on you for your
unstated belief in the equivalence of software that you think you know how
to write and software that has been written and is known to work.
Brian
∂23-Aug-81 2029 pratt@Shasta via Ethernet
Date: 23 Aug 1981 20:22:15-PDT
From: pratt at Shasta
To: equip, reid
Subject: file servers
Seems to me Brian and I are in perfect agreement. I retract my statement that
distributed file servers are not a research project. Yet another reason to
build them.
∂23-Aug-81 2014 pratt@Shasta via Ethernet
Date: 23 Aug 1981 20:13:39-PDT
From: pratt at Shasta
To: Equip@SHASTA, REG@Sail
Subject: Symbolics Lisp Machines
Dick Gabriel has a point. I'd stopped thinking about Lisp Machines because
they were so expensive, but the Symbolics 3600, successor to Symbolics' LM-2
Lisp Machine, with deliveries beginning March, at $60k in quantity, is
comparably priced to a Dolphin, and as Dick says runs the Lisp that is likely
to dominate the coming decade. More significantly, Symbolics claims that the
speed of a Symbolics 3600 is approximately that of a 2060 running Lisp, which
makes it from two to three times faster than a Dolphin. Although the Dolphin's
microcode has been tuned for Lisp, the Symbolics 3600 is a pedigreed Lisp
machine from the ground up. In light of this fact, an order for 15 Dolphins
merely to satisy the Lisp community sounds to me like a big mistake.
∂23-Aug-81 2351 pratt@Shasta via Ethernet
Date: 23 Aug 1981 23:50:25-PDT
From: pratt at Shasta
To: equip
Subject: msg from rpg
Here's a message from Dick Gabriel, at CMU now but returning in 2 weeks.
From GABRIEL@CMU-20C Sun Aug 23 22:35:32 1981
Mail-from: ARPANET host SU-SCORE rcvd at 23-Aug-81 2234-PDT
Mail-from: ARPANET site SU-AI rcvd at 23-Aug-81 2230-PDT
Date: 24 Aug 1981 0129-EDT
From: GABRIEL at CMU-20C
Subject: Response etc
To: vrp at SU-AI
Vaughn, can you please send this on to the equip mailing list.
There appears to be no way to get this to them via CMU(!).
Also, can you put me on that list? I was on that committee
before, and I think I ought to be on it now (between you and me).
Thanks.
---------------------------
In response to Ullman's message about LISP computing, I would like to
first point out that a very large percentage of the research that
goes on in the department that isn't theoretical is done in LISP. Such
groups are: HPP, Formal Reasoning, Hand-eye, and some of Zohar's students
(YM). In addition, much of the Verification work is done in LISP.
I believe several of these groups have some funding from ARPA.
Thus, the machine of choice that can execute LISP code is nearly
as fundamental as the machine of choice to run file servers.
Second, to buy Dolphins for HPP, which run essentially only
InterLisp, and to then ignore the needs of the rest of the LISP
community seems highly inconsistent. This is especially true
when 25% of the HPP projects use a non-InterLisp LISP (quoted
from Genesereth at the SRI LISP Meeting of last April).
Third, there are enough recent developments in the LISP community
that would influence the direction of LISP computing that I think
to buy a group of machines to (inadequately in my opinion) serve
a small community, when that group of machines may not be the most
forward thinking in terms of long term trends would be folly.
Fourth, the group of people, Feigenbaum, McCarthy, Buchanan,
and Rindfleisch do not represent the greatest expertise in LISP
and recent developments available in the department at this time.
Again, I would like to point out that the Common Lisp effort,
which I believe is unknown to the group of people thus far gathered
to investigate the issues, represents an important enough development
that it ought to be taken seriously.
I am currently at CMU working on the definition, design, implementation,
and documentation for Common Lisp. In 2 weeks I will return and
join the Formal Reasoning group, and at that time I would be more
than happy to help assess the relative merits of hardware for
LISP programming.
As far as LISP on a VAX in concerned, I have talked to Larry Masinter,
Jonl White, and several Franz users at CMU. In addition, I have
chatted with Len Bosack and all have essentially agreed that the
Vax represents a technically uninteresting alternative to the 10-like
computers except insofar as address space in concerned. The ISI VAX
InterLisp will not sustain a software development project (according to
Masinter)in InterLisp, nor will a NIL (read Common Lisp) do much better
except in some numeric code. Franz is barely usable as it is and
more than 1 Franz user on a 780 is sufficiently intolerable to drive
all but one LISP job off the machine. In short, the paging characteristics
render the VAX of any species at or below the 780 level untenable.
If anyone is seriouslyt considering buying some VAXes for LISP,then
a purchase of n/8 where n is the number of LISP users in the department
seems about right to me - 10 should do. I seriously doubt that a
750 would be tolerable even as a personal Lisp machine.
On an unrelated topic, there is an interesting development at CMU
relating to terminals: Leonard Zubkoff's data compression techniques
at the terminal/computer boundary. It has been observed that Concept-100's,
over a 9600 baud line have been driven at the effective rate of
36Kbaud (that's right, thirty-six kilo baud). As much as we would all
like to own the 320i's of terminals, I believe that the department ought
to supply the masses with reasonable, but cheap terminals. This I think
is in contrast to a medium number of high-quality terminals and a medium
number of non-existent ones.
-rpg-
-------
∂23-Aug-81 2354 pratt@Shasta via Ethernet
Date: 23 Aug 1981 23:52:14-PDT
From: pratt at Shasta
To: equip
Subject: rpg
As you can see, Dick has a point of view about Lisp that is not adequately
represented on the committee. I think the majority of his views are
technically sound. He says he was on this committee's predecessor, and would
like to contribute to this one too. What's the vote?
∂23-Aug-81 2118 ullman@Shasta via Ethernet
Date: 23 Aug 1981 21:17:02-PDT
From: ullman at Shasta
To: equip
I'm pleased to find all my colleagues so busy on a Sunday night. I've
found it the best time to get something done, ever since they moved WKRP
to Monday.
Anyway, I would like to run a few propositions by you, and see if we have
agreement.
1. Our committee's role is to come up with an equipment purchase plan.
The amount we propose to ARPA and others will not include maintainence.
Those charges will be collected from one or more contracts as needed.
The paradigm is the way CSD-CF works, or perhaps the way maintainence
for Shasta and Diablo are included in the Baskett and Feigenbaum ARPA
contracts, resp. We must be careful, of course, not to buy anything we
cannot afford service for.
2.
2. We ought to get a little more specific on what the .5M for a file
server will buy. I think Mccarthy, Pratt, and Reid ought to confer
first and report to the committee on the 4th.
3. While we are making up a plan for the whole department, our immediate
concern is what to propose to ARPA. It seems appropriate that the
component of equipment that comes from ARPA should be as close as possible
to the contracts we now have with ARPA, especially if the proposal is
formally a supplement to existing contracts. Thus we should probably
be asking for HPP's equipment, the file server, and SUN-related stuff.
4. The issue of the equipment on which to run LISP is primarily a matter
for the users of that language, provided their requests are not so
extreme that it crowds the rest of the department's needs out.
I don't pretend to understand the relative merits of Dolphins, Symbolics,
or others. I think that as a first crack, Engelmore and Rindfleisch
should get together with McCarthy, Feigenbaum, and Buchanan, to try
to make a decision in this area, also with an eye toward reporting to
the committee on the fourth.
∂24-Aug-81 1402 ullman@Shasta via Ethernet
Date: 24 Aug 1981 13:51:21-PDT
From: ullman at Shasta
To: equip
Subject: Dick Gabriel
I suggest that we ask Dick to join the committee.
I would like to know from those who did study the matter at the beginning of
the year whether it was determined that there were different communities
of LISP users that had incompatable hardware needs.
Despite what RPG claims, it seems that LISP users in the department are
in a substantial minority (remembering that "department" includes CSL).
We right now are talking about 15% of our equipment budget DEDICATED to
this one language. If the 15 Dolphins or equivalent are not going to
serve the entire LISP community, are we going to have to raise that percentage
substantially?
∂24-Aug-81 1734 pratt@Shasta via Ethernet
Date: 24 Aug 1981 17:34:09-PDT
From: pratt at Shasta
To: equip
Subject: solicited input
Here's a solicited input from Jeff Mogul. I wonder to what extent Dick
Gabriel's concerns about the 780 not being adequate for Lisp would be
met by a larger Vax.
From mogul Mon Aug 24 11:41:31 1981
To: avb pratt
Subject: committee report
I have one comment on the Vaxen in your proposal:
The proposed purchase of a 2080 reflects (I presume) a desire to
buy DEC's most powerful model in the 20 product line, in spite
of the fact that they are somewhat of an unknown. I suggest that
rather than buying 4 11/780s within the next three years, we buy
one or two 780s and attempt to buy the forthcoming larger Vax
when it becomes available. My experience with DEC's pricing
policies is that it will give more "bang for the buck" than
a 780 when it is introduced. I suspect especially that it
will allow a larger complement of physical memory (currently
the 780 is limited to 8Mb [or 12Mb with 4Mb of dual-port memory],
and the 750 is limited to 4 (2?) Mb.) For properly-written
VLSI and AI software, this could be an advantage.
-Jeff
∂24-Aug-81 1800 REG@Sail via Ethernet
Date: 24 Aug 1981 1756-PDT
From: Ralph Gorin <REG at SU-AI>
To: Equip at SHASTA, (sunet) at Sail
I agree that the new larger Vax (named Venus) will be more cost-effective
than the 11/780. However, Venus is further behind schedule than the 2080;
we'll be fortunate to see it in '85.
The 2080 is not a totally unknown quantity: its prototype IBOX is running.
It looks quite promising. The 2080 will IMPLEMENT 30-bit virtual
addresses (word addresses). It will probably be available with 4 Million
words of physical memory, with 16 M word versions available when 256K
chips are ready.
∂24-Aug-81 2203 pratt@Shasta via Ethernet
Date: 24 Aug 1981 22:01:56-PDT
From: pratt at Shasta
To: equip
Subject: msg 2 from rpg
Date: 25 Aug 1981 0015-EDT
From: GABRIEL at CMU-20C
Subject: Response to Ullman (part II)
To: equip@shasta at SUMEX-AIM
cc: gabriel
My estimate is that 40% of the users of computers at Stanford
who program (as opposed to edit only) are LISP users. First,
there is HPP, which accounts for 20 (?) LISP users, 25% of
whom do not exclusively use InterLisp (the only thing that
Dolphins run). Formal reasoning has about 10 people, all of whom
use a non-InterLisp. Hand eye has 6 or so non-InterLisp users.
There are 6 random Ph.D students (approx) who use one Lisp
or another. Kaplan's crew uses InterLisp and accounts for
3 people (?). Verification (CSL) uses MacLisp and has 3 or 4
active users. The S-1 (non-ARPA) uses MacLisp for NIL development.
A good rule of thumb is to estimate the proportion of ARPA contracts
doing AI to those that don't, and assume that 95% of the AI people
use Lisp exclusively and 5% of the non-AI people use it extensively.
Having maintained MacLisp at SAIL for 6 years, there is a constant 20
user community at any given time on SAIL. 206 is nearly always taught
using MacLisp rather than InterLisp.
I can state unequivocally that 15 Dolphins will be a boon to HPP and
virtually worthless to the rest of the LISP community, and the
non-InterLisp community is at least as numerous as the InterLisp.
Vaughn pointed out that the Symbolics Lisp machine is a substantial
performance leader over the Dolphins, and there is the distressing
fact that the Dolphins strictly run Lisp, while the Lisp Machines
run a total system with editors, network servers, mail servers,
document compilers and the like.
However, let me be clear on this: HPP needs those Dolphins, even
if just as a stopgap measure. We all face address space limitations
and the Dolphins give a few precious bits of address. The VAXes
will not be able to run development InterLisp jobs (Masinter Report).
HPP uses InterLisp. THerefore they need them now. The rest of us
(50%) do not need these machines and they are worhtless for our purposes.
The reasons that we cannot use them include such things as the fact that
they only run InterLisp, but range to such taste oriented things as an alien
programming style and environment to such technical points as horrendously
slow boxed number arthmetic, slow function call, and the arcane, useless
spaghetti stacks.
My guess is that 30% of the total user community uses Lisp and that 15% of the
budget is too low. And, Lisp Machines are general purpose insofar as
editting etc go.
However, my own personal bias is away from personal machines, at least
as envisioned and/or configured these days. I would prefer to see
a fair amount of money go into a very large machine that will be able
to run Common Lisp (and InterLisp) well and with absurd speed. Moreover,
I do not wish to see all of our timesharing machines burdened with
outdated and unfriendly user interfaces such as what TOPS-20 and
Unix offer at the moment*. Bizarre as it may seem there are one or two
minor features of WAITS that make it quite usable, and cause those
of us forced to use other systems (like the ones at CMU) to wail
in utter despair. Of course, the SUN terminals will (or should) have
screen editors locally so that one never has to retype something
nearly correct.
So, I hope that there is room in the considerations for a general
purpose machine (stock hardware) that is amenable to Lisp and
is usable in and of itself. One possibility (remote, slightly,
and expensive) is the S-1, which, if it ever makes it outside
of the dreams of men, will outclass anything else around by far.
-----------
*Unix is elegantly programmable and can achieve wondrous things,
but so is/can TECO (since you can derive EMACS from it). But, would
you be happy to have a TECO without EMACS, just with the knowledge that
EMACS could exist?
-------
---------------
-------
∂25-Aug-81 1410 pratt@Shasta via Ethernet
Date: 25 Aug 1981 14:05:51-PDT
From: pratt at Shasta
To: equip
Subject: 2nd mogul input
>From mogul Tue Aug 25 11:47:39 1981
To: avb pratt
Subject: committee report
I don't have direct experience with Vaxen running large lisp jobs,
but I think RPG's concern needs some examination.
"Is the Vax inadequate to run more than about one Lisp job at a time?"
Well, even the Cray-1 cannot provide complete service to two
simulataneous CPU-bound jobs, if that is the criteria being used.
Further, an 11/780 is not going to provide as much compute power
as a KL-10A.
The more important issue is, "does the operating system on the
Vax (i.e., Unix) degrade badly under the load of several large
Lisp jobs?" I think the answer here is a qualified "yes", the
qualification being that I don't think anyone has released a
lisp that really interacts properly with the Vax architecture
and secondly, that UCB is promising to improve the memory management
facilities so that, e.g., they will not try to maintain a working
set when Lisp is doing a non-compacting GC.
-Jeff
∂01-Sep-81 1643 JMC
To: EQUIP at SHASTA (sunet)
equip.mem[e81,jmc] SOME CONSIDERATIONS ON COMPUTER EQUIPMENT
I hope for general agreement on the first group of considerations.
A second group contains opinions which run somewhat counter to
what seems to be the spirit of the times.
1. We must be realistic about the amount of money we can
expect to get. The wish list of the previous memo totals more than
$6 million, and Bob Engelmore says we can realistically hope to
get $2 million from DARPA with a possibility of more later. We
should make our purchase plans taking into account the possibility
that we don't get more later.
My opinion is that we must run scared about money. My experience
at M.I.T. when we asked for more money than anyone planned to give
us is that we didn't get a definite negative answer; matters just
dragged on indefinitely. Therefore, we must maintain contact with
DARPA and other sources after a preliminary proposal.
2. The proposed file server from Foonly is $205K to $220K,
not $500K as mentioned in the memo. This includes an F5 processor
at $55K to $65K, a 6250 bpi tape unit at $30K to $35K, 6 disks of
670 megabytes each at $20K totalling 4020 megabytes and $120K. An
ethernet interface is included in the processor as standard.
Some questions remain about which I have been refraining from
contacting Poole until the dust settles. These are
(1) Within a year or so, the most cost-effective disks are likely
to be the IBM 3380 or some cheap imitation. Will the F5 handle
the data rate, and what will it cost for a controller or can
that be included in the processor?
(2) Will the system have adequate error detection and correction?
(3) This one is for us - especially Brian Reid. Is the F5 fast
enough or would we need an F4?
I hope this is non-controversial. Our goal with file
system should be to be able to buy whatever is the most cost-effective
large scale storage. Already the Foonly proposal will permit
us to give up purges of material prepared by hand. We will be
able to keep reports and theses on-line indefinitely. Eventually
we want to be able to keep a library on-line, starting with all
worthwhile computer-represented reports and theses we can lay
our hands on.
While I haven't mentioned database work, I don't wish to
exclude it from the above.
3. I see replacing SAIL first and then SCORE. The
candidates are the 2080 and the S-1, possibly both.
Since neither machine is now available, we need to be sure
that we reserve enough money for one or the other.
4. The 6 megabuck plan includes $1.5 million for 200
SUN terminals, but there has been a response that it is unrealistic
to expect to get them so cheaply. If we cannot afford this
much money, then we need a plan for 200 cheaper terminals. I am
sure we can have bit mapped displays with arbitrary character sets
for a cost of between $1000 and $1500 apiece provided we accept
525 line resolution and use them as terminals only.
I have a detailed proposal for this that needs refurbishment. An
opus on the use of arbitrary character sets is KEYBOA.PRE[W81,JMC].
5. Here's the controversial point. The comparison of
individual workstations and time-sharing has always compared
futuristic versions of the workstations with present time-sharing
in cost-effectiveness. It was said that an Alto would cost $500
in 5 years, but that was said more than 5 years ago. The prices
now being charged for LISP machines are realistic estimates of
what a high capacity individual station costs.
Time-sharing isn't dead yet. Moreover, if we get good terminals
and a good file server on the Ethernet, we can put Lisp crunchers
and number crunchers on the net very cheaply.
6. Second controversial point. If some donor gave the
Stanford CSD $6 million to spend as we choose, I think I would
put only about $2.5 million into equipment. The rest would go
into faculty and space.
∂02-Sep-81 1540 JMC
To: mdp at MIT-AI
Please drop me from human-nets. It's interesting but I'm short of
time. Should a more tightly edited and shorter digest become
available at some time, I'd be interested.
∂02-Sep-81 1928 JMC terminals
To: ME
I forgot to tell you that I have the Doyle terminal (9528 on the keyboard),
and the one I had is in the shop again. I think I'll keep this one and
let someone else have the other after it is finally fixed.
∂02-Sep-81 1930 JMC file server
To: csl.bkr at SU-SCORE
First the good news:
Martin Frost thinks SAIL can be readily made to work page by page from
the file server. All he has to do is let the programs think the file
server is a user disk pack.
Now the questionable:
I have been inquiring about Ethernet actual transmission speeds. Mark
Roberts, at my and Ralph's request, transferred a large file between
the two Vaxen. The best speed he could get was 127 kbaud. Doubtless
the problem is in the Ethernet software in the Vaxen, but an Ethernet
based file server may also have problems. Have you looked into the
question of what transfer rates can actually be expected taking
everything into account?
∂02-Sep-81 1937 JMC
To: FFL
Can you get me a new Stanford ID card?
∂03-Sep-81 0208 JMC McGrath
To: pourne at MIT-MC
His plan says he'll be at Denver Hilton till September 8. Perhaps he'll
be back after that. The command FINGER, perhaps called something else
at MC finds when someone has logged out most recently and his plan if he
has filed one.
∂03-Sep-81 1552 JMC
To: RWW
2. The most important cognitively oriented phenomena may be facts
about information. For example, a computer may say, "That information
is in our catalog which will be available tomorrow?".
Referring to information in a file or document is a weak form of an
assertion of knowing. Belief may have to be treated but knowledge
is the first approximation.
Notice that going from a question (say "What pencils do you manufacture?")
to a reference to information (say "That information will be in our
Fall catalog.") requires a reification. The question is conveniently
expressed as a logic sentence with pencils and kinds in the ontology,
whereas the reply requires that information be an object.
The ability to answer a question in this way should be a major
characteristic of CBCL and the programs that use it. One might hope
to refer to information in the original question, so that the program
doesn't have to do reification, but my intuition suggests that this
won't work. Further reifications will always be wanted.
Richard Weyhrauch's METAFOL ideas may be applicable here.
∂04-Sep-81 0128 JMC
To: minsky at MIT-AI
Did I leave a small black notebook in your study when I stayed there?
∂07-Sep-81 1157 JMC
To: bobrow at PARC-MAXC
How about Colmerauer as editor of issue on non-monotonic reasoning?
∂08-Sep-81 1415 JMC Re:
To: FFL
∂08-Sep-81 1259 Bobrow at PARC-MAXC Re:
Date: 8 Sep 1981 12:59 PDT
From: Bobrow at PARC-MAXC
Subject: Re:
In-reply-to: Your message of 07 Sep 1981 1157-PDT
To: John McCarthy <JMC at SU-AI>
cc: bobrow
What is Colmerauer's address? I have to write and ask him to join the board, so I
can ask him if he wants to do it (and that you recommended him). You can put
pressure on him too if you think that would help.
Do you think he would be better than Ray Reiter?
danny
∂08-Sep-81 1428 JMC
To: bobrow at PARC-MAXC
Colmerauer is very talented but somewhat self-isolating and given to feeling
unappreciated. He is the originator of Prolog (1971), the first and main logic
programming language, but he didn't write a paper on it at the time, and
so his role is often ignored, because he isn't referred to in bibliographies
as the originator. It is as though I had neglected to write a paper on
LISP, and its effects are similar to the those of the fact that I didn't
write a paper proposing time-sharing in 1959, and my evidence for being
the inventor of it is my January 1959 memo to the director of the M.I.T.
Computation Center. Anyway, there may be a little lost motion in Colmerauer
getting started, but I think he'll do as good a job as Reiter in the end.
There is no great rush, and it will bring Colmerauer out into more contact
with the AI public, which will be good for him and for the AI public.
The following address will get him, but I think it isn't up-to-date in
terms of French zip codes.
Prof. Alain Colmerauer
Groupe d'Intelligence Artificielle
Unite d'Enseignement et de Recherche
70, Route Leon-Lachamp
13 - Marseille (9 e)
FRANCE
∂10-Sep-81 0000 JMC
bboard
∂09-Sep-81 2322 Richard M. Stallman <RMS at MIT-AI>
Date: 10 September 1981 02:21-EDT
From: Richard M. Stallman <RMS at MIT-AI>
To: jmc at SU-AI
Would you please post this notice at SAIL and SCORE for me?
Before Friday afternoon (Washington time), call Western
Union and ask to send a public opinion telegram to Edward P.
Boland, Congressman from Massachusetts, telling him "Support
$4million funding for the Solar Electric Propulsion System",
or some paraphrase.
SEPS is required for the mission to intercept Halley's
comet, and it was cancelled this spring. If we don't do it
now, the next chance is in 2061.
The public opinion telegram costs about two dollars.
You can send one even in the middle of the night, and
charge it to your home phone.
Don't mention that you are an engineer, or that you work
with computers, or that you went to college, or any
organization you are affiliated with. Politicians are
looking for the opinions of the "average man", and discount
anyone who they know is educated.
Please pass this on.
∂10-Sep-81 0029 JMC
To: FFL
HALL.1[LET,JMC]
∂10-Sep-81 0910 JMC
To: bundy at MIT-AI, FFL at SU-AI
∂10-Sep-81 0131 Alan Bundy <BUNDY at MIT-AI>
Date: 10 September 1981 04:29-EDT
From: Alan Bundy <BUNDY at MIT-AI>
To: jmc at SU-AI
John
You were going to send me some stuff on DIALNET
and Mr S&P. What you actually sent me was stuff on the history
of timesharing. Vey interesting, but ....
Alan
Then I must have sent Stoyan the stuff on Dialnet. It will
be fixed.
∂10-Sep-81 1714 JMC your latest
To: reid at SHASTA (sunet)
I entirely agree with your message, but I fear that your memo (and Vaughan
Pratt's also) may help bring about the result you and I fear. It won't
probably take much to cause DARPA to take the view that it shouldn't
aid the Department at all, but only the projects it supports. In particular,
if HPP doesn't support Departmental resources, this seems very likely. I
also think that Feigenbaum has some separatist feelings that may be
accentuated by his departure from the Departmental Chairmanship.
I think tact is called for. Unfortunately, I don't have a clear idea of
what constitutes tact in this case.
∂11-Sep-81 1208 JMC salary deposit
To: FFL
Please find out what I have to do to get my salary deposited into my
new bank account, same bank. The new account number is 64828 0503.
∂11-Sep-81 1247 JMC
To: FFL
I have left a check for my daughter Sarah on my desk. She'll come for it.
∂11-Sep-81 2355 JMC
To: LLW
Are you expecting me Monday?
∂12-Sep-81 1158 JMC
To: PMF
Yes. I would like to see it.
∂13-Sep-81 1044 JMC
To: LLW
CC: TM, CEG, RPG
Let's make my start Tuesday then.
∂13-Sep-81 1719 JMC
To: FFL
dahl.1
Dear Dr. Dahl:
I haven't made any progress in arousing a desire to have
Prolog taught here this year, so you had better accept the German
offer. Perhaps something can be worked out some time in the future.
To tell the truth, I have little administrative energy at present.
.sgn
∂14-Sep-81 1430 JMC
To: FFL
nguyen.1[let,jmc].
∂14-Sep-81 1826 JMC
To: RPG
Let's go to Livermore together tomorrow.
∂14-Sep-81 2332 JMC
To: RPG
OK, ten it is. Do they generally start late there?
∂15-Sep-81 0146 JMC
To: rms at MIT-AI
I looked at it some, and I didn't think it was on the right track.
However, I'll look at it again, send you a message about it. Right
now I'm thinking about different questions, and I'm not in a position
to interact. Maybe I will be able to send you a message about coming
out or not in by the end of the month.
∂15-Sep-81 0218 JMC
To: rms at MIT-AI
I just FTPed another version of your paper. Tomorrow I'll
try to SCRIBE it. However, it looks a lot better and
more comprehensible than the previous version. However,
so far it shows no sign of comprehending the issues
involved in circumscription.
Here is one preliminary comment.
meta[e81,jmc] Comments on Stallman's META paper
1. Non-monotonicity doesn't prevent a system from being
recursively enumerable. If the opportunities for default
reasoning at any point in the argument and the requirements
for taking back previous results are both effectively
computable, then the set ofstates of belief are
recursively enumerable just as is the set of theorems
of an ordinary theory. The difference is that the set of
all sentences that might be derived is not so interesting
as in conventional logic. Since some are incompatible with
others, the important thing is the set of states of belief.
I think mathematician will be interested in non-monotonic
systems once they get used to the idea.
2. I still think one should say "non-monotonic reasoning" instead
of "non-monotonic logic".
∂15-Sep-81 2352 JMC
To: FFL
Please call Elliott Bloom 282469 at SLAC and ask him or his secretary to
put Mike Genesereth on SE2 list.
∂16-Sep-81 1035 JMC
To: FFL
I think there are some people. Les Dugan should be contacted.
∂16-Sep-81 1103 JMC
To: FFL
∂14-Sep-81 2111 JJW EKL Reference Manual
To: JK, JMC, JMM
The EKL manual is now in a relatively complete state. I've included
examples of most of the commands, and reorganized some of the sections.
There isn't much time left before it gets reproduced for 206 class
notes, but if you are able to proofread it, please send me any corrections
or suggestions. You can Dover my file EKLMAN.PRE[EKL,JJW] to get a copy.
∂16-Sep-81 1110 JMC multi-processing in common LISP
To: RPG
After talking to with Erik and pondering for a while, I began to conjecture
that the best way to get a multi-processing experiment of the kind I want
might be in LISP after all even though LISP programs aren't likely to be
the ones that will need it most. I haven't looked through the Common LISP
manual to see what there might be yet but will do so when I get to Livermore
towards which I'm leaving now. Can you phone me there after 1pm or MAIL me
a message with a phone number.
∂16-Sep-81 1351 JMC badge
To: untulis at SRI-AI
I just remembered that one of the objects in my wallet that was stolen
in Vancouver was my SRI badge. This was on the Wednesday of IJCAI week -
therefore August 26. Can you inform whoever should be told? I'm at
Livermore temporarily but can be MAILed to as usual.
∂16-Sep-81 1528 JMC Foonly
To: csl.bkr at SU-SCORE
I think that the facilities committee has shown sufficient seriousness
about the file server to justify further discussions with Foonly. I
suggest that you and I visit Foonly armed with questions. I can't
do it in the next two weeks unless it turns out to be unexpectedly
urgent, but I guess I could manage a phone call. I'm mainly at
Livermore for a couple weeks, but I can be reached by messages to
SAIL or by phone at 422-0758 day or evening or at home 857-0672
evening or at Stanford 7-4430 evening.
∂16-Sep-81 1541 JMC EKL manual
To: JK
I'll have substantive comments on the new version, but I would suggest
that you include among the acknowledgements a nice sentence about FOL.
I believe it would be appreciated and might have other advantages.
∂16-Sep-81 1628 JMC
To: FFL
Can you pick up META from Dover? It's Stallman's latest.
∂17-Sep-81 0114 JMC compatibility
To: rms at MIT-AI
If you aren't careful M.I.T. will lose its reputation for incompatibility.
All I had to do with META was to put @make(Report) at the beginning and
SCRIBE compiled it without complaint, and our Dover printed it. I even
expect to read it fully soon.
∂17-Sep-81 1349 JMC
To: NCR at SU-AI
SPACE[E81,JMC] contains bboard remarks probably deleted before your return.
∂17-Sep-81 1406 JMC
To: RPG
Is there a document describing differences between S-1 LISP and common LISP?
∂18-Sep-81 1610 JMC
To: FFL
I suppose I'll give you a check Monday or even Tuesday. I'll leave you a
message over the weekend about the ticket and about those appointments you
have tentatively made for me.
∂18-Sep-81 1612 JMC
To: FFL
For you and urgent messages, my phone at LLL is 422-0758.
∂18-Sep-81 1613 JMC
To: CLT
For you and urgent messages, my phone at LLL is 422-0758.
∂18-Sep-81 1617 JMC terminals
To: JRP
I don't want to sell our terminal. However, at Livermore this week, I have
been using the Ambassador terminal made by Ann Arbor Terminals, Inc.,
6175 Jackson Road, Ann Arbor, Michigan 48103, (313) 663-8000. It costs
$1300 in quantities of one without modem. It can display up to 60 lines
of very legible characters. It doesn't have the Stanford special characters
like logical symbols, although they can be used with tricks. In my opinion,
you could (as I have) easily get used to it, and SAIL supports it for the
benefit of the Livermore S-1 project. A suitable modem would cost a few
hundred dollars - Ralph would know more accurately.
∂19-Sep-81 1032 JMC
To: JRP
Please return the terminal to Martin Frost in Room 030 Margaret Jacks
Hall. If he isn't there leave a note stating that it is one I lent you.
If you plan to continue using SAIL, discuss what terminal you buy with
Martin or Ralph unless it is one that SAIL already supports. The point is
that each terminal has its peculiarities, and its use is more or less
convenient depending on whether SAIL know about that terminal, and Martin
will probably decline to do a lot of work for a terminal of which only one
is bought. Of course, the same considerations apply to any other system
you may contemplate using. I'm glad you got good use from the terminal.
Tell me more about the new Product Associates terminal.
∂19-Sep-81 2027 JMC
To: RPG
CC: JRP
What a coincidence that you should ask. John Perry of the philosophy
department (JRP) is returning one momentarily. I suggest you contact
him and arrange to pick it up.
∂19-Sep-81 2028 JMC
To: CLT
Sorry I forgot to tell you I would be late tonight.
∂19-Sep-81 2145 JMC
To: CLT
OK for now. I'm also getting harassed, but I think we can afford to
increase our allocations according to our needs.
∂19-Sep-81 2257 JMC
To: pourne at MIT-MC
I will come with Lowell Wood Thursday afternoon.
∂20-Sep-81 1440 JMC Emacs
To: rms at MIT-AI
"If you type a question mark when an error message is on the screen, you
enter the EMACS error handler. You probably don't want to do this unless
you know how to write TECO programs. Enough said."
Not enough. It is common to type question mark when in trouble. When
I found myself in the error handler, the problem was how to get out again.
Especially since I had a large unsaved edit, had inadvertently entered
minibuffer mode by hitting a wrong key and only wanted to get out of that.
There was a happy ending, since delete got me out of the error handler,
and ↑] got me out of the minibuffer, but I nearly lost a large edit.
The fact, if it is a fact, that <delete> will get out of the error handler
should be in the manual at that place if not elsewhere also. Many people
have complained that it is hard to get out of inadvertently caused
trouble. You should consider the problem of a beginner whose cat has
walked on the keyboard and put EMACS into an unknown state.
∂20-Sep-81 1532 JMC
To: CLT
OK, I'll be there at 5pm with shopping list.
∂20-Sep-81 1644 JMC
To: TOB at SU-AI
CC: REG at SU-AI, csd.feigenbaum at SU-SCORE
My immediate reaction is that six Vaxen is enormous overkill, and
you probably won't get them, and you'll be sorry if you do. We need
to discuss it.
∂21-Sep-81 1153 JMC
To: TOB
I guess I got over excited, and I'll get back to you with more considered
reasoning about why I think we should be cautious about acquiring too
much equipment and how it might apply to the present case. The argument
about VAX becoming standard for vision and assembly work has great force,
however.
∂21-Sep-81 1154 JMC
To: TOB
Something will have to be done. I'll be in tomorrow.
∂21-Sep-81 1155 JMC
To: REG
Will you be in tomorrow to discuss finances of SAIL?
∂21-Sep-81 1253 JMC
To: RPG
Please inform ME that you have it now.
Are you planning to go to LLL on Thursday? I need to get there,
to go to L.A. with Lowell, but I will be returning to SFO.
∂21-Sep-81 1308 JMC various
To: FFL
Please print REW.DOC[EKL,JK] and leave it for me tonight. It's short.
I'll be in tomorrow and will keep those appointments you made for me
and will arrange to pay for my tickets. Since I will drive to L.A.
with Lowell Wood, the trip should start there.
∂21-Sep-81 1313 JMC
To: JK
I'll try to figure it out, but perhaps we'll have to try it and see
what proofs it shortens.
∂21-Sep-81 1439 JMC multi-processing
To: RPG
MULTI[E81,JMC] contains a beginning of proposals for LISP multi-processing.
Please tell me if these ideas seem to be re-inventing some wheel. If
there is some totally different concept, I would like to know. Also
if the sketch is incomprehensible.
∂21-Sep-81 1859 JMC
To: REG
If you haven't already, look at p. 66 of bboard.
∂21-Sep-81 2055 JMC
To: admin.mrc at SU-SCORE
I have forgotten what PROLOG tape it is.
∂21-Sep-81 2244 JMC
To: LGC
I will be in tomorrow (Tuesday) and then not for a week.
∂22-Sep-81 1147 JMC
To: csd.golub at SU-SCORE
I'll be down in a few minutes.
∂22-Sep-81 1510 JMC
To: FFL
∂14-Sep-81 2111 JJW EKL Reference Manual
To: JK, JMC, JMM
The EKL manual is now in a relatively complete state. I've included
examples of most of the commands, and reorganized some of the sections.
There isn't much time left before it gets reproduced for 206 class
notes, but if you are able to proofread it, please send me any corrections
or suggestions. You can Dover my file EKLMAN.PRE[EKL,JJW] to get a copy.
∂23-Sep-81 0025 JMC
To: RPG
What are prospects of your going to Livermore Thursday?
∂23-Sep-81 0026 JMC
To: EJG
What are prospects of your going to Livermore Thursday?
∂23-Sep-81 0959 JMC
To: RPG
Thanks for your ride offer. It's a bit of a last resort. Can I phone
you this evening in case I want to take you up on it and at what number?
∂23-Sep-81 1429 JMC ride
To: EJG
9:30 would suit me fine. If you can come by my house, 846 Lathrop, Stanford,
that would be best, but the Department is also possible. It is correct that
I won't want a ride back.
∂23-Sep-81 1448 JMC student support
To: DPB at SU-AI
I'm content at present with just supporting Yoram Moses. If a student
should turn up strongly motivated to work in one of my areas, I would
consider supporting him.
∂23-Sep-81 1457 JMC extended numbers
To: RPG
The S-1 LISP implementation guide mentions RJF's design as described
in SYMSAC '76 proceedings. Who is RJF, and do you happen to have
this article?
∂23-Sep-81 1552 JMC
To: FFL
It doesn't matter to me from what accounts we borrow. If possible, when the
money comes through I would prefer to change the charges retro-actively
rather than get involved in deals.
∂23-Sep-81 1554 JMC
To: FFL
Goad is 100 percent for the fall. Check with Gabriel, but I think the
way we left it is that he would stay on Livermore payroll for the Fall.
∂23-Sep-81 1634 JMC
To: RPG
Erik is leaving for LLL at 9:30 which suits me better. Thanks though.
∂23-Sep-81 1711 JMC
To: RPG
Consider multiv[e81,jmc] at your leisure.
∂23-Sep-81 2157 JMC
To: RPG
I now think that the examples in MULTIV don't suffer much from the
restrictions in the COMMON LISP document.
∂23-Sep-81 2158 JMC
To: RPG
I don't want to support the LISP timing project without promise of
additional ARPA funds.
∂23-Sep-81 2234 JMC
To: RPG
I have heard nothing. I suggest you ask Englemore whom to contact in
ARPA.
∂23-Sep-81 2241 JMC
To: RPG
No. I won't be there Friday. I'd like to see the E/Lisp demo, but
not tonight. What would be your reaction to the idea that the
Formal Reasoning Group pass (or almost pass) the first year in
exchange for a commitment to make up its share the following year.
Then we could see what worked out well.
∂23-Sep-81 2322 JMC
To: RPG
We would have to get adequate guarantee that we would get ours the following
year, and this might be hard to arrange. At present it's just an idea.
∂24-Sep-81 0101 JMC cs206
To: JMM
I will be away until Wednesday evening, so I won't have a chance to
talk to you about getting ready for cs206. Fortunately, the first
week won't require the computer. I have been dithering but have
finally decided to use LOTS for the programming part of the course
and, if it turns out to be possible, for the EKL work as well. The
problem is the lack of available terminals on SAIL.
Discuss with Jussi getting EKL to run at LOTS. Since it is written
in MACLISP and we use MACLISP on LOTS, there shouldn't be any
fundamental problem, except that the interaction with the line
editor won't work.
Also we should reprint "Using LISP at LOTS". It is among the files
of 206,LSP, I suppose. Carolyn can give you some advice, since she
has taught the course recently.
The course will be on TV this time, and Fran has received various
communications from the TV people, and you should ask here about
it.
I rely on the TA to take on himself quite a lot of responsiblity.
∂24-Sep-81 0110 JMC Who's who
To: FFL
I have left a form from them in my out box. Remind me to correct the
errors when I get back.
∂24-Sep-81 0125 JMC
To: JMM
I have left some LOTS stuff for you on Fran's desk.
∂24-Sep-81 0139 JMC telegram to IBM Japan
To: FFL
Unfortunately I am unable to accept your kind invitation.
∂24-Sep-81 0902 JMC faculty club bill
To: FFL
The charge was 201.07 for Colmerauer and 150.00 for Stoyan.
∂24-Sep-81 0907 JMC
To: cequip at DIABLO (sunet)
An S-1 may be $650K. See you all next week.
∂24-Sep-81 0907 JMC
To: equip at DIABLO (sunet)
An S-1 may be $650K. See you all next week.