perm filename OUTGO.MSG[1,JMC]9 blob
sn#692366 filedate 1982-12-23 generic text, type C, neo UTF8
COMMENT ⊗ VALID 00002 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00002 00002
C00003 ENDMK
C⊗;
∂01-Oct-82 1657 JMC industry professorship
To: csd.golub at SU-SCORE
1. Maybe there is a better name.
2. It occurs to me that if we decide on the people for 1983-84 by
February, we can get the specific course descriptions in the catalog.
I would propose a single permanent number, and a course description that
would give the three course descriptions for the following year with
a disclaimer recommending that students check with the Department
the preceding quarter to be sure that there has been no change in
lecturer or offering.
∂01-Oct-82 1702 JMC industry lectureship
To: bmoore at SRI-AI
Remember your discussion with me last year about desire to teach
courses. I have got the Department to agree to a one year experiment
with an "Industry Lectureship". It is described in the file
INDUS[E82,JMC]. I hope this meets the need. The committee consists
of me and Gene Golub. The first lectureship will be in the Spring
Quarter of this year.
∂01-Oct-82 1809 JMC unnecessary messages
To: csd.armer at SU-SCORE, pratt at SU-SHASTA
Through some bug, mechanical or human, your replies stating your respective
desires to have lunch with Barbara Liskov are going to the entire department.
Such affairs are best arranged more discreetly. Moreover, they distract
everyone's attention, and clutter the disks.
∂01-Oct-82 1819 JMC
To: FFL
steele.re1
∂01-Oct-82 1922 JMC
To: RPG
The common lisp file seems to be breaking checksum, and the problem is
unlikely to be fixed. I suggest splitting the file.
∂01-Oct-82 1913 BUG-CKSUM problems with enormous files
Sender: Woods at PARC-MAXC
Date: 1-Oct-82 19:12:07 PDT (Friday)
From: BUG-CKSUM
Subject: problems with enormous files
To: JMC at SAIL
cc: ME at SAIL
Due to the way CKSUM works, it is unable to handle files larger than about 102
disk blocks. The SAIL compiler compounds the problem by failing to provide
graceful error recovery if an attempt is made to allocate more memory than
exists.
Fixing this problem would require substantive changes to CKSUM and so is
unlikely to occur in the near future. A quick fix could be provided to make
CKSUM ignore all but the first 100 disk blocks of any single file.
-- Don.
∂01-Oct-82 1924 JMC
To: woods at PARC-MAXC
Thanks for the prompt reply. The solution is to split the file.
∂01-Oct-82 2310 JMC
To: FFL
Do I have a parking permit application?
∂01-Oct-82 2331 JMC
To: csd.broder at SU-SCORE
Please put me on AFLB list.
∂02-Oct-82 2133 JMC
To: FFL
LEWIS.1 together with reprint of circumscription paper.
∂04-Oct-82 0226 JMC
To: ARK
I'll look for them, but you might try FIND REITER IN LIBRAR[1,JMC].
∂04-Oct-82 1521 JMC
To: gjs at MIT-AI
Could you send me a copy of the notes for your Scheme course. KMP was here.
∂04-Oct-82 1650 JMC
To: REG
Do you think Brian Tolliver could do McSun?
∂05-Oct-82 0006 JMC Lisp Machine Manuals
To: MLB
CC: JJW
There seems to be a market for between 30 and 40 Manuals in my class,
for which I believe you quoted a price of $12.00 each. Since it is
better than the Maclisp manual for users of Maclisp, I'm sure you'll
sell more if you get them.
∂05-Oct-82 0008 JMC
To: MLB
You might consider updating your plan.
∂05-Oct-82 1241 JMC
To: LGC, RPG
3pm is ok.
∂05-Oct-82 2236 JMC your paper and programming project
To: YOM
1. I have read your paper, and it solves the problem of the Berliner position.
On the other hand, the notion of "while W is following the path" was
ignored. Perhaps it's unnecessary for the problem, and perhaps it's
unnecessary in general. Certainly, it's not clear at present. Anyway, I'll
pay off, but I'm not sure that I can regard having supported a student
away from Stanford to have been a win. Perhaps you could think about "while"
a bit more and we could talk.
2. I have two possible programming projects - both involve checking out
possible improvements to Lisp. One involves using a certain kind of tree
instead of an a-list and would involve writing a version of eval. The
other involves a macro analogous to DO as implemented in Maclisp and
the Lisp machine but aimed at backtracking.
∂06-Oct-82 0102 JMC trip to Washington
To: FFL
What flight to Washington, D.C. can I get leaving Thursday after my
class which ends at 2:30?
∂07-Oct-82 1136 JMC
To: cl.boyer at UTEXAS-20
Sounds fine to me. Congratulations on your prompt solution to the
problem, and on your good luck that AAAI was also prompt.
∂07-Oct-82 1440 JMC
To: csd.golub at SU-SCORE
No honors.
∂09-Oct-82 0040 JMC
To: CLT
The Math. Society meets Jan. 5 - 9 in Denver. I speak. Maybe you want come.
∂09-Oct-82 1740 JMC possible honor
To: csd.golub at SU-SCORE
Come to think of it, maybe getting elected president of the American
Association for Artificial Intelligence counts as an honor.
∂11-Oct-82 1209 JMC Common Lisp Effort
To: RPG
∂07-Oct-82 1304 Raj.Reddy at CMU-10A Common Lisp Effort
Date: 7 October 1982 1602-EDT (Thursday)
From: Raj.Reddy at CMU-10A
To: Kahn at USC-ISI
Subject: Common Lisp Effort
CC: JMC at SU-AI, Allen.Newell at CMU-10A,
Scott.Fahlman <FAHLMAN at CMU-20C>, Raj.Reddy at CMU-10A
Message-Id: <07Oct82 160207 RR29@CMU-10A>
Dear Bob,
We have had some discussions here about the Common Lisp effort in the wake
of Guy Steele going to Tartan Labs. The following represent our current
views on the subject.
1. CMU is committed to Common Lisp and wants to make sure it is running on
as many different machines as possible, with Decsystem 20, VAXs, PERQs and Lisp
machines being the first 4 systems on which it should be running within the
next year.
2. Guy Steele won't be leaving until about January. By that time he expects
to leave things in a stable state.
3. It would be good to have Stanford active in this area with Dick Gabriel
spending full-time and John McCarthy spending part of his time, along with
one or two graduate students also working on it. We are strongly in favor
of cooperating with such an activity.
4. Scott Fahlman and a number of others at CMU will continue to spend time
but Scott is expected to concentrate on his NETL machine and parallel AI
architectures research as soon as Common Lisp begins to be useable.
5. At some point it would be good to get the manufacturers to take
responsibility for future maintenance of Common Lisp.
In summary, we are strongly in favor of supporting Dick Gabriel and others
at Stanford and we will work closely with them to make sure that the
effort is successful. I am also forwarding a more detailed response from
Scott.
Best,
Raj
- - - - Begin forwarded message - - - -
Mail-From: ARPANET host CMU-20C received by CMU-10A at 29-Sep-82 22:07:13-EDT
Date: Wednesday, 29 September 1982 22:07-EDT
From: Scott E. Fahlman <Fahlman at Cmu-20c>
To: Raj.Reddy at CMU-10A
Cc: Allen.Newell at CMU-10A, fahlman at CMU-20C
Subject: Netmail from Kahn
Raj,
Thanks for passing this along. It is interesting to see how John
McCarthy views all this. My reactions are as follows:
1. I would welcome the increased involvement of McCarthy and Stanford in
the Common Lisp effort. One more tuba on the bandwagon.
2. I would rather have the administrative control of Common Lisp
(maintaining the manual, calling the meetings, maintaining the yellow
pages, etc.) remain in Pittsburgh, where we can be sure of quality
control. But unless we get Gabriel or someone comparable to come here,
I don't see any way to do this in the long run. Guy will stay involved
for awhile, but that will dwindle; I don't want to be the main person in
this effort indefinitely -- a year from now I hope to be spending most
of my cycles buiding a parallel thinking machine; of the people now in
our department, I see nobody else who is both interested and competent
to handle this, and I can't think of anyone we might be able to hire for
this task. So maybe we have to let go and hope for the best. We'll
still have the ability to veto any changes to the language, since we
will control the Spice and Vax implementations.
3. Of the people currently heavily involved in Common Lisp, minus Guy
and me, Gabriel is by far the best person for the task of ongoing
coordination. He has pretty good taste overall, is respected by the
group, and is not snarled up in the MIT/Symbolics/LMI political
hurricane.
4. Of the goals McCarthy suggests, the important one is to maintain the
Common Lisp library and its documentation. There needs to be a center
for this, run by someone with enough stature to reject things that do
not meet the standards. The task of evaluating, cataloging, organizing,
and testing modules submitted for the library is a big one, and I don't
see the manpower to do it here without Guy's help. In addition, there
needs to be a group somewhere that works on library modules that are
needed but that nobody else wants to do. If McCarthy and Gabriel want
to do this, fine. ARPA should fund them for this.
5. I think that by the time Guy gets to Tartan, the Common Lisp manual
will be pretty stable. There should be much less need for "maintaining
the discussion and organizing meetings" and for supplements to the
manual. Guy can probably handle any ongoing administrivia, but if there
is a need for additional meetings or standardization efforts beyond what
Guy wants to do, Gabriel would be a good person to deal with this. This
function would probably not be needed in the next year or so, but might
become useful after that.
6. McCarthy's comment about Guy losing his impartial status is strange.
Tartan will not be doing a Lisp until long after Common Lisp is settled,
if ever, and even then there is little likelihood of any commercial
rivalry developing. Our involvement with DEC and with 3RCC has not
impaired our ability to act as referees. All of the implementations
(except Symbolics/LMI/MIT) occupy separate, non-conflicting niches.
So, while I disagree with a few of McCarthy's points and comments, I
welcome the initiative. I think that Gabriel would handle the administrative
control over Common Lisp in a responsible manner, and that the coordination
process would work better in the long run with his full-time involvement than
with decreasing shares of Guy's time and mine. And without such a center
somewhere, I fear that I would become the default coordinator, which
would really hurt my ability to concentrate on AI.
Feel free to share all or part of this with Bob Kahn, if you think that
would be useful.
-- Scott
- - - - End forwarded message - - - -
∂11-Oct-82 1401 JMC
To: RPG
Wednesday is ok.
∂11-Oct-82 1742 JMC Kahn
To: RPG
He was mainly interested in something else, but anyway he would like
a budget strongly hinting that it should be ≤ $100K per year. He is
reluctant to make it a line item, fearing attack, and hopes to divert
some money freed by the departure of Wiederhold.
∂12-Oct-82 1844 JMC
To: JJW
Yes on Hertz.
∂13-Oct-82 1126 JMC
To: FFL
tarnlu.1 is to Sten-Ake Tarnlund.
∂13-Oct-82 1128 JMC
To: ullman at SU-HNV
Unfortunately, I can't come to Peter's Bar Mitzvah.
∂13-Oct-82 1452 JMC badge
To: pjb at S1-A
My badge expired September 30. My contract has just been renewed.
What has to be done about a new badge?
∂13-Oct-82 1457 JMC
To: llw at S1-A
My LLNL contract is renewed. When is a good time to catch you?
∂13-Oct-82 1459 JMC
To: GIO at SU-AI
What are your co-ordinates for phone or other conversation?
∂13-Oct-82 1519 JMC
To: FFL
CC: JDH
∂13-Oct-82 1513 JDH incompletes
I still have incompletes from you for 293 Spring 80-81, 390 Summer 80-81,
and I still have an "N" for Autumn 81-82
We should do something about this.
John Hobby
Please make out the cards to change John Hobby's incompletes to A's.
∂13-Oct-82 1526 JMC Szeredi note
To: warren at SRI-AI
If you still have the note from Peter Szeredi about a Prolog interpreter
permitting introspection, I'd like it back. Talking with Brian Smith
has revived my interest in the subject.
∂13-Oct-82 1530 JMC proposal
To: reid at SU-SHASTA
Now I remember what I had intended to discuss with you before
we sent of the proposal to D.E.C. I thought it should contain
some bragging about your past and future work on the leaf server.
However, if it wins as is ...
∂13-Oct-82 1618 JMC
To: bscott at SU-SCORE
Fran is resigning, and it won't be easy to replace her.
∂14-Oct-82 0000 JMC Saturday if a problem can be solved
To: llw at S1-A
I have appointments Thursday and Friday but could come Saturday.
However, there would have to be some way of getting them to make
the badge that expired Sept. 30 good for that day, since I don't
suppose a new one can be made on Saturday. If that problem doesn't
appeal to you, we'll make it when you have time again.
∂14-Oct-82 1604 JMC
To: briansmith at PARC-MAXC
I'm interested in the reflective Prolog project and can come.
If Bob Moore persuades you to change the time, I prefer a
later hour, since 9am is always difficult for me, since I'm smarter later
in the day. As I mentioned to you Peter Szeredi wrote, at my
instigation, a Prolog interpreter with some "introspective"
capability, and I think this overlaps your notion of reflectivity.
After thinking about "introspective Prolog" in connection with
what I called the Kowalski doctrine, I couldn't convince myself that
introspection, in the sense of the ability to examine the state, was
much more than a curiousity, i.e. that it had applications. It
isn't apparent that it corresponds to human reflective abilities,
since humans don't examine the "raw state", but only certain aspects
of the state.
∂15-Oct-82 1413 JMC
To: llw at S1-A, pjb at S1-A
I am leaving for LLL now and will be there before 4pm.
∂15-Oct-82 1414 JMC 1. and 2.
To: FFL
1. I have gone to Livermore.
2. Could you make some notes for your successor about the computer and other
files you keep for me?
∂18-Oct-82 1439 JMC
To: briansmith at PARC-MAXC
Whether unintentional or not, it messes things up for me. I teach
from 1:15 to 2:30 on Tuesdays and Thursdays.
∂18-Oct-82 1824 JMC
To: LGC, RPG
3pm will be fine.
∂19-Oct-82 1201 JMC
To: briansmith at PARC-MAXC
3 on Thursday would be fine.
∂19-Oct-82 2336 JMC
To: csd.golub at SU-SCORE
I don't know what you mean by "positive action", but I would oppose it
if it involved open or disguised quotas or any other
form of positive discrimination.
∂19-Oct-82 2342 JMC
To: csd.dorio at SU-SCORE
jmc - Nick Littlestone was a graduate student here in 60s or 70s. I don't
remember if he ever got a PhD. Perhaps he is known to the alumni people.
∂20-Oct-82 2228 JMC datamedia problem
To: ME
This has been a problem for a long time, but I wonder what you know
about it. When I type to E, I often get bits of the WHO line
in my output. I suppose it represents some interference between
echoing my characters and writing the WHO line. Is there anything
to be done about it?
∂21-Oct-82 0035 JMC
To: JJW
I think we should give a few easy proving problems tomorrow due in
a week and a half. See what you can come up with. I am thinking of
giving a more substantial pure Lisp programming problem also.
∂21-Oct-82 0144 JMC
To: FFL
reiter.1 noing tsymbols and phi should come out as the Greek letter.
∂21-Oct-82 0146 JMC
To: FFL
Don't bother with the letter to Reiter. Netmail works.
∂21-Oct-82 0148 JMC
To: reiter at RUTGERS
Dear Ray:
Presumably the your proof that circumscription does
predicate completion involves a systematic method for finding
the phi in the case covered by predicate completion. Could
you see if the method will find the phi in some other cases -
for example, the simple disjunctive case of
is-block A ∨ is-block B?
I have some new results on circumscription including
a treatment of your Tale of Two Cities, but I get
results that make me revise the formalism almost as fast
as I can write up what I've got. Nothing on how to invent
phis though.
What are your phone numbers?
Regards,
John
∂21-Oct-82 1309 JMC Wiederhold
To: kahn at USC-ISI
CC: RPG at SU-AI
There won't be Wiederhold money. We have a budget of 90K which we
can send you.
∂21-Oct-82 1108 Gene Golub <CSD.GOLUB at SU-SCORE> Wiederhold
Date: 21 Oct 1982 1047-PDT
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Wiederhold
To: faculty at SU-SCORE
cc: gorin at SU-SCORE, bscott at SU-SCORE
The Wiederhold situation has changed again!
It appears that it will all be resolved satisfactorily with Gio
having a joint appointment. This was resolved at the Provost's
level so that there seems to be no further problems.
GENE
-------
∂25-Oct-82 1204 JMC Common Lisp Support
To: RPG
∂25-Oct-82 0922 OHLANDER at USC-ISI Common Lisp Support
Date: 25 Oct 1982 0917-PDT
Sender: OHLANDER at USC-ISI
Subject: Common Lisp Support
From: OHLANDER at USC-ISI
To: McCarthy at SU-AI
Cc: Kahn at USC-ISI, Adams at USC-ISI
Message-ID: <[USC-ISI]25-Oct-82 09:17:25.OHLANDER>
Prof. McCarthy,
Bob Kahn has come to me for my thoughts on your
proposed support of a Common Lisp effort at Stanford. I know
Dick Gabriel has been involved with the general Common Lisp work
that has gone on thus far and I have no qualms about his
competence or dedication. I have also talked with Raj Reddy and
Scott Fahlman and they believe that it would most likely be a
good thing if Dick Gabriel were to become more active in this
area, essentially taking over the leadership of the ad hoc Common
Lisp group. However, before launching a significant development
in implementing Common Lisp, refining it, and constructing
suitable tools for a rich environment, I would like to reach some
assessment as to how "common" Common Lisp is or is likely to be.
I believe that there is enormous benefit in reaching some
standardization within the MACLISP community. I also know that
there are many independent Lisp development efforts going on with
some tender egos involved.
Another concern that I have is that the Common Lisp language and
any tools that are developed for$ a programming environment have
as much portability as possible. We don't want to get the
intermingling of environment and language that has occurred in
Interlisp. This will require that certain user site dependent
functions, such as graphics, be kept isolated from the language
kernel.
I would appreciate your thoughts concerning these issues and your
opinion on whether and how we might develop a Common Lisp effort
that will serve as a major standard within the DARPA community.
Ron Ohlander
P.S. Dick Gabriel has proposed that we continue the Lisp timing
efforts. This is certainly no problem so$ far as time for
extension of effort$ is concerned because he is covered under
your contract. The little additional money that he requires is
not worth a contractual action. I would suggest that you examine
the existing contract to see if the travel can be covered.
∂26-Oct-82 0120 JMC
To: RPG, ME
R SCHEME doesn't work any more.
∂26-Oct-82 1457 JMC
To: RPG
202 694-5922
∂26-Oct-82 1459 JMC Nov 3
To: adams at USC-ISI
I'll be here, but Dick Gabriel will be in Boston. He'll phone you.
∂26-Oct-82 1934 JMC
To: FFL
marchu.1
∂27-Oct-82 1444 JMC
To: adams at USC-ISI
CC: RPG at SU-AI
Dick Gabriel can meet on Wednesday after all.
∂27-Oct-82 1450 JMC
To: adams at USC-ISI, RPG at SU-AI
How about 11am, my office 356 Margaret Jacks?
∂28-Oct-82 1452 JMC
To: cl.boyer at UTEXAS-20
I have no objections to M.I.T. Press
∂28-Oct-82 1749 JMC
To: ME
So you say.
∂29-Oct-82 0259 JMC
To: CLT
semina[f82,jmc].
∂30-Oct-82 1357 JMC Ritchie and Hanna paper
To: PJH
I think it should be published. I would not delay its publication in
order to include a reply; that could be in a subsequent issue.
The tone is ok. I might suggest omitting the adverb "slightly" at
the top of p. 4. It looks ironical in view of the rest of the sentence.
If the word "spectacular" in the abstract was not in the original draft,
I would suggest omitting it.
I am doubtful about your having involved Lenat in your consideration
of whether to publish the paper. Once a decision to publish has been
made, it is certainly appropriate to discuss a possible replay with
the person whose work has been criticized, but it puts him in an awkward
position to ask him whether he thinks the criticism should be published.
I don't think the paper is unfair or "pillories Lenat unmercifully". It
cannot be described as a personal attack.
The one criticism I would make is that they assumed that the program itself
was unavailable. I don't think this should be the case for major AI
programs, and I don't know whether it was. I suppose Ritchie and Hanna
don't have a TOPS-20 available for testing the program, but some of their
remarks would be less conjectural if supported by a look at the program.
It further seems to me that the paper and Lenat's reply (if any) might well
be made the occasion for opening a discussion, in the AI Journal or at
IJCAI or in AAAI of what should be the standards for experimental work
in AI in terms of adequacy of description and availability of the program
for repetition. Ritchie and Hanna make an important point that repeating
certain work can be important in AI as it is in other sciences.
For example, major efforts were made to repeat Weber's claimed detection
of gravitational waves, the work on neutrino oscillations, Fairbanks's
detection of fractional charges, etc.
I would favor asking Ritchie and Hanna to ask for the program, but I fear
they might regard that as putting them off further and delaying a decision
on publication. Anyway I think it important to start more critical
discussions of research, so I urge prompt publication of essentially the
present draft.
If I knew your home phone, I'd phone you.
∂31-Oct-82 1724 JMC
To: JJW
I have the rest of the exa
∂02-Nov-82 1519 JMC comments on balzer msg??
To: RPG
∂02-Nov-82 1019 KAHN at USC-ISI comments on balzer msg??
Date: 2 Nov 1982 1017-PST
Sender: KAHN at USC-ISI
Subject: comments on balzer msg??
Subject: [BALZER at USC-ISIF: FYI]
From: KAHN at USC-ISI
To: jmc at SU-AI
Message-ID: <[USC-ISI] 2-Nov-82 10:17:57.KAHN>
john,
Would you care to comment on the above msg from Bob Balzer?
bob
Begin forwarded message
Mail-From: ARPANET host USC-ISIF rcvd at 25-Oct-82 1058-PDT
Date: 25 Oct 1982 1053-PDT
From: BALZER at USC-ISIF
To: KAHN at USC-ISI
Subject: FYI
Date: 25 Oct 1982 1045-PDT
From: BALZER at USC-ISIF
Subject: LISP COMPATIBILITY
To: ADAMS at USC-ISI
cc: OHLANDER at USC-ISI, LYNCH, BALZER
DUANE,
I UNDERSTAND THROUGH DAN THAT YOU ARE QUITE INTERESTED IN ENDING THE DIVERSITY
THAT EXISTS AMONG THE VARIOUS LISP DIALECTS AND OF YOUR PLAN TO ENLIST
MCCARTHY'S SUPPORT.
WHILE I SHARE THIS OBJECTIVE WITH YOU AND BELIEVE THAT MCCARTHY'S SUPPORT WOULD
BE HELPFUL, I'M CONCERNED THAT ANY TOP-DOWN APPROACH WILL FAIL BECAUSE OF THE
TREMENDOUS INERTIA INVOLVED IN THIS AREA. THIS INERTIA ARRISES FROM:
1. BODIES OF EXISTING CODE
2. IMPLEMENTATION AVAILABILITY(INCLUDING HARDWARE PRICE/PERFORMANCE)
3. EXPERIENCE
4. RELIGIOUS CONVICTION
WHAT THE COMMUNITY NEEDS IS A FAMILY OF MACHINES WHICH RUN COMPATIBLE LISP
SYSTEMS RANGING FROM (AFFORDABLE) PERSONAL MACHINES TO MUCH MORE POWERFUL
SHARED (PROBABLY SEQUENTIALLY) LISP SERVERS. IT SEEMS CLEAR THAT LONG TERM
THE AFFORDABLE PERSONAL MACHINES MUST COME FROM THE XEROX'S, HP'S, DEC'S,
AND IBM'S OF THE WORLD WHO ARE PRODUCING VAST QUANTITIES OF SOME POWERFUL
PERSONAL MACHINE WHICH JUST HAPPENS TO RUN LISP, AND THAT THE HIGH-END
LISP SERVERS MUST COME FROM THE SYMBOLICS AND LMI'S OF THE WORLD WHO BUILD
SPECIAL PURPOSE ENGINES FOR THIS SPECIFIC HIGH PERFORMANCE TASK. THE
SMALL SPECIALIZED COMPANIES CANNOT COMPETE WITH THE ECONOMIES OF SCALE THAT
THE BIG MANUFACTURERS GET FROM PIGGY-BACKING ON GENERAL PURPOSE PERSONAL
MACHINES. NOR CAN THE BIG MANUFACTURERS RESPOND TO THE LIMITED MARKET FOR
"SUPER" LISP SERVERS.
WE ARE REASONABLY CLOSE TO SATISFYING THIS FAMILY NEED THROUGH THE XEROX
DANDILION AND THE SYMBOLICS 3600, EXCEPT FOR THE OVERWHELMING PROBLEM THAT
THEY DON'T RUN COMPATIBLE LISPS. I THINK MOST OF THE LISP RESEARCH COMMUNITY
WOULD FIND SUCH A COMBINATION OF PERSONAL AND SERVER LISP MACHINES VERY
ATTRACTIVE.
FURTHERMORE, I BELIEVE THAT THE BEST STRATEGY FOR ENDING THE DIVERSITY AMONG
THE VARIOUS LISP DIALECTS IS TO UTILIZE THE ATTRACTIVENESS OF A DANDILION/3600
COMBINATION TO JUSTIFY A "BOTTOM-UP" APPROACH OF MAKING LISP-MACHINE LISP
AND INTERLISP COMPATIBLE, OR AT LEAST COMPATIBLE ENOUGH SO THAT THESE
MACHINES ARE INTEROPERABLE (I.E. PROGRAMS BEING DEVELOPED COULD BE MOVED
BACK AND FORTH BETWEEN THESE MACHINES FOR FURTHER DEVELOPMENT AND/OR USE).
THIS OPTION IS ATTRACTIVE ENOUGH THAT WE ARE PUSHING SYMBOLICS VERY HARD
TO EXTEND THE ALREADY QUITE SUCCESSFUL COMPATIBILITY PACKAGE, AND ARE PLANING
TO AUGMENT THEIR EFFORTS WITH SOME OF OUR OWN. PERHAPS, XEROX CAN BE
PERSUADED TO FOLLOW AN ANALOGOUS APPROACH.
MUCH COMPATIBILITY NOW EXISTS AT THE PROGRAM LEVEL. THE MAJOR EXCEPTION IS
THE WINDOW PACKAGE. IN ADDITION, A REMOTE GRAPHICS PROTOCOL NEEDS TO BE BUILT
SO THAT LISP SERVERS CAN EMPLOY GRAPHICS/WINDOWS THROUGH DEDICATED PERSONAL
MACHINES(I.E. THE USERS ONLY HAVE ONE TERMINAL IN THEIR OFFICE). BEYOND THE
PROGRAM COMPATIBILITY AREA IS THE INTEGRATION OF "ENVIORNMENT" CAPABILITIES.
HERE, BECAUSE PEOPLE RATHER THAN PROGRAMS ARE THE MAIN(BUT NOT ONLY) USERS,
EQUIVALENT CAPABILITIES RATHER THAN COMPATIBILITY IS MOST CRUCIAL.
I WOULD SUGGEST THAT IPTO HELP FORM AN APPROPRIATE GROUP OF USERS AND VENDORS
TO FORMULATE A PLAN FOR ENABLING THIS INTEROPERABLITY. PERHAPS ISI CAN/SHOULD
PLAY A MAJOR ROLE IN BOTH FORMULATING SUCH A PLAN AND CARRING IT OUT.
REGARDS,
BOB
-------
-------
--------------------
End forwarded message
∂02-Nov-82 1746 JMC
To: csd.yearwood at SU-SCORE
Tomorrow afternoon after 2pm would be ok.
∂02-Nov-82 1747 JMC
To: kanerva at SUMEX-AIM
I did think it was important and still do, especially in mathematics.
∂02-Nov-82 2253 JMC
To: RPG
How's this for a reply to Kahn?
Balzer seems to have forgotten about the Common Lisp effort, which, thanks
in large measure to Guy Steele, has succeeded in enlisting the co-operative
efforts of almost everyone in the field, except that its representation
from the Interlisp community is rather weak, though not non-existent.
I would imagine that Balzer's idea that Xerox should supply the small
machines and symbolics the larger would founder, if seriously advanced,
on the rock of Dorado, the Xerox candidate for the large machine. The
Common Lisp effort isn't top down; I know I had nothing to do with
starting it, and I believe ARPA's role was also secondary. I'm joining
it, because I think it has a good chance of succeeding, and I think I
can add to the probability of it succeeding. There are more than 600
messages in the Common Lisp file at SAIL that Dick Gabriel keeps for
the group, and I think the manual is converging, and the various
implementers seem to be saying that they can live with the result.
Certainly Symbolics will produce an Interlisp version for their machine
if they get enough orders, but judging from their strong participation
in the Common Lisp effort, I would conclude that Commmon Lisp is
their main compatibility effort.
Moreover, I believe that the Stanford effort to concentrate
on building a first class public library for Common Lisp will
be the major thing that can be done to reverse the fragmentation.
∂03-Nov-82 0058 JMC
To: RPG
It's balzer.rep[f82,jmc]. Modify it if you want.
∂03-Nov-82 0828 JMC
To: kahn at USC-ISI
Balzer seems to have forgotten about the Common Lisp effort,
which, thanks in large measure to Guy Steele, has succeeded in enlisting
the co-operative efforts of almost everyone in the field, except that its
representation from the Interlisp community is rather weak, though not
non-existent. I would imagine that Balzer's idea that Xerox should supply
the small machines and Symbolics the larger would founder, if seriously
advanced, on the rock of Dorado, the Xerox candidate for the large
machine.
The Common Lisp effort isn't top down: it was started by 4
implementation groups - Spice Lisp, Vax Nil, S-1 Lisp, and Symbolics Lisp
Machine Lisp - agreeing to move their dialects closer together. Besides
these it now includes LMI, two D.E.C. sponsored LISP efforts (for VAX
and TOPS-20), and Portable Standard LISP. I doubt the wisdom of a
fresh start on forming a compatibility effort. I'll talk to Ed
Feigenbaum and explore the attitude of his community to the possibility
of their supporting the Common Lisp effort.
I'm joining it because I think it has a good chance of succeeding,
and I think I can add to the probability of it succeeding. There are more
than 600 messages in the Common Lisp file at SAIL that Dick Gabriel keeps
for the group, and I think the manual is converging. Also the various
implementers seem to be saying that they can live with the result.
No doubt Symbolics will produce an Interlisp version for their
machine if they get enough orders, but judging from their strong
participation in the Common Lisp effort, I would conclude that Commmon
Lisp is their main compatibility effort.
Finally I believe that the Stanford concentration on the tools for
building a first class public library for Common Lisp will be the major
contribution to reversing the proliferation of LISP variants.
∂04-Nov-82 1507 JMC
To: csd.dorio at SU-SCORE
Thanks, Nancy.
∂04-Nov-82 1622 JMC
To: csd.golub at SU-SCORE
.require "letter.pub[let,jmc]"
∂CSD Professor Gene Golub↓Computer Science Department∞
Dear Gene:
This is to add my recommendation to those for Mike Genesereth.
I would have written earlier if asked, but frankly I didn't think that
there could be any question of his appointment at the assistant
professor level, since I think he is the most promising AI
appointments we have made in a long time.
My reason for saying this is that I believe he is working
on precisely the problem in AI that I think has the greatest
potential for a major breakthrough. This is the representation
of information about situations and goals in predicate logic
coupled with the use of predicate logic to express the statements
that control the search for a solution to the problem.
This is a tough problem, and he may not succeed. However,
it is the fundamental problem that almost all other researchers
are evading. Stanford is in the fortunate position of getting
an almost free ride. If he succeeds we win big. If he doesn't
succeed we may have a hard time making up our minds when a tenure
decision has to be made if the byproducts are at all interesting.
To make explicit comparisons, I rate Genesereth's chance
of making a major breakthrough as better than that of Jerry Feldman,
Cordell Green or Raj Reddy - compared as of similar stages in their
careers. Our other candidate for a major breakthrough in AI is
Lenat, but my estimates of what is likely to succeed would put
Genesereth ahead of him.
I also consider Genesereth to have the personal qualities
of responsibility, intelligence and co-operativeness required for
success in an academic environment, and I see no important
weaknesses.
I have not had much interaction with Mike except for some
discussions about non-monotonic reasoning. When I get circumscription
into good intellectual shape, then I may be pressing him to put
some of the formalism into his system MRS.
.reg
∂05-Nov-82 1351 JMC
To: csd.yearwood at SU-SCORE
Please have her come on Tuesday. I want to look further than those
I have so far interviewed. Sorry about yesterday.
∂05-Nov-82 1527 JMC
To: csd.yearwood at SU-SCORE
11:30 Tuesday is ok.
∂05-Nov-82 1726 JMC
To: csd.golub at SU-SCORE
That's the second copy of U.S. China co-operation memo.
∂06-Nov-82 1141 JMC term project
To: JJW
I have a good term project, but I need your help in writing it up for
the class. A first draft will be in LET1[F82,JMC], but since I haven't
started it yet, I can't say how much there will be when you look at it.
Anyway please phone 857-0672 and/or find me in the office when you get
this.
∂06-Nov-82 1816 JMC
To: RPG
I forgot what you said about when SCHEME will again be available.
∂07-Nov-82 1620 JMC
To: kahn at USC-ISI
I have no objections to your forwarding my message to Balzer.
I talked to Tom Rindfleisch and Ed Feigenbaum about the attitude
of HPP to Common Lisp. Their view is hopeful but still somewhat
skeptical that the divergent interests will come together. No
objections to the proposed content of Common Lisp were expressed,
and this seems to augur well for the acceptance of Common Lisp
if the coalition holds together long enough to finish it. As I
said before, this looks hopeful. We'll get the letter proposal
in as soon as we hear from Ron.
∂07-Nov-82 1839 JMC
To: CLT
Phone when you return.
∂07-Nov-82 2249 JMC
To: RPG
We need to talk about ARPA and about let-by-need.
∂08-Nov-82 0156 JMC transfinite chess problems
To: BYY
Some time ago you mentioned that you knew of chess problems on an
infinite board that white could win in a transfinite number of moves.
Could you (a) cite a simple example and/or (b) give a pointer to the
literature or a person who knows about it?
∂08-Nov-82 0856 JMC
To: REG
Your book is reviewed in March 1982 Computing Reviews.
∂08-Nov-82 2307 JMC
To: CG
The proposal seems adequate. Misspelled perceive, briefly,personnel.
∂09-Nov-82 0054 JMC
To: pony
Send bills for jmc to jmc.
∂09-Nov-82 1619 JMC circumscription
To: CLT
What is your opinion of Reiter's proosal for relating induction schemas
to theories Horn in P and extending it to theories non-Horn in P?
∂09-Nov-82 1521 REITER at RUTGERS circumscription
Date: 9 Nov 1982 1612-EST
From: REITER at RUTGERS
Subject: circumscription
To: jmc at SU-AI
cc: reiter at RUTGERS
Some thoughts on circumscription:
1. As I remarked in my AAAI talk, but not in the paper, the circumscription
schema for an inductively defined predicate P is provably equivalent to the
axiom of structural induction for P. Let me ellaborate by considering as an
example the sufficient conditions defining an S-expression:
(x)ATOM(x) --> S-EXPR(x)
(xy).S-EXPR(x) & S-EXPR(y) --> S-EXPR(cons(x,y))
Circumscribe S-EXPR, replace PHI(x) by S-EXPR(x)&PSI(x), do some simplific-
ation and fidling, and you get
[(x).ATOM(x) --> PSI(x)] & [(xy)S-EXPR(x) & PSI(x) & S-EXPR(y) & PSI(y)
--> PSI(cons(x,y))] --> (x).S-EXPR(x) --> PSI(x)
This is the usual axiom of structural induction for proving properties of
S-expressions. This is the same construction you used for getting the
induction axiom for the natural numbers in your circumscription paper.
(by the way there is a slight error in your statement of the induction
axiom. What you wrote does not follow logically from your axioms.)
Now, what exactly do we mean by an inductive definition? Logicians mean
any set of wffs which are Horn in P. See Barwise's Handbook of Math.
Logic, especially the principle of phi-induction. So my result for theories
Horn in the predicate P says that circumscription is equivalent to the
principle of induction.
2. Maybe that's obvious, although it wasn't to me for a long time. But
notice that if T(P) is a theory Horn in P, then
T(P & PHI) --> (x).P(x) --> PHI(x) (1)
is equivalent to the axiom of induction for the predicate P. So, let's
generalize the induction axiom to arbitrary theories in P to be exactly
the same formula (1). Maybe logicians have considered this although I
doubt it since they are normally interested in inductively defined sets.
So, my proposal is to define the circumscription schema to be (1)
rather than your definition.
3. Does it work? Well, it does for theories Horn in P as I said earlier.
It also works for the two thorny examples I know about:
(a) Circumscribing a disjunction. e.g. your blocks example.
Take PHI(x) = x=A then x=B as you do in your paper, fiddle a bit, and
you'll get your formula.
(b) Circumscribing an existential. e.g. (Ex)P(x)
Intuitively, one wants to derive that there is a unique x s.t. P(x)
i.e.
(Ex)P(x) & [(y)P(y) --> y=x]
Take PHI(x) = (Ey)P(y) & y=x, fiddle a lot, and you'll get this.
What do you think?
Best wishes,
Ray
-------
∂10-Nov-82 1354 JMC postponability
To: SGF
According to Carolyn's interpretation of your question, I didn't answer
it correctly. The right answer is that postponing one variable can
never make a previously postponable variable non-postponable. Since
a variable is postponable if the goals in which it occurs are
simultaneously satisfiable by giving it a value regardless of
what values the other variables have, postponing another variable
can only make postponing it easier by removing some of the goals
that have to be satisfied.
∂10-Nov-82 1719 JMC
To: csd.golub at SU-SCORE
Ah, a new word.
John! This is errific. Can you have it set up as a letter?
∂10-Nov-82 2112 JMC
To: feigenbaum at SUMEX-AIM
There were two Stanford studends, Tom McWilliams and L. Curt Widdoes,
but the S-1 project was launched, on the basis of their ideas,
several years before they completed their theses. Also Mike Farmwald
did a thesis based on his work in designing the arithmetic unit of the
S-1. Widdoes runs a small company, and McWilliams and Farmwald are
still with the S-1 project. I believe that all of them had Baskett
as thesis adviser.
∂10-Nov-82 2239 JMC
To: reiter at RUTGERS
reiter[f82,jmc] To Reiter re circumscription applied to Nixon paradox
Dear Ray:
Here is my treatment of the Nixon paradox. It uses my new notion
of prioritized circumscription and also the idea of circumscribing a formula
rather than a predicate. Thus suppose E{P,x} is some formula involving
a predicate P and a variable x with P satisfying an axiom A{P}.
We write
A'{P} ≡ A{P} ∧ ∀P'.A{P'} ∧ (∀x.E{P',x} ⊃ E{P,x}) ⊃ (∀x.E{P,x} ≡ E{P',x})
and call A'{P} the circumscription of E relative to A with P
as a variable. This readily generalizes to several variables. The formula
A'{P} can be used as the axiom of a further circumscription.
In the Nixon paradox, the axiom A{nnly-pacifist,nnly-non-pacifist,pacifist} is
(1) ∀x. quaker x ∧ ¬nnly-pacifist x ⊃ pacifist x
∧ ∀x. republican x ∧ -nnly-non-pacifist x ⊃ ¬pacifist x
If we make the definition
(2) A(pp,pnp,p) ≡ (∀x.quaker x ∧ ¬pp x ⊃ p x)
∧ (∀x.republican x ∧ ¬pnp x ⊃ ¬p x),
then our axiom (1) is A(nnly-pacifist, nnly-non-pacifist, pacifist). It
makes the convention that a Quaker is to be considered a pacifist
unless some fact prevents this inference,
and Republican is to be considered a non-pacifist
unless the inference is prevented. nnly stands for "not necessarily",
but I hope to find a more perspicuous abbreviation later.
We circumscribe the expressions nnly-pacifist x and
nnly-non-pacifist x with equal priority using nnly-pacifist,
nnly-non-pacifist and pacifist as variables of minimization.
The circumscription formula is then
(3) A(nnly-pacifist,nnly-non-pacifist,pacifist) ∧
∀pp pnp p.(∀x.quaker x ∧ ¬pp x ⊃ p x)
∧ (∀x.republican x ∧ ¬pnp x ⊃ ¬p x)
∧ (∀x.pp x ⊃ nnly-pacifist x)
∧ (∀x.pnp x ⊃ nnly-non-pacifist x)
⊃
(∀x.nnly-pacifist x ≡ pp x)
∧ (∀x.nnly-non-pacifist x ≡ pnp x).
We show that (3) is equivalent to
(4) A(nnly-pacifist,nnly-non-pacifist,pacifist) ∧
∧ ∀x.nnly-pacifist x ≡ quaker x ∧ republican x ∧ ¬pacifist x
∧ ∀x.nnly-non-pacifist x ≡ quaker x ∧ republican x ∧ pacifist x,
which agrees with intuition.
Going from (3) to (4) requires choosing
the predicates to be subsituted for the variables
pp, pnp, and p. We make the substitutions
pp x ≡ quaker x ∧ republican x ∧ ¬pacifist x
pnp x ≡ quaker x ∧ republican x ∧ pacifist x
p x ≡ quaker x ∧ ¬republican x
∨ quaker x ∧ pacifist x
∨ ¬republican x ∧ pacifist x.
A tedious computation, or the propositional calculus routines of EKL,
show that the left side of (3) is then satisfied, and the right side
then tells us.
(5) ∀x.nnly-pacifist x ≡ quaker x ∧ republican x ∧ ¬pacifist x
∧ ∀x.nnly-non-pacifist x ≡ quaker x ∧ republican x ∧ pacifist x.
Plugging these into (1) gives
(6) ∀x.quaker x ∧ ¬republican x ⊃ pacifist x
∧ ∀x.¬quaker x ∧ republican x ⊃ ¬pacifist x,
which is what our intuition tells us should be the result of
taking into account just the facts expressed in (1). In order
to show that (3) doesn't imply anything about people who are both
Quakers and Republicans, we need to show that (6) implies (3) which
haven't yet got around to doing.
I'll send you a paper copy of this, because I suppose you
may have difficulty printing the logical symbols, although Bob
Smith, to whom transmit my regards if he's still there, may know
how to print the SAIL character set.
For reference, the special characters are
∀ for all
∃ there exists
∧ and
∨ or
¬ not
{ left curly bracket
} right curly bracket
I'm pondering your more recent message and will reply later.
Apropos of it Carolyn Talcott remarked that she didn't see how your
form of circumscription would apply to circumscribing an expression.
Also do you have an example where yours and mine give different
results?
Best Regards,
∂11-Nov-82 0044 JMC
To: RPG
There are duplicate pages in common.msg.
∂11-Nov-82 0223 JMC previous message
To: reiter at RUTGERS
1. I can prove equivalence of your and my versions
of circumscription.
2. I don't see the bug in circumscribing (Ex)P(x) by either method.
∂11-Nov-82 0940 JMC schedule
To: "@EXAM.DIS[F82,JMC]"
It turns out I have a meeting at 2pm. Therefore, I propose the
following schedule.
Finlayson 9am
Moses 10:30
Weening 12:30
If this is inconvenient, we could schedule one of the exams for Tuesday.
∂11-Nov-82 1141 JMC
To: "#JMCJNK.MSG[1,JMC]"
foo
∂12-Nov-82 0922 JMC name change on lists
To: chappell at SRI-AI
Please change my name on the lists AIC-ASSOCIATES and TINlunchers from
JMC@SAIL to JMC-LISTS@SAIL. For individual messages, I'm still
JMC@SAIL, but the change will separate my mail coming to lists from
individual messages. Many thanks.
∂12-Nov-82 0928 JMC name change for mailing lists
To: CLT
Please change my name on the seminar notice list to JMC-LISTS. This will
separate announcements from individual mail.
∂12-Nov-82 0935 JMC
To: cschmidt at SUMEX-AIM
Please change my name on the list HPP-Lisp-Machines to JMC-LISTS@SAIL.
I'm still JMC for individual messages, but I want to receive mail
directed to mailing lists separately.
∂12-Nov-82 1643 JMC
To: ARK
On p.72 yumyum gives a 708 area code where 408 is meant.
∂12-Nov-82 2004 JMC distribution list name
To: reid at SU-SHASTA
For the purpose of inclusion on distribution lists, I am JMC-LISTS@SAIL.
This to separate mail received as part of a mailing lists from individual
messages.
For the latter I remain JMC@SAIL.
∂12-Nov-82 2009 JMC
To: ZM, RPG, LGC, CG, CLT, LGC, JJW, YOM
CC: ARK
Diana Hall, our new secretary, will start in about a week.
∂12-Nov-82 2253 JMC
To: JK
Diana Hall, our new sec'y, will start in about a week.
∂13-Nov-82 1629 JMC
To: kmp at MIT-MC
Of your message of 0232, 13 Nov., copying common-lisp, only the header
made it.
∂14-Nov-82 0114 JMC message for Gloria
To: minsky at MIT-AI
Tell her there's a new P.D. James mystery, and that her books are becoming
a sort of grownup Nancy Drew.
∂14-Nov-82 1601 JMC
To: CLT
I just woke up to fact that I forgot Guarneri. Sorry! Please phone.
∂14-Nov-82 1624 JMC
To: JJW
Are the LIsp axioms at LOTS?
∂14-Nov-82 1916 JMC change of name in mailing lists
To: csd.broder at SU-SCORE
Please replace my name in the AFLB mailing list by JMC-LISTS@SAIL. For
messages sent individually, it is still JMC@SAIL.
∂14-Nov-82 2018 JMC
To: RPG
bad M.I.T., naughty M.I.T.
∂15-Nov-82 0005 JMC
To: ZM
OK, though the students were expecting you.
∂15-Nov-82 0050 JMC bug
To: JK
The permut.prf attempts a rewrite as its last step. I don't understand
why nothing happens.
∂16-Nov-82 1140 JMC
To: kmp at MIT-MC
Yes, I received them, and thanks. I haven't received any T material though.
∂16-Nov-82 1546 JMC
To: JK
Proof? ;Done.
;LISPAX read in
;PERMUT read in
Proof? switched to PERMUT
20.
(get-proofs permut)
(proof permut)
;;; permut.lsp[f82,jmc] ekl axioms and proofs for permutation functions
(get-proofs lispax)
(proof permut)
(decl rac (unaryname: rac) (type: |ground→ground|) (syntype: constant)
(bindingpower: 950))
(decl rdc (unaryname: rdc) (type: |ground→ground|) (syntype: constant)
(bindingpower: 950))
(decl snoc (type: |(ground⊗ground)→ground|) (syntype: constant))
(axiom |∀u.rdc u = if null cdr u then nil else car u . rdc cdr u|)
(label definfo)
(axiom |∀u.rac u = if null cdr u then car u else rac cdr u|)
(label definfo)
(axiom |∀x u.snoc(x,u) = u * x.nil|)
(label definfo)
(label snoc_def)
(decl reverse (unaryname: reverse) (type: |ground→ground|) (syntype: constant)
(bindingpower: 950))
(decl rev (type: |ground⊗ground→ground|) (syntype: constant))
(axiom |∀u.reverse = rev(u,nil)|)
(label reverse_rev)
(axiom |∀u v.rev(u,v) = if null u then v else rev(cdr u, car u . v)|)
(label definfo)
(axiom |∀u.reverse u = if null u then nil else reverse cdr u * car u . nil|)
(label reverse_def)
(label definfo)
(axiom |∀u.reverse reverse u = u|)
(label simpinfo)
(label reverse_reverse)
(axiom |∀u v.reverse(u*v) = reverse v * reverse u|)
(label reverse_append)
(decl lcycle (unaryname: lcycle) (type: |ground→ground|) (syntype: constant)
(bindingpower: 950))
(decl rcycle (unaryname: rcycle) (type: |ground→ground|) (syntype: constant)
(bindingpower: 950))
(axiom |∀u.lcycle u = if null u then nil else snoc(cdr u, car u)|)
(label definfo)
(axiom |∀u.rcycle u = if null u then nil else rac u . rdc u|)
(label definfo)
;end_defs
(trw |reverse snoc(x,u)|
((use snoc_def (mode: t))
(use (reverse_append lispax#sortinfo) (mode: t))) sortinfo)
REVERSE SNOC(X,U)=REVERSE U*REVERSE (X.NIL)
(trw |reverse (x.nil)| (use reverse_def (mode: allways)) sortinfo simpinfo)
(trw |listp (x.nil)| nil sortinfo simpinfo)
∂17-Nov-82 0120 JMC permut again
To: JK
Would you take a look at the last bit of permut.lsp and tell me why
(trw |x.nil * reverse u| (use append_def (mode: t)) sortinfo simpinfo)
does nothing. I want, of course, x.reverse u. I'll be back from
L.A. tonight.
∂17-Nov-82 0126 JMC a bit more on permut
To: JK
It looks like one needs the to use the definitions separately as well
as as part of definfo.
2. It may be grumbled at that you allow and use "-" in names like
get-proofs, but require the user to use "_" in labels.
∂17-Nov-82 1843 JMC
To: bobrow at PARC-MAXC
I'll look for Dgano and Sirovich paper.
∂18-Nov-82 0003 JMC
To: PJH
Does the message about the Ritchie paper meet your need?
∂19-Nov-82 1018 JMC mailing list names
To: csd.golub at SU-SCORE
Please change my name on the tenured faculty electronic mailing list
to JMC-LISTS@SAIL. I have found it convenient to receive mail from lists
separately from mail sent individually to me. You might find something
similar convenient.
∂19-Nov-82 1024 JMC tenure
To: faculty at SU-SCORE
My immediate reaction is that the present system of tenure should be
continued. The reason is to not dilute the effort now put into tenure
decisions. The Dean suggests a "hard decision" after 4 years, which
would presumably require an effort similar to that involved in a tenure
case. My view is that the effort involved in doing an honest job in
the present system is barely sustainable. The effect of going to a more
elaborate system is that both the mini-tenure decision and the real
tenure decision might tend to become perfunctory.
∂19-Nov-82 1202 JMC
To: DCL
CC: HST
I put the Stoyan manuscript in interdepartmental mail yesterday.
Hopefully, you'll get it within a week.
∂19-Nov-82 1445 JMC
To: RWW
Certainly at SRI-AI, but I had one put up on SCORE a year ago. Maybe
it's still available. It was on MRC's disk pack, and I forget how
to call it. If you have the time, you could put the SRI version
on SAIL.
∂22-Nov-82 0955 JMC
To: LMG at SU-AI
I have created a directory for Diana Hall as dfh
∂22-Nov-82 1345 JMC refereeing
To: bobrow at PARC-MAXC
I have passed on the Degano and Sirovich paper to Chris Goad here, who
has agreed to referree it. I don't know enough about the proposals
for inductionless induction to do a good job on it myself. I the
paper by Conrad Kuck "A logic for default reasoning" still under
consideration. I remember that you told me some paper had already
been rejected, but I don't think this was it.
∂23-Nov-82 2009 JMC references
To: bmoore at SRI-AI
I'm teaching CS226, epistemology of AI, next quarter.
1. What are the best references to your relevant work for the
students to read?
2. What other work might you recommend that they read?
∂24-Nov-82 1257 JMC Business Language
To: DFH
∂24-Nov-82 0917 MCCARTY at RUTGERS Business Language
Date: 24 Nov 1982 1217-EST
From: MCCARTY at RUTGERS
Subject: Business Language
To: jmc at SU-AI
cc: mccarty at RUTGERS, nilsson at SRI-AI
Nils Nilsson was here at Rutgers last week, and he told me about some work
you were doing on a Common Business Communication Language (CBCL). Could you
please send me anything you have written on the subject? From Nils'
description, it sounds as if there might be some relationship between CBCL
and the representations of legal concepts which we have developed here in the
TAXMAN project. If so, I would very much like to pursue this further.
Regards,
L. Thorne McCarty
-------
∂24-Nov-82 1446 JMC
To: PJH
What's your home phone? I get lively too late to phone at work.
∂25-Nov-82 2336 JMC
To: TOB
I was formerly a member and dropped out about 15 years ago.
They are interested in academic freedom, and also tend
to take a trade unionist - us and them - attitude towards
university administrations. I might have dropped out when,
as I recall, they supported Bruce Franklin, but I had dropped
out some years previously. I suggest you look at some issues
of their bulletin in the library in order to determine whether
their current concerns interest you.
∂26-Nov-82 0953 JMC
To: bobrow at PARC-MAXC
I have been following Brian's work to some extent. At the moment I don't
see the connection. Can you say a bit more about what you think it is?
∂27-Nov-82 1326 JMC
To: JJW
Looks like you're on the right track.
∂27-Nov-82 1745 JMC more
To: JJW
I agree that fixed numbers of arguments will cover the useful cases.
Why don't you carry your reverse example just a bit farther and
get out the minimization schema by simplification also? I can talk
about the minimization schema in class if things are in good shape
in time.
∂29-Nov-82 1054 JMC
To: JJW
I had already noticed that the minimization schema "essentially contained
in line 17". I guess I want an example, e.g. loop x, where line 17 gets
expanded.
∂29-Nov-82 1146 JMC oversight
To: PJH
I forgot to say that if one is conveniently available, I would like
another copy of Haas's thesis to put on reserve for my class this
winter. Mine is covered with rude remarks in red ink.
∂29-Nov-82 1324 JMC
To: bmoore at SRI-AI
I would be grateful for one copy of each of the references you mentioned.
How about your paper with Hendrix also. By the way have you seen a
Rochester PhD thesis by Andrew Haas entitled "Planning Mental Actions",
and, if so, what is your opinion of it?
∂29-Nov-82 1628 JMC
To: csd.atkinson at SU-SCORE
The smaller blackboard will be ok.
∂29-Nov-82 1629 JMC consulting bill
To: DFH
Invoice
Decision focus
attn: Steve Barrager
consulting: Friday, Novermber 26, 1/2 day at $500 per day, $250.
address in phon[1,jmc]
∂29-Nov-82 1631 JMC
To: DFH
Use my home address 846 Lathrop Dr. Stanford 94305 for consulting bills.
∂01-Dec-82 1125 JMC
To: BH
Surely you meant % before ISID on p.92 of bboard?
∂01-Dec-82 1653 JMC thesis
To: ARG
Here are some issues that you should address. They should be
addressed in general without bothering to mention those reasons,
such as the state of PDP-11 software, that induced the particular
decisions you made when the project was undertaken.
1. Should environments for special purpose programming, e.g.
manipulator control, be entirely separate from general purpose
programming? Is a special editor and operating environment
really required?
2. What features must a general purpose environment have in
order to be adaptable to special purpose physical world
control programming? What would LISP or Algol require for this?
3. As a separate question, is structured editing really a winner?
∂01-Dec-82 2048 JMC
To: jmc at S1-A
foo
∂02-Dec-82 1619 JMC
To: csd.armer at SU-SCORE
Jan. 11 is ok for me.
∂02-Dec-82 2039 JMC
To: CLT
Susie's number is (56) 293-7498, i.e. 408 293-7498.
∂03-Dec-82 1711 JMC
To: ME
;/from wed/after 10pm says illegal date range
∂04-Dec-82 1829 JMC
To: JDH
I would like a copy of 204 problem writeups for this quarter.
∂05-Dec-82 0003 JMC
To: ME
Downtime: 0 min., availability: 100%
No service breaks
Outside air temperature:
High -572 C., -999 F. at 0:00 on 1 Jan '64
Low 537 C., 999 F. at 0:00 on 1 Jan '64
Mean -17 C., 0 F.
∂05-Dec-82 1338 JMC
To: reiter at RUTGERS
Ray:
Here is Carolyn's proof that your schema implies mine. We
have yours
(R) (phi)(T(P&phi) ⊃ (x)(P(x) -→ phi(x))
and we want to prove mine
(J) (phi)(T(phi) & (x).(phi(x) -→ P(x)) -→ (x)(P(x) -→ phi(x)).
So assume T(phi) and (x)(phi(x) -→ P(x)).
The latter formula is equivalent to (x)(P(x)∧phi(x) iff phi(x)),
from which follows T(phi) iff T(P&phi). Hence we can use (R)
to derive the conclusion of (J).
Sorry moving to Rutgers didn't work out, and I hope absence from
Vancouver has made its virtues more apparent.
Regards,
John
∂06-Dec-82 0005 JMC
To: PJH at SU-AI, reiter at RUTGERS
Each of the several papers that introduces a mode of non-monotonic reasoning
seems to have a particular application in mind. Perhaps we are looking
at different parts of an elephant. Here is a typology of non-monotonic
reasoning about which I solicit comments. This is intended as part of
a larger paper. The orientation is towards circumscription, but I suppose
the considerations apply to other formalisms as well.
Circumscription may have several uses.
1. As a communication convention. Suppose A tells B about
a situation involving a bird. If the bird may not be able to fly, and this
is relevant to solving the problem, then A should mention
the relevant information. Whereas if the bird can fly, there is
no requirement to mention the fact.
The circumscriptions to be made by the recipient of the communication
are those described in this paper. I have not yet considered how
the sender computes what to say and how this refrains from saying
that the bird can fly - presumably without any explicit decision.
2. As a database or information storage convention. It may be a
convention of a particular database that certain predicates have
their minimal extension. This generalizes the closed world
assumption. One such would be nnly-can-fly in the case of information
about birds.
Neither 1 nor 2 requires that most birds can fly.
Should it happen that most birds that are subject to the communication
or about which information is requested from the data base cannot fly, the
convention may lead to inefficiency but not incorrectness.
3. As a rule of conjecture. This use was emphasized
in (McCarthy 1980). The circumscriptions may be regarded as expressions of some
probabilistic notions such as "most birds can fly" or they may be
expressions of simple cases. Thus it is simple to conjecture that
there are no relevant present
material objects other
than those whose presence can be inferred. It is also
a simple conjecture that a tool asserted to be present is usable
for its normal function. Such conjecture sometimes conflict, but there
is nothing wrong with having incompatible conjectures on hand. Besides
the possibility of deciding that one is correct and the other wrong, it
is possible to use one for generating possible exceptions to the other.
4. As a representation of a policy. The example is Doyle's "The meeting
will be on Wednesday unless another decision is explicitly made".
5. As a very streamlined expression of probabilistic information when
numerical probabilities, especially conditional probabilities, are
unobtainable. Since circumscription doesn't provide numerical probabilities,
its probabilistic interpetation involves
probabilities that are either infinitesimal, within an
infinitesimal of one, or intermediate - without any discrimination
among the intermediate values. The circumscriptions give conditional
probabilities. Thus we may treat the probability that a bird
can't fly as an infinitesimal. However, if the rare
event occurs that the bird is a penguin, then the conditional probability that
it can fly is infinitesimal, but we may hear of some rare condition
that would allow it to fly after all.
Why don't we use finite
probabilities combined by the usual laws? That would be fine
if we had the numbers, but circumscription is usable when we can't
get the numbers or find their use inconvenient. Note that the
general probability that a bird can fly may be irrelevant, because
we are interested in particular situations which weigh in favor or
against a particular bird flying. Of course, circumscription
does not provide a means of weighing evidence; it is appropriate
when the information permits snap decisions. However, many
cases nominally treated in terms of weighing information are in fact
cases in which the weights are such that circumscription and other
defaults work better.
6. We might also speculate that certain laws of common sense physics
or common sense psychology is inherently non-monotonic or, more specifically,
involves circumscription. The speculation is that this common sense
information has some inherently preferred form.
Six different uses for non-monotonic reasoning seems
too many, so let's see if we can condense them. Maybe
the first two and even the fourth are linguistic conventions.
∂06-Dec-82 1035 JMC
To: darden at SUMEX-AIM
I used the phrase "epistemological engineering" in early 60s, and I
remember that Jerome Lettvin had "epistemological engineering" on
the door of his laboratory a little later. I don't know that he
got it from me, however. "Experimental epistemology" seems to me
quite recent. Also "applied epistemology" has been used.
How are things?
∂06-Dec-82 1239 JMC
To: csd.stewart at SU-SCORE
I am interested in your SAIL Datamedia.
∂07-Dec-82 1250 JMC
To: LGC
PACWAR is in 204,jdh
∂08-Dec-82 0942 JMC change of name on tinlunchers list
To: chappell at SRI-AI
Please change my name on the TINLUNCHERS mailing list from
JMC@SAIL to JMC-LISTS@SAIL. I prefer to get announcements
separately from individual messages.
∂08-Dec-82 1903 JMC riding
To: sally at 10444
My daughter Sarah McCarthy, 916 756-1103, at Davis knows a lot about it.
∂09-Dec-82 1147 JMC
To: YM
fing cbf%s1
Person Last logout
CBF Charles Frankston 10:02 on 4 Dec 1982. Plan:
Going back to the 'tute. Don't expect any logins until early Feb.
Send mail to rmf@Multics to contact me; or call (617) 969-1997.
↑C
.
∂09-Dec-82 1148 JMC terminal recommendation
To: jbr at S1-A
I want to replace my Datamedia home terminal by something to be used
with SAIL. Do you have a current recommendation of a terminal
or of an expert? I was quite happy with the Ambassador with the M.I.T.
keyboard, but I don't know how hard it is to get a suitable combination
for use over phone lines. Am I correct that the S-1 people all prefer
to come in rather than work from home?
∂09-Dec-82 2324 JMC
To: jbr at S1-A
Does the Ambassador so equipped handle the SAIL character set?
∂10-Dec-82 1643 JMC telephone terminal
To: EJG
I want to replace my Datamedia by something better for using SAIL over
a phone line. Jeff Rubin thought you might know something about whether
there is some way of getting an Ambassador or something else better than
a Datamedia to handle the SAIL character set.
∂10-Dec-82 1646 JMC
To: jbr at S1-A
Thnaks.
∂10-Dec-82 2016 JMC
To: IAZ
I'd like to look at your bike - the one that doesn't need work. Do you
bring it to MJH?
∂10-Dec-82 2133 JMC Christmas
To: darden at SUMEX-AIM
I'm sorry to hear about your father. I'm indeed the grandfather
of Katherine Rose McCarthy; the next will have last name Gunther.
No doubt this will confuse some institutions.
If you are in a mood to be telephoned, MAIL a number.
∂12-Dec-82 1457 JMC ideas for final
To: JJW
I'll get to it tomorrow. HOwever, you might look at old finals which
are in directories, [fxy,jmc]. I don't want to use old problems, but
they are often suggestive.
∂13-Dec-82 0025 JMC
To: TVR
What's new about mail?
∂14-Dec-82 0017 JMC
To: jmc at S1-A
foo
∂15-Dec-82 0036 JMC
To: jmc-lists
ffo
∂15-Dec-82 0038 JMC
To: walker at SRI-AI
My address for lists is jmc-lists not jmc-list.
∂15-Dec-82 1602 JMC reply to letter from Harry Stanton
To: bradford@mit-oz at MIT-MC
Harry:
Thanks for your letter of 1982 December 9. I am now writing
a new paper on non-monotonic reasoning, and it comes before finishing
the material on knowledge which I want to include in the collected
papers. Progress is slow, and I don't think it is appropriate to
start any clocks ticking at your end.
Progress on the LISP book seems to await Carolyn completing
her thesis, although bits of it are being revised in accordance with
my classroom experience. Most of the comments you have obtained are
well taken, and we will take many of the suggestions.
I think it is appropriate to include "Inversion of functions
defined by Turing machines", and I'll do it.
Many thanks for your patience, but I fear lots more of it
will be required.
Please let me know when you get this. I'm jmc@su-ai, but
there may be some problems for a while after the first of the
year if they really change protocols on schedule.
Best regards,
∂15-Dec-82 1618 JMC
To: berwick@mit-oz at MIT-MC
For Harry Stanton:
Harry:
Thanks for your letter of 1982 December 9. I am now writing
a new paper on non-monotonic reasoning, and it comes before finishing
the material on knowledge which I want to include in the collected
papers. Progress is slow, and I don't think it is appropriate to
start any clocks ticking at your end.
Progress on the LISP book seems to await Carolyn completing
her thesis, although bits of it are being revised in accordance with
my classroom experience. Most of the comments you have obtained are
well taken, and we will take many of the suggestions.
I think it is appropriate to include "Inversion of functions
defined by Turing machines", and I'll do it.
Many thanks for your patience, but I fear lots more of it
will be required.
Please let me know when you get this. I'm jmc@su-ai, but
there may be some problems for a while after the first of the
year if they really change protocols on schedule.
Best regards,
John
∂15-Dec-82 2116 JMC
To: berwick.bradford@mit-oz at MIT-MC
This is a test to see if address works.
∂16-Dec-82 1341 JMC
To: pattermann at SUMEX-AIM
If the JMCCARTHY@SUMEX is me, mail should be (and should have been)
forwarded to JMC@SAIL. Actually I didn't know I had an account at
SUMEX. I might use it once in a great while if I knew its password.
∂18-Dec-82 1817 JMC new ekl
To: JK, JJW
Is the new EKL called EKL, and what is the state of writeup?
∂19-Dec-82 0123 JMC
To: JJW
I'll come in and help finish off this morning or afternoon.
∂19-Dec-82 1437 JMC
To: DFH
jonas.1
∂20-Dec-82 1046 JMC
To: bboard
Programmer wanted to adapt Dover version of Pub to WAITS. Money.
∂20-Dec-82 1525 JMC
To: aaai-office at SUMEX-AIM
Please change my name on AAAI-DISTRIB from JMC@SU-AI to JMC-LISTS@SU-AI.
I prefer to receive mail sent via lists separately from person-to-person
mail.
∂20-Dec-82 1633 JMC British Museum
To: RJT
Is the British Museum a library or is it a museum that contains a library -
for a reference to the British Museum algorithm?
∂20-Dec-82 2247 JMC
To: RPG
The formal proposal has to contain exactly what they decide to give you.
Indeed whenever they want to change something in a proposal, they make
us submit a new one proposing what they intend to give us. I suppose
the idea is that when the proposal goes up to the Director of ARPA,
there should be no confusion about "they propose A, but we want to
give them B".
However, you can send Ohlander a message saying whatever you think
appropriate.
∂21-Dec-82 1542 JMC
To: MRG
Remember you owe me a list of the requesters and users of MRS.
∂21-Dec-82 1556 JMC
To: berwick.bradford@mit-oz at MIT-MC
Harry:
I had "Inversion of functions defined by Turing machines" typed
into the computer. As you will recall, I intend to provide each of
the papers with a commentary giving my present point of view on its
subject matter. Here is a draft of the commentary for Inversion.
Have you, as editor, comments on the commentary?
John
turing.com[f82,jmc] Commentary on "The inversion ..."
Commentary:
This paper was written in the summer of 1952 and elaborated
somewhat before it was published in 1956. It was written to clarify
the meaning of "problem" and to discuss problem solving methods that
work directly from the definition of problem. A problem was
was called well-defined if a procedure was given for testing a
proposed solution, and the problem solving methods worked backwards
from this test.
Thus no auxiliary facts of general common sense knowledge or of
the specific domain are used.
Even at the time the idea didn't seem promising. I cannot
reconstruct my state of mind at the time, but I spent most of my
time thinking about methods that involved knowledge of facts, and
the ideas that were first published with my 1958 paper "Programs
with Common Sense". However, I was unable to make these ideas sufficiently
definite in the summer of 1952, so I decided to explore this more
limited domain.
There is nothing about computers
in the "Inversion ... " paper. This is because I, in common at least
with Minsky and Shannon with whom I discussed the ideas,, thought about
artificial intelligence in terms of machines, not programmable computers,
about which we knew little. Even von Neumann, who did know about
programmable computers, seemed to think about thinking machines
exclusively as objects analogous to the human brain. To my knowledge,
only Turing then considered programming computers as the technique
for realizing artificial intelligence. Newell and Simon were next
in 1954, and I switched my attention to computers in 1955, during a
summer spent working for IBM in Poughkeepsie, New York.
Perhaps some part of the idea could be rehabilitated, but
it would seem appropriate to use Lisp functions or logic programs
as the way to represent the function to be inverted. The paper
does not discuss different representations of the same function
nor does it discuss the natural restriction (as long as one hand
is to be tied behind the back) of using the function to be inverted
only extensionally, i.e. by computing values, although some of the
methods observe this restriction.
∂21-Dec-82 1653 JMC Ordering conjuncts ...
To: de2smith at SUMEX-AIM
CC: MRG at SU-AI
Mike gave me a copy, and I have skimmed it. You can refer to my
AI Memo rather than to "personal communication". Besides that I
know considerably more about meta-reasoning in this case that will
further complicate your considerations. You should also refer to
the Pereira and Porto "Intelligent backtracking ..." and to several
other Prolog papers (Pereira and also Gallaire). Brian Smith's
ideas on reflection are also relevant.
I like the paper, and after I read the formulas in detail I may have
specific comments. When you have time to come around I can tell you
about the "semantic meta-reasoning" required to supplement the
"syntactic meta-reasoning" involved in postponement when we want to
postpone countries that have 4 neighbors in addition to those with
3 or fewer neighbors.
∂21-Dec-82 2149 JMC
To: TOB
CC: GHG
∂21-Dec-82 2010 TOB industrial professor
To: JMC, TOB
I have some candidates. I will contact them if
it is still a possibility.
I have only one definite candidate so far, and I want to
get them in time to put the course descriptions in the catalog, so
go ahead.
∂22-Dec-82 1048 JMC
To: pourne at MIT-MC
Jerry: According to finger, JXP still exists as Jerry Pournelle and
last logged in on Oct. 28. Can't check further without a password.
If you get this in time before I leave for Seattle Thursday at 2pm
phone me at 415 497-4430 or 857-0672 and we'll investigate further.
I don't remember about Silicon Brains, but I guess everything but
timing is evident from the title. It's possible.
∂22-Dec-82 1049 JMC incidentally
To: pourne at MIT-MC
Incidentally, I once heard the phrase "silicon based life" from
Barney Oliver. It makes a nice switch on the old science fiction
notion of the slow Martian monsters based on silicon pseudo-organic
chemistry.
∂23-Dec-82 1309 JMC cs226
To: DFH, YOM
I'm going to be at the Math Society meeting in Denver on Thursday, Jan 6,
so the first meeting of CS226 should be on the following Tuesday.
It should be called an organization meeting, and it needs a room and
time. 1:15 is a suitable time, but another will do also. The announcement
should be on CSD physical and electronic bulletin boards and should also
go to the philosophy department which has cross-listed the course. It
will be mainly based on papers, mostly mine. Yoram, please have 20
copies each of two papers "Programs with Common Sense" and "Inversion
of Functions Defined by Turing Machines" ready for distribution at the
time of the first class. Yoram, also look at CS226[f82,jmc] which contains
some notes. I'll be back Dec. 31, and we should get together as soon as
possible after that.