perm filename E85.OUT[LET,JMC] blob
sn#807002 filedate 1985-10-01 generic text, type C, neo UTF8
COMMENT ⊗ VALID 00002 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00002 00002
C00003 ENDMK
C⊗;
∂01-Jul-85 1606 JMC
To: RA
MAIL me Zohar's numbers; I may call from home. Also what's the time difference.
∂01-Jul-85 1644 JMC Next week's PLANLUNCH -- WEDNESDAY July 3
To: VAL
∂27-Jun-85 1228 @SU-CSLI.ARPA:LANSKY@SRI-AI.ARPA Next week's PLANLUNCH -- WEDNESDAY July 3
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 27 Jun 85 12:27:50 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Thu 27 Jun 85 12:21:43-PDT
Date: Thu 27 Jun 85 12:21:03-PDT
From: LANSKY@SRI-AI.ARPA
Subject: Next week's PLANLUNCH -- WEDNESDAY July 3
To: aic-associates@SRI-AI.ARPA, folks@SU-CSLI.ARPA
"HOW TO CLEAR A BLOCK"
or
Unsolved Problems in the Blocks World #17
Richard Waldinger -- SRI AI Center
11:00 am, WEDNESDAY, July 3 (notice change in day)
Room EJ232, SRI International
ABSTRACT:
Apparently simple problems in the blocks world get more complicated
when we look at them closely. Take the problem of clearing a block.
In general, it requires forming conditionals and loops and even
strengthening the specifications; no planner has solved it.
We consider how such problems might be approached by bending a
theorem prover a little bit.
-------
∂01-Jul-85 1721 JMC re: yet another letter to the president (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - What is required to get the Lebanese government or the Syrians or
the Shiites to co-operate in defeating terrorists is a credible threat.
Curiously enough we have one. We can credibly threaten to change our
Mideast policy from one (mostly) of justice to one of alliance. Our
policy has been one of calling each situation as we see it. We can change
to one of getting allies and sticking to them even when they pursue their
interests in ways we find unjust. In this case the obvious allies are the
Israelis and the Lebanese Christians (or some faction of them). The
threat to more frankly ally with Israel is not very strong at present,
although a year ago it would have been, because they have withdrawn from
Lebanon, and we're already helping them to get military equipment.
However, the threat to take sides with the Christians and supply them with
modern equipment, training and perhaps air and naval artillery support has
plenty of teeth. It may be that the release of the hostages achieved so
far has involved such threats sotto voce.
As a general remark, as long as we were powerful enough and determined
enough to play the role of world policeman, such a policy was inappropriate.
A policeman should be fair. However, now that we are not strong enough
to be policeman, we have to revert to the more primitive and more usual
form of foreign policy - a policy of alliance.
∂01-Jul-85 1735 JMC re: Editors, hacking, etc. (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - Consider the in-group advantages of using the old meaning of "hacker",
while the general public, the media, IBM and the official publications
of the ACM use the new meaning. When some newcomer to the programming
group or the department or club uses the meaning in the vulgar sense,
he can be put down for weeks. He'll soon learn to mind his manners.
A kind reply would be, "Oh, you mean a password hacker", but surely the
you bboard hackers can think of more witty retorts.
∂01-Jul-85 1803 JMC re: question on very long term electronic storage (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - Obsolescence is more of a problem for present electronic media than
decay. Therefore, an electronic archive should plan to copy its files
every so often. For example, SAIL still has 7 track tape drives 20 years
old and its archives on 7 track tape. I don't know any other tape drives
at Stanford that can read them. My guess is that copying them to
6250 bpi 9 track tapes will keep them readable for another 30 years.
However, in a few years we may be able to keep all those files on-line.
It will then be a routine part of installing an improved storage
technology to copy the archives.
∂01-Jul-85 1857 JMC Rider wanted for Throughbred hunter
To: SMC
∂26-Jun-85 0910 @SU-SCORE.ARPA:jbn@FORD-WDL1 Rider wanted for Throughbred hunter
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 26 Jun 85 09:10:37 PDT
Received: from FORD-WDL1 by SU-SCORE.ARPA with TCP; Wed 26 Jun 85 09:09:42-PDT
Return-Path:<>
Date: 25-Jun-85 15:38:45-PDT
From: jbn@FORD-WDL1.ARPA
Subject: Rider wanted for Throughbred hunter
To: SU-BBOARDS@SU-SCORE.ARPA
I need an experienced rider to help keep my Thoroughbred hunter in shape.
I ride him on the weekends and one or two evenings per week, but it isn't
enough; he needs to be ridden every day to keep him in shape. So I can offer
someone the opportunity to ride a good horse at no cost. My previous rider
has just graduated from Stanford, so I need a replacement. The horse is
a big, athletic animal with a strong personality; he needs a strong rider
with jumping and eventing experience. He also needs more flat work and
dressage, and has a tendency to rush his fences. He's an excellent trail
horse, and is good in open country. He has never been hunted, though.
You must be 21 or over and have good medical insurance.
I need someone who can ride at least three weekdays per week; evenings
are fine. A covered, illuminated arena is available, and there is excellent
trail access; the horse is stabled in Portola Valley, near Stanford.
I can be reached at 415-856-0767 evenings, or at "jbn@FORD-WDL1.ARPA"
or "wdl1!jbn".
It would be appreciated if someone would repost this to LOTSland
(exactly once).
John Nagle
∂01-Jul-85 1904 JMC re: yet another letter to the president (from SAIL's BBOARD)
To: BAMPI@SU-SIERRA.ARPA
[In reply to message sent Mon 1 Jul 85 18:37:03-PDT.]
Your message did not get on the SAIL bboard, because it was mailed to
OTHER-SU-BBOARDS@SAIL. Had it been mailed to SU-BBOARDS@SAIL, the
SAIL bboard would have been included. I'll put it there and get around
to answering it.
∂01-Jul-85 2114 JMC Unfortunately, the victim didn't have a gun.
To: su-bboards@SU-AI.ARPA
a259 1710 28 Jun 85
AM-Ramseur, 1st Ld, a225,0233
By GARY LANGER
Associated Press Writer
NEW YORK (AP) - James Ramseur Jr., one of the four teen-agers shot
by Bernhard Goetz on a subway train, was arrested Friday on charges
of raping and robbing a young woman at gunpoint last month,
authorities said.
Police arrested Ramseur, 18, after the victim - who was hospitalized
for four days after the attack - identified him in a lineup at the
48th Precinct in the Bronx, law enforcement sources told The
Associated Press.
Bronx District Attorney Mario Merola, who announced the arrest, said
in a statement that Ramseur allegedly accosted the 18-year-old victim
in an elevator in her building at the Claremont Houses, a Bronx
housing project.
Ramseur, who lives in another building in the same project,
allegedly forced the young woman to a landing leading to the roof of
the building, raped her, took her earrings and three rings, and fled,
Merola said.
Police said Ramseur was arrested on charges of rape, sodomy,
robbery, assault and criminal possession of a gun. A source close to
the investigation said police were seeking a second man said to have
participated in the attack.
Ramseur was one of the youths wounded by Goetz in a widely
publicized shooting on Dec. 22. Goetz, 37, said he fired in
self-defense as the four youths menaced him, but the four said they
were only panhandling for money.
A initial grand jury refused to indict Goetz for shooting the
youths, but a second grand jury impaneled on the basis of new
evidence indicted him on charges including attempted murder and
assault. An appeals court refused to dismiss that indictment
Thursday.
∂02-Jul-85 0152 JMC letter to Prof. Mitchell Roth
To: RA
Enclosed is a copy of the Qlisp proposal that may interest you. I assume
you've mailed the papers I lent you. Sorry I missed you on Thursday.
∂02-Jul-85 0152 JMC more to roth and another letter
To: RA
Thanks for your hospitality and my regards to Diane.
to: Professor Thomas Head
(same address as Roth)
Many thanks to you and Eileen for your hospitality in
Fairbanks. Perhaps we'll see you here or in Washington.
Best Regards,
∂02-Jul-85 1333 JMC re: my paper on CWA and circumscription
To: VAL
[In reply to message rcvd 02-Jul-85 13:20-PT.]
That's correct. Rutie will tell you the exact form.
∂02-Jul-85 1409 JMC re: Fred Riss, IBM
To: RA
[In reply to message rcvd 02-Jul-85 09:54-PT.]
11 on Tuesday is ok.
∂02-Jul-85 1453 JMC for Claudia
To: aaai@SUMEX-AIM.ARPA
It occurs to me that one of the points we discussed in our phone
conversation was left slightly at a loose end. Namely, my view
is that since there is no other organization involved in sponsoring
the conference, we could avoid that part of AAAI's machinery and
that Chudnovsky and Jenks would deal directly and only with you.
This isn't a matter of principle with me so that you should
let Marty know if you think that otherwise he might be offended.
I just want to keep matters as simple as possible.
David will or already has phoned you.
∂02-Jul-85 2040 JMC re: yet another letter to the president (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - Neither of the criticisms is relevant to the question of whether
threatening to support the Christians in Lebanon would be effective
in getting the various Moslem factions to discourage attacks on our
citizens.
Responding to the criticisms made, I won't defend every U.S. policy, but
I will claim that our policy in Lebanon made attempts to be fair at the
cost of hindering our partial ally Israel from accomplishing its objectives.
It seems to me that this was a mistake. As to Mr. Bampi's journey through
space and time, I have defended many U.S. actions, including some of those
he cites, on BBOARD, and I won't repeat these defenses now.
∂02-Jul-85 2059 JMC reply hack
To: su-bboards@SU-AI.ARPA, juliao@SU-GREGORIO.ARPA,
treitel@SU-SUSHI.ARPA
When someone uses the REPLY hack in TOPS-20 or UNIX to reply to a message
on SAIL's BBOARD, then the reply address is OTHER-SU-BBOARDS@SAIL, and
guess what? The message goes to all other BBOARDs but SAIL's. The reason
the address is OTHER-SU-BBOARDS@SAIL, is that SAIL users read BBOARD and
make replies with the editor and put the message on the SAIL BBOARD
directly. Therefore, you should change the return address from
OTHER-SU-BBOARDS@SAIL to SU-BBOARDS@SAIL. I received three messages in
the last two days that didn't get on the SAIL BBOARD apparently for that
reason.
∂02-Jul-85 2115 JMC ps on the letter to Roth
To: RA
The "Workshop on coupling symbolic and numerical computing in expert systems"
will be held at
Boeing Computer Services AI Center
Bellevue, Washington 98008
Aug 27-29 1985
The chairman is Janusz Kowalik,Boeing Computer Services,
Advanced Technology Applications Division
M/S 7A-03- P.O. Box 24346
Seattle, Washington 98124
(206) 763-5392
∂03-Jul-85 2241 JMC interview by Slater
To: minsky%oz@MIT-MC.ARPA
Robert Slater, who claims to have interviewed you for a Prentice-Hall
book wants to interview me. Does he seem to be a reasonable person?
∂03-Jul-85 2350 JMC
To: RA
newrep.4 is another to the New Republic; they printed newrep.3.
∂04-Jul-85 0915 JMC
To: RA
Please check whether the men who called me really work for KQED.
∂04-Jul-85 1721 JMC Thank you
To: hobbs@SRI-AI.ARPA, bmoore@SRI-AI.ARPA
Thanks very much for the book and for the kind dedication.
∂04-Jul-85 1727 JMC
To: SJG
a255 1610 04 Jul 85
AM-Young Graduate,0192
13-Year-Old Gets Oxford Math Degree
OXFORD, England (AP) - A 13-year-old girl was awarded a mathematics
degree at Oxford University on Thursday after two years of study.
Before that, she had never even gone to school.
Ruth Lawrence learned the good news when exam results were posted at
the 800-year-old university in central England. She was one of two
math students who received special commendation from the examiners.
Ruth finished college in two years instead of the normal three, but
she cannot pick up her degree until next year because the university
has a three-year residency requirement.
She said she would be staying in Oxford for at least another three
years to do research toward a doctorate.
In an interview with the British Broadcasting Corp., Ruth said: ''I
don't think I am a genius. I think it's got out of a lot of hard
work. It's not sort of born in you. I think you just have to work
hard and have good teachers.''
Before college, her father, Harry Lawrence, taught Ruth at home.
Her tutor at Oxford, Mary Lunn, told the BBC that Ruth was now a
better mathematician than she.
AP-NY-07-04-85 1909EDT
***************
∂05-Jul-85 0948 JMC reply to message
To: munnari!mulga.oz!jwl@SEISMO.CSS.GOV
[In reply to message sent Fri, 5 Jul 85 18:54:01 EST.]
I want to co-ordinate with Nils Nilsson who returns Monday. I'll let you
know Monday or phone UBC. Will you be at the Computer Science Department
there?
∂05-Jul-85 1058 JMC cost of trip
To: HST
The business class round trip to Frankfurt is $2888. That plus hotel
at the meeting is all I anticipate.
∂07-Jul-85 2350 JMC re: fundamentalist Christians reveal anti-Semitic bias (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - While I have not listened to fundamentalist radio programs, I would
suppose that they aren't all of one mind. In any case there is so much
inconsistency in both the Old and New Testaments internally and with
each other that it is unnecessarily worrisome to look for the most gloomy
interpretations even if they are supported by some individual radio program.
∂08-Jul-85 2152 JMC
To: pat@IMSSS
Reminding you to send me your papers on quantum mechanics.
∂09-Jul-85 2158 JMC
To: FY
10 hours
∂09-Jul-85 2302 JMC re: Message from Jerry Pournelle
To: crash!bblue@NOSC.ARPA
[In reply to message sent Tue, 9 Jul 85 21:23:47 PDT.]
Thanks for the message from Jerry.
∂09-Jul-85 2345 JMC
To: TK@SU-AI.ARPA
What's this I hear about Pournelle being zapped at MC for favoring SID?
∂09-Jul-85 2350 JMC re: Reviews of "Not in Our Genes."
To: LIBRARY@SU-SCORE.ARPA
[In reply to message sent Tue 9 Jul 85 23:43:08-PDT.]
Many thanks for the references to reviews. I would be interested in further
non-Marxist reviews, and thanks for continuing to look for Marxist reviews.
Incidentally, I saw a review in Commentary, but perhaps that was too recent.
∂10-Jul-85 0026 JMC reply to message
To: PGS@MIT-MC.ARPA
[In reply to message sent Wed, 10 Jul 85 03:17:27 EDT.]
OK, thanks for that assurance. What if he were to agree to be good?
∂11-Jul-85 0054 JMC report
To: RA
Please send "First order theories of individual concepts and propositions"
and CBCL to Professor David Kaplan
Department of Philosophy
UCLA
∂11-Jul-85 1225 JMC
To: ME
There are long disk queues at low loads, suggesting disk trouble.
∂11-Jul-85 1231 JMC
To: RA
Did Stanford think there was conflict in MCC proposed contract?
∂11-Jul-85 1251 JMC
To: RA
rabbat.1, his letter in my out basket
∂11-Jul-85 1355 JMC
To: RA
eindor.1
∂11-Jul-85 2134 JMC
To: ME
Would it be much trouble to give AP the whole of the HOT screen?
∂12-Jul-85 1351 JMC review of your paper by Minc
To: JK
CC: GLB
Have you seen Minc's review (CR, May 1985) of your
"EKL - A mathemtically oriented proof checker"?
He awaits the EKL version of Landau's Grundlagen.
∂12-Jul-85 1807 JMC
To: RA
Note to Roth at U. Alaska. "I received the papers".
∂14-Jul-85 1309 JMC countering Parnas
To: llw@S1-A.ARPA
I see that the computer issue in connection with SDI is heating up with
the Parnas resignation reported in the New York Times on Friday.
If it will be useful I am willing to tell any suitable forum that
there is no law of computer science asserting that even very complicated
programs can't be debugged and that formal mathematical methods can
help. The main requirement enough smart enough people being willing
to work on it.
∂14-Jul-85 1310 JMC P.S.
To: llw@S1-A.ARPA
If it is available, I am willing to examine the 17 pages of memoranda
accompanying the Parnas resignations and comment on them.
∂14-Jul-85 1903 JMC re: visit by senator glenn
To: KAHN@USC-ISI.ARPA
[In reply to message sent 14 Jul 1985 21:58-EDT.]
Thanks for the notice. I'll keep the day open and await further information.
∂14-Jul-85 2114 JMC 3rd thought
To: llw@S1-A.ARPA
CC: CLT@SU-AI.ARPA, JK@SU-AI.ARPA
Some of us here, interested in program verification, are willing to go
further than merely asserting that there is no law of computer science
preventing the verification of SDI programs. Namely, we are willing
to give some thought to how large programs which interact with the
outside world, and SDI programs in particular, may be verified.
Everybody here is heavily committed already, but this seems important.
If you think it's worthwhile, I'll talk to more people, concentrating
solely on people who are already favorable to the SDI enterprise and
who are interested in the verification of large programs. Otherwise,
we might find only illustrations of the proverb "Where there's a will
to fail, obstacles can be found".
∂14-Jul-85 2116 JMC p.p.s.
To: llw@S1-A.ARPA
In case it needs to be tied in, I received today a message from Bob
Kahn saying that he was bringing Senator Glenn August 27. He thought
the senator was also visiting Livermore at that time.
∂15-Jul-85 1708 JMC re: Doris Hyde, Inference
To: RA
[In reply to message rcvd 15-Jul-85 17:03-PT.]
I mailed some papers to Doris Hyde of Information International on Friday
or Saturday.
∂16-Jul-85 1422 JMC re: NY Times article
To: RA
[In reply to message rcvd 16-Jul-85 14:16-PT.]
It turns out I need the article in a file. Would you please type it
into the file parnas.nyt[e85,jmc] and with the entry in files[let, jmc]
with the comment
Computer scientist thinks SDI programs impossible.
Yes on the interview.
∂16-Jul-85 1423 JMC
To: RA
steen.1[let,jmc]
∂16-Jul-85 1443 JMC re: J.W. Lloyd
To: RA
[In reply to message rcvd 16-Jul-85 14:40-PT.]
Thanks for tracking down Lloyd. Please go ahead. There may be an abstract
in my message file or in my previous messag file s85.in[let,jmc]. Please
announce his talk to bboard, faculty and on bulletin boards.
∂16-Jul-85 1545 JMC
To: RA@SU-AI.ARPA, nilsson@SU-SCORE.ARPA, yao@SU-SCORE.ARPA
A Basis for Deductive Database Systems
J.W. Lloyd
Department of Computer Science
University of Melbourne
This seminar is concerned with a theoretical basis for deductive database
systems. A deductive database consists of closed typed first order logic
formulas of the form A<-W, where A is an atom and W is a typed first
order formula. A typed first order formula can
be used as a query and a closed typed first order formula can be used
as an integrity constraint. Functions are allowed to appear in formulas.
Such a deductive database system can be implemented using a PROLOG system.
The main results are concerned with the soundness,
completeness and non-floundering of the query evaluation process,
the soundness of the implementation of integrity constraints,
and a simplification theorem for implementing integrity constraints.
An alternative query evaluation process will be presented.
Some remarks comparing this approach with the standard relational database
approach will also be made.
∂16-Jul-85 1554 JMC send paper
To: RA
Mr. Jim Bartimo
Personal Computing
1625 The Alameda, Suite 600
San Jose, CA 95126
a reprint of my paper on the history of LISP
Some Expert Systems Need Common Sense
∂16-Jul-85 1635 JMC re: Lloyd's talk
To: RA
[In reply to message rcvd 16-Jul-85 16:21-PT.]
Yes, of course.
∂17-Jul-85 0022 JMC re: Austin trip
To: RA
[In reply to message rcvd 15-Jul-85 16:17-PT.]
No car and MCC will make hotel reservation.
∂17-Jul-85 0030 JMC
To: llw@S1-A.ARPA, JK@SU-AI.ARPA
NY TIMES Friday, July 12, 1985
By Charles Mohr
SCIENTIST QUITS ANTIMISSILE PANEL, SAYING TASK IS IMPOSSIBLE
A computer scientist has resigned from an advisory panel on antimissle defense,
asserting that it will never be possibble to program a vast complex of battle
management computers reliably or to assume they will work when confronted
with a salvo of nuclear missiles.
The scientist, David L. Parnas, a professor at the University of Victoria in
Victoria, British Columbia, who is a a consultant to the Office of Naval Research
in Washington, was one of nine scientists asked by the Strategic Defense Initiative
Office to serve at $1,000 a day on the ``panel on computing in support of battle
management.''
Professor Parnas, an American citizen with secret military clearances, said in
a letter of resignation June 28 and in 17 pages of accompanying memorandums
that it would never be possible to test realistically the large array of computers
that would link and control a system of sensors, antimissle weapons, guidance and
aiming devices, and battle management stations.
Nor, he protested, would it be possible to follow orthodox computer program-
writing practices in which errors and ``bugs'' are detected and eliminated in
prolonged everyday use.
`Money spent will Be Wasted'
----------------------------
``Because of the extreme demands on the system and our inability to test it,
we wil never be able to believe, with any confidence, that we have succeded,''
he wrote. ``Most of the money spent will be wasted.''
Comdr. James H. Offutt of the Navy, an official with the strategic defense office,
said in an interview Wednesday that it was ``probably true'' that what he called
current techniques of writing computer programs would not be reliable, but he
added that Professor Parnas and other had been asked to explore new methods to
``unlock'' workable procedures.
``That is what S.D.I. is all about -- it is a reserach program'' said Commander
Offutt, who is deputy director for battle management, command, control and
communications. He said the goal was a ``realistic testing program'' and said he
was convinced that this could be done through computer simulation of possible
events.
In his letter to Commander Offutt, Professor Parnas took note of President Reagan's
1983 request to the scientists to work toward making nuclear weapons obsolete and
impotent.
`No Technological Magic'
------------------------
``I believe,'' Professor Parnas said, ``that is is our duty, as scientists and
engineers, to reply that we have no technological magic that will accomplish
that. The President and the public should know that.''
Professor Parnas, who worked full time at the Office of Naval Research from
1980 to 1982 and helped develop the battle computer program for the A-7
fighter-bomber and other Navy systems, said that his objections were not
ideological and that he had no ``objections to weapons development in general.''
The professor, who grew up in New York City and previously taught at the University
of North Carolina, remains a consultant to the Navy. In a telephone interview
Tuesday, he said he reached his conclusions about the program, widely known as a
``Star Wars'' plan, after attending the first meeting of the experts panel and
after pondering the scientific dimensions of the antimissile problem.
The strategic defense office is hoping to spend $33 billion over six years for
research and testing on components of an antimissile defense, although Congress
cut $400 million from the request last year and is likely to cut at least $700
million this year. Work has only begun on a design of defense.
Computers and the programs, or software, that control them are a vital element
in any defense. They would be used to help space-based sensors detect and count
a hostile missile launching, sort decoys from warheads, calculate trajectories
and track targets, aim directed energy or high-speed projectile weapons, assess
hits and misses and re-aim weapons and for other tasks.
In the interview, Professor Parnas said he thought it was impossible to believe
that the programs would be free of error or that they could meet the short ``hard''
deadlines inherent in a missile attack that might last only 30 minutes.
``The worst thing is that we wouldn't trust the system if we did build it,''
he remarked.
In his memorandums, the professor put forth detailed explanations of his doubts.
He argued that large-scale programs like that envisioned for the program
require modifications based on realistic use, saying, ``It is only through such
modifications that software becomes reliable.'' He contended this would never
be possible with the strategic missile defense.
He wrote at length about the possibilities, often discussed, that workable
programs could be written by computers themselves, or that ``artificial
intelligence'' could solve the problem or that programs could be adequately
verified through mathematical simulation. In each case, he said he believed that
experience with such methods showed they were unrealistic as a solution.
Some other scientists have recently expressed public doubts that large-scale
programs free of fatal flaws can be written. Herbert Lin, a research fellow at
the Massachusetts Institute of Technology, said this month that the basic lesson
was that ``no program works the first time.''
Professor Parnas wrote to Commander Offutt, ``I am quite certain that you will
be able to find software experts who disagree with my conclusions.'' He said that
for many the antimissile program was seen as a ``pot of gold just waiting to
be tapped'' for research projects and also an exciting puzzzle that was ``fun
to work on,'' but that he had concluded the plan was not feasible.
∂17-Jul-85 1113 JMC parallel computers
To: LES
∂17-Jul-85 1023 fateman%ucbdali@Berkeley parallel computers
Received: from UCB-VAX.ARPA by SU-AI.ARPA with TCP; 17 Jul 85 10:23:09 PDT
Received: from ucbdali.ARPA by UCB-VAX.ARPA (4.24/5.2)
id AA02456; Wed, 17 Jul 85 10:18:50 pdt
Received: by ucbdali.ARPA (4.24/4.48)
id AA29674; Wed, 17 Jul 85 10:23:59 pdt
Date: Wed, 17 Jul 85 10:23:59 pdt
From: fateman%ucbdali@Berkeley (Richard Fateman)
Message-Id: <8507171723.AA29674@ucbdali.ARPA>
To: clt@su-ai, jmc@su-ai, rpg@su-ai
Subject: parallel computers
Encore seemed to be able to demonstrate something at NCC.
∂17-Jul-85 1817 JMC re: Pournelle
To: tk@MIT-MC.ARPA
[In reply to message sent Tue, 16 Jul 85 16:02 EDT.]
My writing SID was a misprint for SDI - standing for Star Wars.
He thought he was zapped for favoring that, so I see he still didn't
understand the situation.
∂17-Jul-85 1905 JMC re: using english good (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - It seems to me that there are several kinds of misuse of language.
1. semi<time unit> means "twice a". by<time unit> means "every other"
as in "semiweekly" and "biweekly". Some people started using "biweekly"
to mean "semiweekly", and it caught on to the extent that Webster's
Third International Dictionary recognizes "bi" as a synonym for "semi".
These people, eventually abetted by Webster, have made the English language
worse than it was. The word "biweekly" has become ambiguous, so if
one wants to be sure of being understood, one has to use an additional
explanation.
2. Misspelling. Using "occurance" for "occurrence" looks ugly to many
people. Of course, to make something that looks ugly to certain others
is an advantage to some people. There is the further fact
that standardized helps inexperienced people, e.g. children and
foreigners, look up unfamiliar words in dictionaries. There are
also many spelling errors that result in a different word. However,
one needn't be too snooty about spelling, because accurate spelling
is much harder for some people than it is for others.
3. There is a myth among linguists that language cannot be improved
by authorities prescribing improved usage. Such improvements have
often occurred in English, starting, I believe, with Samuel Johnson.
4. I think it is somewhat harmful when companies make themselves
brand names by taking an ordinary word and misspelling it. Besides,
the results are usually ugly.
5. Finally, beauty and elegance are legitimate considerations - at
least for those who prefer them. We could start a campaign
to be spared bad English just as self-righteous as the present
campaign to be spared cigarette smoke.
∂18-Jul-85 0920 JMC re: gwai85
To: HST
[In reply to message rcvd 18-Jul-85 04:09-PT.]
I'm sorry to be so vague up to now. The problem is that the meeting
is right around the start of classes, so the default would be that
I would hurry right back. I have about decided to get someone,
most likely Carolyn, to begin my class for me so I can stay a week
in Germany, but I haven't talked about it yet.
As for the talk itself, it could be either "What is common sense?"
or "Non-monotonic reasoning with applications to formalizing common
sense". The latter would be more mathematical. Which do you think
the audience would prefer.
Incidentally, the trip will probably cost quite a lot less than I
said, because I'll combine it with a trip to Paris and Marseilles,
but I can't be quite certain of that yet.
∂19-Jul-85 0038 JMC
To: JK
There have recently been claims by various scientists that
the computer programs that would be required to realize the
Strategic Defense Initiative (SDI) are impossible to make reliable,
and therefore the project must be abandoned. These claims have
been made both by a few physicists and by a few computer scientists.
As far as we can see, these scientists are speaking beyond their
competence in order to lend support to the anti-SDI cause.
The actual requirements for SDI computer systems are completely
undetermined, because they depend on what physical devices are to be
used for detecting missiles, distinguishing them from whatever decoys
may be used, and whatever physical effects may be used to destroy
they missiles. Since these matters are still in the research stage,
the computer requirements are completely unknown.
Therefore, those opposed to the SDI are raising arguments
of a general character. Someone has come up with the idea that
programs of 10 million instructions will be required. As long as
the requirements are unknown, that number is as good or bad as
any other. There are only a few programs that big in operation.
The argument continues by asserting that the only way to
make the program reliable is to have a nuclear war in order to
test all parts of it. This is a mistake.
First of all, there is no principle of computer science
that limits what size programs can be made reliable. Making a
big program reliable that cannot be repeatedly tested in a
all situations is difficult, but it has been done, and techniques
exist and more can be developed. Many present programs have parts
designed to deal with emergencies that are too dangerous or expensive to
allow to happen, and many techniques are available to make them
reliable.
These techniques include the following. First, simulated
environments can be created that are even more demanding than
the real world is expected to be. Second, mathematical techniques
exist for proving parts of a program correct and checking these
proofs by computer.
If the SDI progams prove to be very demanding, and this
is quite possible, there is time to develop the necessary program
verification techniques in parallel with the development of the
SDI hardware. As with other parts of the SDI project, the
computing part will require dedicated work by competent scientists.
We are sure that the computer scientists who work on this
vital project will do their share.
Those who oppose this effort to end Mutually Assured
Destruction have no basis in computer science for predicting
its failure. They must rely on the general principle, ``Where
there is a will to fail, obstacles can always be found''.
∂19-Jul-85 1349 JMC re: reaching Sarah Macarthy
To: RA
[In reply to message rcvd 19-Jul-85 11:20-PT.]
(9)968-5792) and leave message.
By the way, that's Gene Golub.
∂19-Jul-85 1353 JMC re: Dinner with Gordon Bell
To: RA
[In reply to message rcvd 19-Jul-85 10:40-PT.]
Please tell Gordon Bell that I'm tied up tonight, but tomorrow night
would be fine. I'll call him tonight or he can call me at 857-0672
between 6:30pm and 8pm or between 10pm and midnight.
∂19-Jul-85 1536 JMC
To: RA
Please prepare an invoice for MAD for 4 hours consulting today.
∂20-Jul-85 0959 Mailer failed mail returned
To: JMC
In processing the following command:
MAIL
The command was aborted because these Host Name(s) are Unknown:
RUSSIAN.SPA.SYMBOLICS.COM
------- Begin undelivered message: -------
∂20-Jul-85 0959 JMC re: nusum
[In reply to message sent Sat, 20 Jul 85 02:27 PDT.]
This is mostly a test to see if replying works with this address. However,
HST is, so far as I know, in Germany now.
------- End undelivered message -------
∂22-Jul-85 0909 JMC re: J.W.Lloyd
To: RA
[In reply to message rcvd 22-Jul-85 09:06-PT.]
I have to take my car to the garage, but I'll be in by 10.
∂22-Jul-85 0957 JMC benevolence of Ho Chi Minh
To: mrc@SU-SCORE.ARPA
See the article "Reeducation Kills" in Freedom at Issue in the lounge.
∂23-Jul-85 0908 JMC
To: VAL
July 31 ok for Przymusinski seminar.
∂24-Jul-85 0808 JMC
To: sandra@SRI-KL.ARPA
OK, please phone early Thursday afternoon.
∂24-Jul-85 1202 JMC reservations to L.A. on Friday
To: RA
Please make reservations to leave San Jose as soon as possible after
8am and to return after 9pm on PSA.
∂24-Jul-85 1431 JMC
To: RA
No car needed. Inference is 5 min from airport.
∂24-Jul-85 2343 JMC
To: CLT
Lloyd's book is 1984.
∂25-Jul-85 0936 JMC
To: fowler@SU-SUSHI.ARPA
About 250 yen to the dollar.
∂25-Jul-85 1212 JMC re: NSF 810 4877 final project report
To: RA
[In reply to message rcvd 25-Jul-85 11:35-PT.]
Please find out what that number stands for. They always cite these
meaningless numbers.
∂25-Jul-85 1254 JMC re: Robert Slater
To: RA
[In reply to message rcvd 25-Jul-85 12:45-PT.]
I called him.
∂25-Jul-85 1507 JMC re: Common knowledge of common knowledge
To: barwise.pa@XEROX.ARPA
[In reply to message sent 25 Jul 85 13:41 PDT.]
So far as I know the point is original with me. However, I didn't publish
it. Moreover, my reading, even of Hintikka's work, is sufficiently sketchy
that I wouldn't be confident that no-one else published it. I have talked
about it since 1975. My earlier formalization of knowledge involving
0 or "any fool" used a common knower, but 0 knows what everyone knows.
This dates from the 1960s. Most of what I did isn't published, but I'll
send you one AI memo based on improvements by Sato et al. of matters I
lectured on in Kyoto in 1975.
∂25-Jul-85 1557 JMC NSF report
To: RA
BASIC.REP[e85,jmc] contains the body of the report. Please put
a heading on it identifying it with the NSF numbers, add the
bibliography including all my publications of the time period
and TEX it.
∂25-Jul-85 1830 JMC Split atoms, not wood.
To: su-bboards@SU-AI.ARPA
a233 1515 25 Jul 85
AM-EPA-Wood Stoves, Bjt,0595
Wood Stoves To Come Under New Federal Regulation
By MATT YANCEY
Associated Press Writer
WASHINGTON (AP) - The government intends to place household wood
stoves, an ancient source of heat revived by high energy prices,
under new regulations because of the millions of tons of
health-damaging pollutants they pour into the air.
The Environmental Protection Agency, blaming the stoves for the
failure of several areas of the nation to meet federal air quality
limits, said Thursday it plans to develop pollution-reducing
performance standards for the units that could be issued within 18
months.
The new standards would effectively require any wood-burning stoves
and fireplace inserts sold after a yet-to-be-established date to
include catalytic converters similar to those on automobiles.
The converters, costing $200 to $300 each, can effectively reduce
the pollution from wood burners by 50 to 80 percent by recycling and
reburning the smoke the stoves produce.
''We estimate that residential wood stoves are exposing large
numbers of people to pollutants which are harmful to human health,''
acting Assistant EPA Administrator Charles Elkins said Thursday in
announcing the proposal.
Without controls, Elkins said, the stoves annually could be pouring
7 million tons of soot, dirt and other particulate matters, 19
million tons of carbon monoxide, 159,000 tons of hydrocarbons and
52,000 tons of cancer-causing polycyclic organic compounds into the
atmosphere by the year 2005.
Some experts have estimated that the polycyclic compounds and other
particulate products of incomplete combustion are causing 800 cancer
cases a year nationwide. EPA officials blame wood burning stoves for
half of that pollution.
''Wood smoke is one of the largest unregulated sources of pollution;
it's killing a lot of people,'' said David Doniger, an attorney for
the Natural Resources Defense Council, an environmental group that
has been pressuring the EPA to regulate the stoves.
In addition, the non-carcinogenic pollutants produced by the stoves
can settle in the lungs and cause or aggravate several respiratory
ailments.
Establishing the performance standards is the EPA's first step in
drafting regulations governing the manufacture of wood stoves. David
Cohen, an EPA spokesman, said the agency hopes to issue proposed
regulations by Jan. 1, 1987.
If that ''accelerated'' schedule is kept, new stoves not equipped
with the pollution-reducing converters or that otherwise fail to meet
new pollution limits could be banned from the market as early as two
years from now.
Some states and a handful of communities in areas where wood burners
have become particularly popular - the Northeast, Rocky Mountains and
Northwest - already have moved to reduce the often-visible and smelly
accumulations of smoke that the stoves can produce.
Oregon already has established state limits on the pollution that
new residential wood-burning stoves can emit. Effective next June,
that standard is 6 grams of particulate matter per hour for
catalyst-equipped stoves and 15 grams per hour for those not so
equipped.
State officials in Oregon estimate its standards will cut pollution
from the stoves by 50 percent. Even tighter Oregon standards to be
implimented in 1988 are expected to yeild an 80 percent reduction in
pollution compared with conventional wood-burning stoves.
Communities such as Butte, Mont., and Aspen, Colo., have passed
local ordinances limiting use of wood stoves during inversions and
other weather conditions that tend to trap pollution in place.
Worried that more states and local governments will establish varied
standards of their own, manufacturers and distributors of the stoves
have endorsed the adoption of a nationwide federal regulations by the
EPA.
AP-NY-07-25-85 1815EDT
***************
∂26-Jul-85 0731 JMC
To: RA
Please send Doug Lenat at MCC "Some expert systems need common sense".
∂28-Jul-85 0328 JMC send reports
To: RA
Please send to Paul Haley at Inference
1. Gabriel and McCarthy paper
2. the map coloring paper and the Szeredi letter
∂28-Jul-85 1801 JMC The star wars computer system.
To: gnelson@DECWRL.ARPA
Are you the co-author of a report of that title claiming it can't be
done? If so please send me a copy ASAP. I am entering the lists on the
opposite side.
∂28-Jul-85 2131 JMC
To: CLT
p.29 computationly → computationally
∂28-Jul-85 2133 JMC abstract syntax, p.31
To: CLT
A sentence explaining that abstract syntax is defined by the names
and relations of the selectors and constructors and is independent
of the notation used would make the idea clear. It has become
a folk concept and is often used without explanation and without
reference to the explanation in my 1963 paper.
∂28-Jul-85 2242 JMC
To: CLT
Ce sera Mercredi a sept heures et Sarah assistera aussi.
∂28-Jul-85 2254 JMC
To: CLT
∂28-Jul-85 2252 CLT
what do you mean Sarah will assist too?
"assister" means "be present". It doesn't mean "assist".
∂28-Jul-85 2332 JMC goals of theory of computation
To: CLT
It has been one of my main goals, and I would be surprised if
I didn't mention it in the paper you cite, to provide a means of
getting practical computer checkable proofs that programs meet
their specifications. Did I miss it, or didn't you mention this
goal at all? There is also the specification of the correctness
of compilers which requires formal specification of the semantics
of programming languages.
∂29-Jul-85 0122 JMC
To: RA
grauba.1
∂29-Jul-85 1408 JMC re: LA trip
To: RA
[In reply to message rcvd 29-Jul-85 13:27-PT.]
Thanks for the reservations.
∂29-Jul-85 1409 JMC re: The star wars computer system.
To: gnelson@DECWRL.ARPA
[In reply to message sent 29 Jul 1985 1116-PDT.]
Thanks for the reports. Is Dave David Parnas or someone else?
∂30-Jul-85 2124 JMC re: circ'n seminar
To: VAL
[In reply to message rcvd 30-Jul-85 14:04-PT.]
Sure, invite them.
∂30-Jul-85 2125 JMC re: MCC
To: RA
[In reply to message rcvd 30-Jul-85 15:13-PT.]
Right. To Bledsoe.
∂30-Jul-85 2136 JMC a great loss
To: su-bboards@SU-AI.ARPA, CLT@SU-AI.ARPA, sf@SU-CSLI.ARPA,
barwise@SU-CSLI.ARPA
a277 2033 30 Jul 85
AM-Deaths,0216
By The Associated Press
Julia Robinson
BERKELEY, Calif. (AP) - Julia B. Robinson, 86, a mathematics pioneer
who taught at the University of California in Berkeley, died Tuesday
of leukemia.
Mrs. Robinson's research was aimed at solving logic problems with
number theories. In 1976, she became the first woman elected to the
National Academy in Science and served as the first woman president
of the American Mathematical Society. Mrs. Robinson conducted
research and occasionally lectured in the UC Berkeley math
department. She became a professor there in 1976 and retired last
month.
[She certainly wasn't 86; more likely 66].
∂30-Jul-85 2324 JMC re: abortion, religion, etc (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - Men can bare children easily. I've done it many times. It's
especially easy with very young ones now that something like Velcro
is used instead of safety pins on diapers. Just lay the baby down on
a flat surface and remove first the shirt and then the diaper.
It is recommended that all children be bared before giving them
baths, but older children can bare themselves.
It is mere male chauvinism (or possibly female chauvinism) to say
that males can't bare children.
∂30-Jul-85 2340 JMC re: abortion, religion, etc (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - Apropos of men baring children, I should add that there are
numerous restrictions of law, custom and morals. They
depend on whether the child is yours, its age, its sex, whether
it objects, and the purpose for which you are baring it. If uncertain
consult your lawyer, your mother, a sociologist or the church of your
choice.
∂31-Jul-85 1804 JMC
To: llw@S1-A.ARPA
The letter got delayed while Bob Jastrow got some facts about the
AT&T program, but here's the current draft. He is consulting
Abrahamson's office about where to send it. Maybe tomorrow, I hope.
@make(letter)
@style(indent 0)
@begin(center)
John McCarthy
846 Lathrop Drive
Stanford, CA 94305
@end(center)
@tabclear
@tabdivide(3)
@b[Telephone (415) 857-0672@\@>July 31, 1985]
@begin(address)
Mr. .....
@end(address)
@begin (body)
@greeting(Dear Sir:)
A number of scientists have recently claimed that the computer
programs needed to realize the Strategic Defense Initiative (SDI) are impossible
to make reliable, and therefore the project must be abandoned. As far as I can
see, these scientists are speaking beyond their competence in order to lend
support to the anti-SDI cause.
The computing arguments they have raised are of a general
character. The Fletcher report on ballistic missiles defense says a program with
6 to 10 million instructions will be required.* This would indeed be
a very large program. But this doesn't justify the assertion
by these scientists that the program cannot be written, and if
written, could not be verified. Even larger programs have been
written and put into immediate successful use.
For example, the program put into use by AT&T when the nation changed over to
electronic switching has 50 million lines of code. Furthermore, it has
14,000 interconnections between nodes, compared to an estimated 4500
interconnections in the SDI program.
In other words, the telephone program is bigger and in many respects more
complex than SDI requirements. Yet the program worked well from the
moment it was put into operation.
The argument against SDI continues by asserting that the only way to make the
program reliable is to have a nuclear war in order to test all parts of it. This
is a mistake. Many present programs are designed to deal with emergencies too
dangerous or expensive to allow them to happen for test purposes. Yet techniques
are available to make these programs reliable. The mammoth AT&T program worked
when it was turned on although it had never been tested in "battle".
These techniques for checking large programs include the following.
First mathematical techniques exist for proving parts of a program
correct and checking those proofs by computer.
Second,
simulated battle conditions can be created that are even more complex and more
stressing than a real attack.
*Report of the Study on Eliminating the Threat Posed by Nuclear Ballistic Missiles,
Volume 5 (Battle Management, Communications and Data Processing), p. 43.
@newpage
My conclusion, based on 35 years of research in computing and artificial
intelligence, is that
those who oppose Strategic Defense or "Star Wars" have no basis in computer science
for predicting its failure. They must rely on the general principle, "Where
there is a will to fail, obstacles can always be found."
Like the other scientific and engineering problems associated with SDI,
the computer problems are likely to be challenging.
@end(body)
Sincerely
John McCarthy
Professor of Computer Science
Stanford University
[John McCarthy is one of the leaders of American computer science. He was
one of the originators of artificial intelligence research, developed
the programming language LISP most used in AI, originated the concept
of computer time-sharing that is the basis of present computer operating
systems, and originated some of the main methods of mathematically proving
that computer programs meet their specifications. He found the Stanford
Artificial Intelligenc Laboratory and has served as President of the
American Association for Artificlal Intelligence.]
∂01-Aug-85 0011 JMC re: Afterthought
To: LLW@S1-A.ARPA
[In reply to message sent 01 Aug 85 0009 PDT.]
I presently intend to use both office address and phone number, but thanks
for the kind words and the thought.
∂03-Aug-85 0001 JMC re: [The Mailer Daemon <Mailer@SUMEX-AIM.ARPA>: Message of 2-Aug-85 16:11:43]
To: AAAI-OFFICE@SUMEX-AIM.ARPA
[In reply to message sent Fri 2 Aug 85 16:49:19-PDT.]
Your MAIL to JMC-LIST@SU-AI failed, because it should be JMC-LISTS@SU-AI.
∂04-Aug-85 2210 JMC reply to message
To: ucscd.ucscc!ucscc!beeson@UCB-VAX.ARPA
[In reply to message sent Sun, 4 Aug 85 12:43:50 pdt.]
Except for a few typos, I'll let what I said stand if the others let theirs
stand. "Vietnam" is one word. Use the American spelling "behavior"
consistently. Nixon's Science Adviser was Dr. Edward David. I first
wrote about the electronic library in the late 1960s. You have "The
best way way to make progress is to forbid suffering". I said "
forbid something". There are more typos.
∂04-Aug-85 2212 JMC re: gio presentation
To: KAHN@USC-ISI.ARPA
[In reply to message sent 4 Aug 1985 21:41-EDT.]
20 minutes will do it.
∂04-Aug-85 2215 JMC re: Renaming the machine?
To: JJW, LES, CLT, RPG, RTC
[In reply to message from JJW rcvd 04-Aug-85 20:01-PT.]
clumsy, cloddish or catatonic of which I prefer catatonic given
the typical response times.
∂04-Aug-85 2221 Mailer failed mail returned
To: JMC@SU-AI.ARPA
In processing the following command:
MAIL
The command was aborted because these Host Name(s) are Unknown:
XEROX.PA
------- Begin undelivered message: -------
∂04-Aug-85 2221 JMC reply to message
To: ucscd.ucscc!ucscc!beeson@UCB-VAX.ARPA, TW@SU-AI.ARPA
[In reply to message from ucscd.ucscc!ucscc!beeson@Berkeley sent Sun, 4 Aug 85 12:43:50 pdt.]
Except for a few typos, I'll let what I said stand if the others let theirs
stand. "Vietnam" is one word. Use the American spelling "behavior"
consistently. Nixon's Science Adviser was Dr. Edward David. I first
wrote about the electronic library in the late 1960s. You have "The
best way way to make progress is to forbid suffering". I said "
forbid something". There are more typos.
------- End undelivered message -------
∂05-Aug-85 1357 JMC Graubard letter
To: RA
I have redone grauba.1. Please put the decorations in that file
and print it.
∂05-Aug-85 1446 JMC PROLOG Digest V3 #16
To: subramanian@SUMEX-AIM.ARPA
∂08-Apr-85 0048 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #16
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 8 Apr 85 00:47:47 PST
Date: Sunday, April 7, 1985 7:51PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #16
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Monday, 8 Apr 1985 Volume 3 : Issue 16
Today's Topics:
Puzzles - Maps & C-Prolog
Implementations - Bugs & Cases & Cuts & Determinism,
& Snips & Denotational Semantics
----------------------------------------------------------------------
Date: Wed, 27 Mar 85 11:04:31 -0200
From: Ehud Shapiro <Udi%Wisdom.bitnet@WISCVM.ARPA>
Subject: Not the semantics of Concurrent Prolog
I know, I know I should be doing semantics of Concurrent
Prolog, but instead I wrote a Prolog program for map
colouring. It seems that the program doesn't do any work.
The magic is, as usual, in the combination of unification
and non-determinism.
The program defines the relation colour(Map,Colours),
between a map and a list of colours, which is true if Map
is legally coloured using Colours.
A map is represented using adjacency-lists, where with
each country we associate its colour, and the list of
colours of its neighbours. If the program is used in
generate-mode, i.e. to colour an uncoloured map, these
colours would be uninstantiated variables. The program
would then compute all possible colourings. For example,
the uncoloured map
←←←←←←←←←←
| a |
|←←←←←←←←|
|b |c |d |
|←←|←←|←←|
|e |f |
|←←←|←←←←|
is represented by the term:
map( [country(a,A,[B,C,D]),
country(b,B,[A,C,E]),
country(c,C,[A,B,D,E,F]),
country(d,D,[A,B,F]),
country(e,E,[B,C,F]),
country(f,F,[C,D,E])
]
).
And here is the program:
colour←map([Country|Map],Colours) :-
colour←country(Country,Colours),
colour←map(Map,Colours).
colour←map([],←Colours).
colour←country(country(←Name,C,AdjacentCs),Colours) :-
remove(C,Colours,Colours1),
subset(AdjacentCs,Colours1).
Which uses a couple of utilities:
subset([C|Cs],Colours) :-
remove(C,Colours,←),
subset(Cs,Colours).
subset([],←Colours).
remove(C,[C|Cs],Cs).
remove(C,[C1|Cs],[C1|Cs1]) :-
remove(C,Cs,Cs1).
To test the program, use the following code:
test(Map) :-
map(Map),
colours(Colours),
colour←map(Map,Colours).
colours([red,green,blue,white]).
By the way, the running time of the algorithm is
exponential in the size of the map.
-- Ehud Shapiro
------------------------------
Date: Thu, 28 Mar 85 00:23:12 -0200
From: Ehud Shapiro <Udi%Wisdom.bitnet@WISCVM.ARPA>
Subject: Addendum
p.s. An exercise, for all you program-provers out there:
Prove that colour←map(Map,Colours) terminates succesfully,
if Maprepresents a planar graph, and Colours contains four
distinct colours.
Hint: assume that remove/3 and subset/2 are defined
correctly, just concentrate on proving the first three
axioms...
------------------------------
Date: Tue, 2 Apr 85 19:24:06 pst
From: Allen VanGelder <AVG@Diablo>
Subject: Riddle: How did I do this in C Prolog?
In the following script I did what it looks like I did.
cpl gives me C Prolog top level. write/1 has not been
redefined; the write you see actually wrote "thisIsX(6)";
p.pl has no output statements.
RIDDLE: What kind of rule(s) in p.pl account for this
behavior?
Script started on Tue Apr 2 18:50:19 1985
% cpl
C Prolog version 1.4e.edai
yes
| ?- ['p.pl'].
p.pl consulted 84 bytes 0.016667 sec.
yes
| ?- p(X).
X = 6 ;
no
| ?- p(X), write(thisIsX(X)), X=6.
thisIsX(6)
no
| ?- halt.
% Prolog execution halted
% ↑D
script done on Tue Apr 2 18:53:33 1985
For those not fluent in C Prolog syntax, I
loaded "p.pl", then issued goal "p(X)" and
got the solution "X=6". The semi-colon asks
for more solutions; there were none. Then I
issued the more complex goal, which printed
"thisIsX(6)" and failed.
------------------------------
Date: Thu, 28 Mar 85 15:54:12 pst
From: Newton@CIT-Vax (Mike Newton)
Subject: CProlog bug (revisited)
Hi,
There was an error in my posting on the C-Prolog Problem,
instead of
f(A) :- (true, ! ; B=0), A =< 7.
the test case should have been:
f(A) :- (true, ! ; A=0), A =< 7.
↑------ i goofed here.
Here is a transcript of the session using the distributed
C-Prolog 1.5:
Script started on Thu Mar 28 10:09:24 1985
((cit-vax:1)) pd
/usr/src/cit/cprolog ~
((cit-vax:2)) ls
AIDigest STRIPSTANFORD cpl1x4a fprolog
README coling cpl1x5 stanford
((cit-vax:3)) cd cpl1x5
((cit-vax:4)) ./prolog
C-Prolog version 1.5
| ?- [user].
| f(A) :- (true, ! ; A = 0) , A =< 7.
user consulted 128 bytes 0.0500008 sec.
yes
| ?- f(3).
! Error in arithmetic expression: ! is not a number
no
| ?
[ Prolog execution halted ]
((cit-vax:5)
script done on Thu Mar 28 10:10:16 1985
The problem first surfaced in a grammar rule that was
being used in our compiler. I simplified it to the
above test case... The fix that i posted in the earlier
newsletter has not yet caused any noticeable problems, but..
Regards,
-- Mike
[818-356-6771]
------------------------------
Date: 29 Mar 85 14:05 PST
From: Kahn.pa@XEROX.ARPA
Subject: A Case for Cases
The case for Cases is mostly a mater of taste. I think
a Prolog clause should be meaningful on its own. Yet
this often leads to redundant computation. Consider
the common way of writing a predicate to compute the
maximum of two numbers.
max(X,Y,X) :- X > Y, !.
max(X,Y,Y).
The second clause is pretty bizarre. Using Cases one
can package up the two clauses and get some clearer (and
just as efficient code). I admit that in this example and
those that come to mind If would do as well. (Richard
is right that its hard to find examples where the
non-determinism of the test is useful). Cases has a
neater semantics than "If though".
I sometimes think that max should be written simply
as follows
max(X,Y,X) :- X > Y.
max(X,Y,Y) :- not (X > Y).
and let the compiler compile out the redundant test.
The problem with this is that it can be pretty verbose
if there are many cases since one needs to explictly
write the negation of the previous tests.
------------------------------
Date: Wed, 27 Mar 85 18:13:23 pst
From: Peter Ludemann <Ludemann%UBC.Csnet@csnet-relay.arpa>
Subject: Cuts, Determinism.
I agree that "soft" cuts don't seem to have much use.
I think that tests by their nature shouldn't instantiate
any variables, so backtracking over them won't produce
anything new. However, the semantics of soft cuts are
nicer: it is straightforward to rewrite clauses containing
soft cuts into a form which contains no cuts at all.
As to non-determinism of append. What I meant is this.
Suppose we have a clause:
g(A, B, C) :- append(A, B, C), g2(A, B, C).
Unless we know that at least two of the three variables
are ground, we don't know if the call TO append is
deterministic. For example, if we called "g(X, Y,
[1,2,3])", the call to append is clearly
non-deterministic and we cannot do the last call
optimisation with "g2". So, in the absence of mode
declarations and some moderately complicated compile-time
analysis, we can only tell at run time that a given call
is deterministic. This is easy (and reasonably efficient)
to determine at run-time - if a goal is "done with"
(backtracking over it will only result in failure), it can
be removed from the stack. The last call optimisation is
applicable if the backtrack clause is the parent clause.
Warren's SRI Tech Note 309 talks about goal stacking in
the last few pages and he makes the point that tail
recursion optimization (which I prefer the call last call
optimization) is applicable at every procedure call -
"one simply discards the calling goal if it is later than
the last choice point". I think that my (run-time) method
achieves that.
------------------------------
Date: Wed, 27 Mar 85 22:12:56 pst
From: Prolog@CIT-Vax (Paul Prolog)
Subject: Cuts and Snips
We have an implementation of prolog at Caltech
with a new concept we are calling snips, a concept
conceived by Jim Kajiya. This is somewhat like the
"soft" cut described in the last couple of digests.
Our point is to allow more localized control of
backtracking within a clause. Snips allow a backtrack
frame to be created within the scope of a clause,
but then, unlike a cut, allow only the backtrack
points the user wishes, to be removed. It would work
like this. If one had a clause
a :- b,c,[!,d,e,f,!],g
where the [! marks the beginning of a snip region,
and !] marks the end of a snip region, backtracking
would be normal for the b and c clauses. Backtracking
would continue normally into the snip region, but once
the !] was passed, any backtrack points created by
d, e, or f would be deleted from the backtrack stack.
Then, if g failed, any alternatives of c or b would be
tried. If before the !] were executed, backtracking
took execution to the left of the snip region (left
of [!), nothing special would happen. Cut regions can
be nested as deeply as the user wishes.
Snips are useful when one wants to have backtracking
procedures, but also wants to use the same procedure
in deterministic programs. For instance, in a pattern
matcher I am writing to handle associativity and
commutivity, I need to take a list and take elements
out, while keeping a list of the non-selected
elements. This is written in a clause as
..... ,
append(Begin,[TestElement|End],OriginalList),
append(Begin,End,RestList),
.....
(Of course, there may be a better way to write this,
but I wrote it this way). Now the second append
doesn't ever need to backtrack, since it is not
being used for search. However, having two append
predicates, one with no backtrack point, and one
without, would be silly. A cut may not be appropriate,
since the procedure that this clause fragment is a
part of may need to backtrack. A singlet clause
with a cut could be used, but program comprehension
would be better if all logical parts of a clause
could stick together, as in the example above.
So, rewriting it as
..... ,
append(Begin,[TestElement|End],OriginalList),
[!,append(Begin,End,RestList),!],
.....
would remove the problem.
We feel that snips are a valid addition to
standard prolog. Though there are ways of getting
around the need for them, they are useful in their
own right. (A more detailed article may be coming).
Any comments?
-Keith Hughes
hughes@cit-vax
PS. The other members of our group are Mike Newton
(newton@cit-vax) and Jim Kajiya (kajiya@cit-20).
------------------------------
Date: Thu, 28 Mar 85 21:06:54 pst
From: Paul Voda <Voda%ubc.csnet@csnet-relay.arpa>
Subject: Denot. semantics versus the Symbiosis approach.
This is a note on the relation between the
denotational and "logical" semantics of logic
programming languages. Denotational semantics deals
with the meaning and control at the same time. Thus
for a sequential disjunction the formula
"P(5) or 5 = 5" will denote the undefined element if
the computation of "P(5)" does not terminate.
Viewed as a formula of a predicate calculus, i.e.
ignoring the control and concentrating only on the
meaning, the formula is true. This combination of
the meaning and control in the denotational semantics
is not too bad when the control is just the normal
reduction (lazy evaluation) of the lambda calculus.
As the control gets more complicated (explicit
parallelism, input-output annotations, cuts and commits,
etc.) the meaning functions of the denotational
semantics must necessarily grow more and more complicated
(in the case of parallelism there is even no satisfactory
solution).
In my Symbiosis paper I essentially took the step taken
by Hoare when he separated the partial correctness from
the termination. Divide and impera is always a good
methodological principle but in the case of logic (or
functional) languages the gains are quite significant.
First of all, programs of these languages are formulas resp.
terms of a first order logic theory. The meaning of programs
(corresponding to the partial correctness) is obtained
simply from the standard model of the theory. The reasoning
about the programs can utilize the full deductive power
of first order theories (quantifiers, induction). The
termination is separated from the "logical" meaning.
Now, computations are proofs and in the Symbiosis paper
I show on five different programming languages how to set up
a deductively restricted subtheory of the meaning theory
where the computations exactly correspond to the derivations
in the subtheories. This takes care of the operational
semantics and the problem of termination is reduced to
the proof-theoretical question of existence of a proof.
The questions of existence of proofs are much harder
then the quite straightforward questions of meaning.
But I do not share the gloomy view of Uday
(Prolog Digest 3#15) that we cannot do better than "run-it-
and-see". First of all, we are dealing with deductively
restricted subtheories. Thus it is not too difficult to
arithmetize the predicate of the provability (i.e the
provability in the operational subtheory) in the meaning
theory. One can then proceed and prove a couple of
sufficient conditions for the termination. The conditions
are generally of the form "if .... terminates then
←←←← ....←←← also terminates". These theorems can
be then used to reason about the termination.
----Paul Voda
------------------------------
Date: Tue 2 Apr 85 13:46:32-PST
From: Joseph A. Goguen <GOGUEN@SRI-AI.ARPA>
Subject: logic programming semantics
I have been following the discussions of the semantics
of Prolog and Concurrent Prolog in the Digest with a
rising mixture of fascination and despair. If the
designers of a language have trouble explaining (and
perhaps even understanding) some constructs of their
language, then perhaps there is something wrong with
those constructs. I hasten to add that I find the
read-only annotation idea very appealing, and I hope
that somehow its semantics can be made as simple as I
once thought it might be.
There is a point that I would like to add to the
discussion: Having a denotational semantics for a
language does not make it more respectable. There are
many nice things about denotational semantics, besides
its mathematical elegance and the way it handles
potentially infinite control structures; one of these
is that it serves as a pretty accurate measure of
just how horrible a programming language is; in particular,
you can see how many levels of continuations are needed,
which constructs need the extra complications, how large
the definition is (e.g., how many hundreds or thousands
of semantic equations), etc. Of course, you can give a
denotational semantics for almost anything if you really
want to (e.g., Ada). But I claim that if a language
*needs* a denotational semantics in order to explain its
constructs, then that language is too complicated. The
good thing about a real logic programming language is
exactly that it *does* *not* *need* such a semantics:
a program means exactly what it says. It is a symptom
of the constructs of a language being too operational that
one is driven to denotational semantics to explain them.
So I guess what I'm saying is I am afraid that Concurrent
Prolog is getting us further away from the original
ideals of logic programming, closer to PL/I and its cousins.
In my opinion, what we need is more research on how to do
practical programming with pure logic programming
languages; this may mean enriching the logic, or doing some
things in a metalogic, or whatever. But making Prolog less
and less pure does not seem to me the right way to go.
In case you wonder what I mean by a "logic programming
language", here is a general definition: a language
whose programs consist of sentences in a well understood
(and reasonably simple) logical system, with operational
semanticsgiven by deduction in that logical system. In
particular, "well understood" should include a completeness
theorem for deduction. Notice that both "relational" or
"Horn clause" programming (i.e., what is usually called
"logic programming") and functional programming are special
cases, with logics respectively first order Horn clause
logic, and equational logic (in fact, usually higher order
equational logic).
I would also impose another requirement, that every program
has an initial model. This provides a foundation for
database manipulations, since you know exactly what is true
-- namely, what can be proved from the axioms -- and
everything else is false (this is Occam's famous razor); in
fact, the "closed world" that the program is talking about
is exactly the initial model. (For those not familiar with
this terminology, the initial model is actually
characterized (for the usual logics) uniquely up to
isomorphism by the property that only what is provable is
true; it is closely related to the Herbrand universe.
It has very recently been proven (by Tarlecki, and by Mahr
and Makowsky) that the largest sublanguage of first order
logic such that all sets of sentences have initial models
is Horn clause logic. This result extends to many-sorted
logic and/or logic with equality. So that gives us a
good idea of just how far we can go with logic programming
and still provide the programming with a good idea of what
his program is about (namely, whatever is in the initial model).
------------------------------
End of PROLOG Digest
********************
∂05-Aug-85 1448 JMC coloring II
To: subramanian@SUMEX-AIM.ARPA
∂08-Apr-85 1624 avg@diablo coloring II
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 8 Apr 85 16:21:47 PST
Date: Mon, 8 Apr 85 16:22:15 pst
From: Allen VanGelder <avg@diablo>
Subject: coloring II
To: jmc@sail
/*
The program defines the relation color(Map, Colors),
between a map and a list of colors, which is true if Map
is legally colored using Colors.
-- Ehud Shapiro
A map is represented using adjacency-lists, where with
each country we associate its color, and the list of
colors of its neighbours. If the program is used in
generate-mode, i.e. to color an uncolored map, these
colors would be uninstantiated variables. The program
would then compute all possible colorings. For example,
the uncolored map
←←←←←←←←←←
| a |
|←←←←←←←←|
|b |c |d |
|←←|←←|←←|
|e |f |
|←←←|←←←←|
is represented by the term:
*/
map( [country(a, A, [B, C, D]),
country(b, B, [A, C, E]),
country(c, C, [A, B, D, E, F]),
country(d, D, [A, B, F]),
country(e, E, [B, C, F]),
country(f, F, [C, D, E])
]
).
color←map([Country | Map], Colors) :-
color←country(Country, Colors),
color←map(Map, Colors).
color←map([], ←Colors).
color←country(country(←Name, C, AdjacentCs), Colors) :-
remove(C, Colors, Colors1),
subset(AdjacentCs, Colors1).
subset([C | Cs], Colors) :-
remove(C, Colors, ←),
subset(Cs, Colors).
subset([], ←Colors).
remove(C, [C | Cs], Cs).
remove(C, [C1 | Cs], [C1 | Cs1]) :-
remove(C, Cs, Cs1).
colors([red, green, blue, white]).
test(Map) :-
map(Map),
colors(Colors),
color←map(Map, Colors).
∂05-Aug-85 1448 JMC coloring III
To: subramanian@SUMEX-AIM.ARPA
∂08-Apr-85 1653 avg@diablo coloring III
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 8 Apr 85 16:53:08 PST
Date: Mon, 8 Apr 85 16:53:35 pst
From: Allen VanGelder <avg@diablo>
Subject: coloring III
To: jmc@sail
The file in the message "coloring IV" has an added rule for color←map
that applies the postponement heuristic, plus rules for the new
predicate easy←country that selects a country with < 4 neighbors if
possible. It seems to work.
∂05-Aug-85 1448 JMC coloring IV
To: subramanian@SUMEX-AIM.ARPA
∂08-Apr-85 1655 avg@diablo coloring IV
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 8 Apr 85 16:55:35 PST
Date: Mon, 8 Apr 85 16:56:01 pst
From: Allen VanGelder <avg@diablo>
Subject: coloring IV
To: jmc@sail
/*
The program defines the relation color(Map, Colors),
between a map and a list of colors, which is true if Map
is legally colored using Colors.
-- Ehud Shapiro
Postponement heuristic added -- A.V.G.
A map is represented using adjacency-lists, where with
each country we associate its color, and the list of
colors of its neighbours. If the program is used in
generate-mode, i.e. to color an uncolored map, these
colors would be uninstantiated variables. The program
would then compute all possible colorings. For example,
the uncolored map
←←←←←←←←←←
| a |
|←←←←←←←←|
|b |c |d |
|←←|←←|←←|
|e |f |
|←←←|←←←←|
is represented by the term:
*/
map( [country(a, A, [B, C, D]),
country(b, B, [A, C, E]),
country(c, C, [A, B, D, E, F]),
country(d, D, [A, B, F]),
country(e, E, [B, C, F]),
country(f, F, [C, D, E])
]
).
color←map([], ←Colors).
color←map(Map, Colors) :-
easy←country(Map, Country, Map1),
!, % Require use of easy←country when possible (optinal cut)
color←map(Map1, Colors),
color←country(Country, Colors).
color←map([Country | Map], Colors) :-
color←country(Country, Colors),
color←map(Map, Colors).
color←country(country(←Name, C, AdjacentCs), Colors) :-
remove(C, Colors, Colors1),
subset(AdjacentCs, Colors1).
easy←country([Country | Map], Country, Map) :-
Country = country(←Name, C, AdjacentCs),
length(AdjacentCs, N),
N < 4.
easy←country([Country | Map], Country1, [Country | Map1]) :-
Country = country(←Name, C, AdjacentCs),
length(AdjacentCs, N),
N >= 4,
easy←country(Map, Country1, Map1).
subset([], ←Colors).
subset([C | Cs], Colors) :-
remove(C, Colors, ←),
subset(Cs, Colors).
remove(C, [C | Cs], Cs).
remove(C, [C1 | Cs], [C1 | Cs1]) :-
remove(C, Cs, Cs1).
colors([red, green, blue, white]).
test(Map) :-
map(Map),
colors(Colors),
color←map(Map, Colors).
∂05-Aug-85 1448 JMC coloring V
To: subramanian@SUMEX-AIM.ARPA
∂08-Apr-85 1700 avg@diablo coloring V
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 8 Apr 85 17:00:49 PST
Date: Mon, 8 Apr 85 17:01:08 pst
From: Allen VanGelder <avg@diablo>
Subject: coloring V
To: jmc@sail
The file in coloring IV is intended to give the least perturbation
solution to incorporating the postponement heuristic. The final
result does not represent good programming style in my opinion.
If you get a coloring VI message, it will be my idea of the program done
in correct style. I haven't done it yet though.
∂05-Aug-85 1449 JMC coloring V-a
To: subramanian@SUMEX-AIM.ARPA
∂08-Apr-85 1736 avg@diablo coloring V-a
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 8 Apr 85 17:36:17 PST
Date: Mon, 8 Apr 85 17:36:39 pst
From: Allen VanGelder <avg@diablo>
Subject: coloring V-a
To: jmc@sail
The file in coloring VI is the one promised earlier. It is a little
awkward because I wanted to avoid Prolog's "or via semi-colon" and
stick to pure clause form. Compared to the earlier version that has
postponement, in this version easy←country always succeeds, but its last
argument is "no" if there is no easy country left. Then
color←map←choose switches to color←map←brute instead of recurring on
color←map.
The program worked on the one map furnished.
∂05-Aug-85 1449 JMC coloring VI
To: subramanian@SUMEX-AIM.ARPA
∂08-Apr-85 1737 avg@diablo coloring VI
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 8 Apr 85 17:36:49 PST
Date: Mon, 8 Apr 85 17:37:12 pst
From: Allen VanGelder <avg@diablo>
Subject: coloring VI
To: jmc@sail
/*
The program defines the relation color(Map, Colors),
between a map and a list of colors, which is true if Map
is legally colored using Colors.
-- Ehud Shapiro
Postponement heuristic added.
Recoded to use guards instead of cuts to eliminate redundant solns.
-- A.V.G.
A map is represented using adjacency-lists, where with
each country we associate its color, and the list of
colors of its neighbours. If the program is used in
generate-mode, i.e. to color an uncolored map, these
colors would be uninstantiated variables. The program
would then compute all possible colorings. For example,
the uncolored map
←←←←←←←←←←
| a |
|←←←←←←←←|
|b |c |d |
|←←|←←|←←|
|e |f |
|←←←|←←←←|
is represented by the term:
*/
map( [country(a, A, [B, C, D]),
country(b, B, [A, C, E]),
country(c, C, [A, B, D, E, F]),
country(d, D, [A, B, F]),
country(e, E, [B, C, F]),
country(f, F, [C, D, E])
]
).
color←map([], ←Colors).
color←map(Map, Colors) :-
easy←country(Map, Country, Map1, Easy),
color←map←choose(Country, Map1, Easy, Colors).
color←map←choose(Country, Map, yes, Colors) :-
color←map(Map, Colors),
color←country(Country, Colors).
color←map←choose(Country, Map, no, Colors) :-
color←map←brute([Country | Map], Colors).
color←map←brute([], ←Colors).
color←map←brute([Country | Map], Colors) :-
color←country(Country, Colors),
color←map←brute(Map, Colors).
color←country(country(←Name, C, AdjacentCs), Colors) :-
remove(C, Colors, Colors1),
subset(AdjacentCs, Colors1).
easy←country([Country | Map], Country, Map, yes) :-
Country = country(←Name, C, AdjacentCs),
length(AdjacentCs, N),
N < 4.
easy←country([Country], Country, [], no) :-
Country = country(←Name, C, AdjacentCs),
length(AdjacentCs, N),
N >= 4.
easy←country([Country | Map], Country1, [Country | Map1], Easy) :-
Country = country(←Name, C, AdjacentCs),
length(AdjacentCs, N),
N >= 4,
easy←country(Map, Country1, Map1, Easy).
subset([], ←Colors).
subset([C | Cs], Colors) :-
remove(C, Colors, ←),
subset(Cs, Colors).
remove(C, [C | Cs], Cs).
remove(C, [C1 | Cs], [C1 | Cs1]) :-
remove(C, Cs, Cs1).
colors([red, green, blue, white]).
test(Map) :-
map(Map),
colors(Colors),
color←map(Map, Colors).
∂05-Aug-85 1450 JMC
To: subramanian@SUMEX-AIM.ARPA
∂08-Apr-85 2021 VGA avoiding space problems in prolog
Apparently the Prolog compiler does better with space. My files in [1,VGA]
can be examined and copied. prolog.ini is a souped-up version of FY's that
has a compile macro. color.pl is coloring II" with some changes to make it
compilable. The beginning comment should be self-explanatory. That comment
and the "public" lines after it an be opied to other coloring files and
should be sufficient to make them compilable.
This should also help to avoid running out of space before getting to the
first solution on somewhat larger problems.
∂05-Aug-85 1450 JMC Re: doubt
To: subramanian@SUMEX-AIM.ARPA
∂09-Apr-85 1228 avg@diablo Re: doubt
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 9 Apr 85 12:27:20 PST
Date: Tue, 9 Apr 85 12:27:51 pst
From: Allen VanGelder <avg@diablo>
Subject: Re: doubt
To: JMC@SU-AI.ARPA
I intended that easy←country succeed if ANY country in the list is easy,
not just if the first one is easy. Thus if b and d are easy in
map [a, b, c, d, e], the sequence of subgoals should be something like this:
color←map [a, b, c, d, e].
color←map [a, c, d, e], color←cty b.
color←map [a, c, e], color←cty d, color←cty b.
color←map←brute [e, a, c], color←cty d, color←cty b.
* color←cty e, color←cty a, color←cty c, color←cty d, color←cty b.
with liberal abuse of notation and some steps omitted. The starred line
does not exist all at once, but does reflect the final order in which
countries are colored.
If it does not work this way, it can be fixed to do so.
Color IV and Color VI were supposed to do ALMOST the same thing.
However, the simpler Color IV eventually reduces to the more natural order
on the brute-force portion:
* color←cty a, color←cty c, color←cty e, color←cty d, color←cty b.
∂05-Aug-85 1451 JMC
To: subramanian@SUMEX-AIM.ARPA
∂09-Apr-85 2032 VGA Prolog magic
Shapiro's program is meta-inefficient in the way it instantiates colors
via unbound arguments of subset. We really only want to check that the
adjacent countries whose colors have been chosen do not conflict with the
current choice. A one-line fix to the subset procedure using \+ \+
does the trick. See color.pl[1,VGA]. You will see a line
\+ \+ remove(...),
where there used to be just
remove(...),
This enables the program to find all 120 colorings without running out of
space. The same fix should be put in all versions, or the heuristics have
no effect, because it blindly instantiates all countries colors in the
course of satisfying one "subset" subgoal.
Also, there is a typo in map. Country D should be adjacent to [A, C, F],
not to [A, B, F].
Recall that \+ loosely means "not provable" in prolog. \+ \+ is double
negation. If remove(...) can succeed, then \+ \+ remove(...) will also
succeed once but will not do any bindings.
Finally, I checked over my code for Color VI and it does indeed postpone
countries a, b, d, e, f, but not c. I verified this using trace.
∂05-Aug-85 1451 JMC
To: subramanian@SUMEX-AIM.ARPA
∂05-Jun-85 0011 JJW Ideas on my thesis
To: JMC@SU-AI.ARPA, RPG@SU-AI.ARPA, ullman@SU-AIMVAX.ARPA
Here's a summary of some of my recent thoughts on parallel Lisp, and
proposals for what to do next. This is a rather long message, so I hope
you have the time to read it.
Two distinguishing characteristics of existing proposals for parallel
Lisps are the language constructs and the target architectures. I do
not propose to introduce new ideas in either of these areas, though for
convenience I'd like to abstract the existing forms into whatever turns
out to be most useful.
As a basis for language constructs, I'm considering those of Qlambda and
Multilisp.
The main abstraction for language features is what I'll call "conditional
futures". Futures are representations of values being computed by
separate processes. They can be passed around just like other data
objects (in the Lisp style of passing pointers), but if a process performs
an operation that requires the value being computed, it will suspend if
that value is not yet available and automatically resume when it is.
Once the value of a future has been computed, a distinction need no longer
be made between the future and the value it represents. Thus, futures are
semantically equivalent to their values; the only difference is
operational, in that futures have the potential of speeding up computation
which does not immediately need to reference the value, but there is also
an associated overhead cost for using futures.
A conditional future consists of a predicate, and a form that computes a
value. If the predicate is true, the result is a future with a new task
spawned to compute the value. If false, the process evaluating the
conditional future itself computes the value, avoiding the overhead.
Neither Qlambda nor Multilisp directly provides conditional futures. In
Qlambda, the conditional predicates are explicit in the QLET and QLAMBDA
forms, but the futures are implicit. Multilisp has explicit futures, but
only unconditional.
Neither of these obstacles is at all important. In Qlambda, we can create
an unconditional future to compute an expression E by writing
(QLET 'EAGER ((X E)) X)
and a future conditional on P to compute E by writing
(QLET (IF P 'EAGER NIL) ((X E)) X)
In Multilisp, the unconditional future is (FUTURE E), and the conditional
future is
(IF P (FUTURE E) E)
I don't want to waste time arguing whether Qlambda or Multilisp has a better
syntax for expressing conditional futures. Obviously in either language one
could define a macro (CFUTURE P E) that would have the desired effect. The
Qlambda primitives and the Multilisp primitives can also be defined in terms
of each other with no runtime performance penalty.
The next question is whether futures are a sufficient basis with which to
do parallel computation. In an applicative language, I claim this is so.
(I will argue for applicative computation in a moment.) To prove this
claim, I will have to show that the introduction of futures into an
applicative program always leads to a computation that takes a minimum
time given enough processors, or can attain the minimum possible on fewer
processors. If the cost (overhead) of futures is zero, this seems fairly
clear to me, though I haven't proved it yet.
(All of this assumes no change in the algorithm itself. The introduction
of parallel algorithms can of course provide even more speedup, but that
sort of high-level decision is left to the programmer. It will be
important to show that applicative Lisp with futures is a good language
for expressing parallel algorithms.)
When there is a non-zero cost for creating futures, then there are some
computations that run faster when less than full parallelism is used.
The trick, then, is to use conditional futures with predicates chosen to
optimize the running time.
That is basically what I want to do. The open question is what, among the
wide range of possibilities, should be used for the conditions. Some
general classes of conditions are: (1) extra variables introduced into the
program, such as the "depth" in the Fibonacci example in the Qlambda
paper; (2) extra information stored in the data, such as the "size" of the
data used in a subcomputation, or some other estimate of the time needed
for the subproblem; (3) information about the current state of the
multiprocessor, to avoid creating more parallelism when there is enough to
keep all processors busy. There must be other possibilities as well.
I am hoping to find some general principles that will guide programmers to
the proper choice of these conditions, or even better, that will enable
compilers and interpreters to choose them automatically. To look for
these general principles, I will need to try test cases, see what works
and what doesn't, and find out why. I don't think I can come up with the
answers in any other way, and at this time I'm about ready to start the
experiments.
Some sort of simulation of the multiprocessor Lisp environment is needed,
since at present we don't have a running multiprocessor Lisp. I would
like to avoid starting from scratch, and there are two existing tools that
could be used. One is the Qlambda simulator written by RPG for the
Symbolics Lisp Machine. The other is the Multilisp simulator written by
Halstead for the VAX.
I don't yet have enough experience with either of these to say which is
better, both in terms of the statistics-gathering facilities available and
speed of execution. If there's a clear winner, I'll use it. There is
also the possibility of running the Multilisp simulator on the S-1 at
Livermore, since it has Unix. (In theory the Qlambda simulator should be
easy to adapt to S-1 Lisp, but no one has yet brought up S-1 Lisp running
under an operating system there.)
My intention is to parameterize the simulation to allow studying the
effects of different costs for the creation of futures and other primitive
operations. If these variables have a direct and simple effect on the
choice of predicates for conditional futures, there's a useful result.
Both of these simulators are geared towards standard von Neumann
processors with a shared global memory. Other architectures for
multiprocessors allow greater expansion before performance starts to
degrade, and these should be investigated. I think the simulators are
flexible enough so that this is fairly simple. The dataflow model of
computation is also very suitable for futures, and will be considered.
Above, I promised to argue in favor of applicative programming. The
use of side effects in conjunction with parallelism is known to make
writing and debugging programs difficult. Applicative programs have a
clean semantics and are equivalent in their sequential and parallel
implementations. Thus, there is a lot to be said for not using side
effects.
On the minus side, it is sometimes hard to express algorithms efficiently
in a purely applicative language, especially when one uses high-order
abstractions like mapping functions and combinators. But this can often
be solved by using "lazy evaluation", which avoids performing parts of
the computation that are not needed for the result.
It may be going a bit far to say that applicative programming with lazy
evaluation is both expressive and efficient enough for all general
programming tasks. But I think a large class of problems do fall into
this category, and for these it is worth the effort to use this style.
On a parallel processor, though, one may not want to adhere to lazy
evaluation all the time. There is potential gain in computing a
subexpression's value by the time it is needed, although there is a
potential loss in computing it if it is never used. (And there is the
danger of running on forever if the program uses the "infinite data
structures" that lazy evaluation allows.)
This leads to an extension of the concept of conditional futures. As
well as deciding whether to compute a subexpression in a separate
process, we can give it a priority (or assign it a fraction of the
available processing power), thus placing it on a continuum between
strict lazy evaluation and full eager evaluation. The assignment of
these priorities is another question that can be studied in experiments.
Summary:
Here are the things that I propose to do.
1. Show that futures are a sufficient construct to express parallel
execution of applicative programs, and that Lisp with futures is a
good language in which to express parallel algorithms.
2. Experiment with various predicates for conditional futures, to see
what performs best using a variety of assumptions about the parallel
architecture.
3. Experiment with priorities for lazy evaluation in conjunction with
conditional futures.
4. Systematize the results, if possible, into guidelines that can be
applied in writing parallel Lisp programs, and in automating the
introduction of these predicates into programs.
∂05-Aug-85 1636 JMC
To: subramanian@SUMEX-AIM.ARPA
It turns out Wednesday at 2pm isn't suitable. How about Thurs?
∂05-Aug-85 1702 JMC re: Udi Shapiro
To: RICHARDSON@SU-SCORE.ARPA
[In reply to message sent Mon 5 Aug 85 16:42:19-PDT.]
Not that I know of. Please do it.
∂06-Aug-85 0934 JMC Kathy Harkin
To: SMC
called and needs your USCTA number and the owner of the horse
in order to report results of show.
851-7855.
∂06-Aug-85 0937 JMC phone message
To: SMC
Kathy Harkin called and needs your USCTA number and the owner of the horse
in order to report results of show. 851-7855.
∂06-Aug-85 0938 JMC message
To: smc@SU-SCORE.ARPA
Kathy Harkin called and needs your USCTA number and the owner of the horse
in order to report results of show. 851-7855.
∂06-Aug-85 1140 JMC
To: RA
That's Aaron Sloman. Thanks.
∂06-Aug-85 1744 JMC
To: RA
Only one is a senator. Dickinson is a member of the House. Here are
their addresses.
Hon. John W. Warner
U.S. Senate
421 Russell Senate Office Bldg.
Wash. D.C. 20510
Hon. William L. Dickinson
U.S. House of Representatives
2406 Rayburn House Office Bldg.
Washington, D.C. 20515
∂07-Aug-85 0022 JMC
To: VAL
Dead birds don't fly
As long as we are willing to put the fact that dead birds
don't normally fly in the part of our database concerned with
birds and flying there is no problem. We have
∀x.dead x ⊃ ab aspect2 x
and
∀x.bird x ∧ ¬ab aspect7 x ⊃ ¬flies x.
However, when we want to include more information, this
becomes inefficient. Not only don't dead birds fly, they don't
sing, lay eggs or eat. On the other hand, if a bird is black
in color, it normally remains black when dead.
The partial solution of the problem when we are willing
to treat dead birds not flying in a special way, should not be forgotten.
It might be all we know how to do now, and even when we can do better
in general, if a program has to think a lot about whether a bird
flies, it may be useful to keep this information in the database
even though it is a consequence of more general information.
In general, however, we know that dead animals don't act,
birds are animals, and flying is an action. We can get by making
this an absolute, but suppose we want to allow for zombies.
Certainly a child is prepared to treat being informed (misinformed)
about the existence of zombies in the same way as he treats any
other information about exceptions to his previously learned
generalizations. Therefore, we should axiomatize death in a way
that interacts properly with actions of all kinds and which has
the requisite generality.
Remark: Minsky's bird with its feet encased in concrete may offer
additional difficulty. Since this is an ad hoc phenomenon,
there is no information about the ability of entities with
their feet encased in concrete to fly. It has to follow
from an understanding of mechanism. What isn't clear is
how the priorities of the mechanism are established.
∂07-Aug-85 1522 JMC
To: RA
harman.1
∂07-Aug-85 2124 JMC
To: RA
yarima.1 and please retrieve letter from Yarim-Agaev and fix name his center
∂07-Aug-85 2211 JMC re: Forum on Risks to the Public in Computer Systems (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - What a pity that discussion of risks to the public in not using
on-line computer systems isn't solicited.
∂07-Aug-85 2222 JMC re: [Andrew Yao <YAO@SU-SCORE.ARPA>: [Ehud Shapiro <udi%wisdom.bitnet@WISCVM.ARPA>: Re: talk]]
To: ashok@SU-SUSHI.ARPA, Ashok@SU-SCORE.ARPA, YAO@SU-SCORE.ARPA,
udi%wisdom.bitnet@WISCVM.ARPA
[In reply to message from ashok@SU-SUSHI.ARPA sent Wed 7 Aug 85 22:12:53-PDT.]
From what I understand, Shapiro doesn't consider himself as being mainly
in AI. His work on concurrent prolog is systems work but is also relevant
to databases and programming languages. He should meet with Jeff Ullman
if he is available.
∂08-Aug-85 1132 JMC re: Reagan, profanity, liberal clap-trap (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - Touche, Rosenbloom. I defended Reagan against MRC's charge
of quarter-senility by commenting on his effectiveness as
a politician, both with the voters and with Congress. However,
with your kind invitation, it is certainly appropriate to say
what I like about Reagan's policies.
As with every President, what Reagan is accomplishing in carrying out
his program is limited by what he can get Congress to do, as well as
by elements of the program that turn out to be unrealistic. Reagan
has the further problem that the bureaucracy is mostly Democratic.
Nevertheless, there are accomplishments:
1. Inflation is way down, and unemployment is somewhat down. The
relation of this to Reagan's policies is hard to assess, but certainly
the Democrats weren't promising or predicting anything of the kind.
2. The growth of the welfare state has slowed. The view of society
as a bundle of minorities has been somewhat checked.
3. Supreme Court activism has been somewhat checked.
4. The space station project has been started.
5. The Strategic Defense Initiative has started the country moving
away from mutually assured destruction. Whether it will be a technical
success remains to be seen.
6. Reagan understands that communism taking over a country is both a
tragedy for the country and a danger to the free world.
7. In Grenada, he saw an opportunity and took it. Too bad it was
such a small opportunity.
8. The situation in El Salvador has stabilized. Incidentally, this
involved the Reagan Administration going against its prejudices about
what forces within El Salvador are the best defense against communism.
9. The principle that it is ok to revolt against communism and for
us to support such revolts has been established, however shakily,
in Nicaragua and Afghanistan.
10. Reagan, by his manner, has increased the level of civility in
Washington. I didn't have any expectation that this was possible,
and I regard it as one of Reagan's greatest accomplishments.
As an intellectual, I would prefer a more intellectual President.
Jeane Kirkpatrick is my choice for Reagan's successor. However,
this reminds me that once the frogs (liberals) asked God for a
king, and God sent them a log. The log was impressive but inactive,
and the frogs complained to God about the lack of leadership. God
then sent them a stork for king who ate frogs.
∂08-Aug-85 1349 JMC re: Rosenbloom vs. Rosenblum
To: pr@SU-AIMVAX.ARPA
[In reply to message sent 8 Aug 1985 13:18-PDT.]
I knew he was David and you Paul, but I was careless in verifying the
spelling. Sorry about possibly causing confusion.
∂08-Aug-85 1353 JMC
To: pednault@SRI-AI.ARPA
What number can I phone you on?
∂09-Aug-85 1047 JMC re: "treason" & "unpatriotic" (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - It's spelled Fascist. It was originally the name of Mussolini's
Italian (1919) doctrine and comes (via Italian) from the Latin "fasces",
an ax in a bundle of sticks, signifying unity. The doctrine wasn't
originally racist, since it advocated militarism for all nations and
the survival of the most successfully militarist. Nazism added the
German racism, and Italian fascism was influenced by it.
The Brittanica quotes
Mussolini:
"war alone brings up to their highest tension all human energies and puts
the stamp of nobility upon the peoples who have the courage to meet it.
Fascism carries this anti-pacificist struggle into the lives of individuals.
It is education for combat .... War is to the man what maternity is to the
woman. I do not believe in perpetual peace; not only do I not belive in it
but I find it depressing and a negation of all the fundamental virtues of
man."
One of the major favorable development since World War II is that today
no significant politician or literary person admits regarding war as
desirable in itself.
∂09-Aug-85 1636 JMC add my name
To: golsw%cernvm.bitnet@WISCVM.ARPA
Please add my name to the statement on Zelichonok. John McCarthy, Professor
of Computer Science Stanford. Here is the statement I received.
WE HAVE JUST BEEN SHOCKED TO LEARN OF THE IMMINENT OPENING OF THE TRIAL
OF OUR COLLEAGUE ROALD ZELICHONOK. WE WELL KNOW THAT THIS MAN IS HONEST,
SINCERE AND LAW-ABIDING, AND WE ARE ABSOLUTELY CERTAIN OF HIS INNOCENCE.
WE HEREBY REQUEST THAT HE BE IMMEDIATELY LIBERATED AND AUTHORIZED TO
EMIGRATE TO ISRAEL.
∂09-Aug-85 1815 JMC
To: CLT
They accept for 7ish on Sunday.
∂09-Aug-85 2154 JMC re: incredibly funny story about the Reagan administration (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - MRC finds this incredibly funny, but I find it presents interesting
questions. What are the limits of the Freedom of Information Act in terms
of the obligation to release letters? Before you jump to a conclusion,
ask what rule distinguishes citizens with some right to privacy from
lobbyists who might be influencing Government officials improperly.
Common Cause etc. have successfully forced Government officials to keep
logs of telephone calls that are available for inspection. If you have
certain views of who are the good guys, you could require that letters of
"protest" by "citizens" be kept private, while letters "exerting pressure"
from "special interests" be public. But the decision of which is which is
subjective. The second question concerns what an official may do in his
spare time. What counts as using his position to intimidate, etc.?
However, don't worry. Someday there will be a different Adminstration and
people will get computer mail from Government officials calling them
cretins.
∂10-Aug-85 1029 JMC bboard goof
To: Crispin@SUMEX-AIM.ARPA
You misused the REPLY macro in your latest with the effect that
SAIL had the work of redistributing your message, and it ended up
on every BBOARD except SAIL's. I leave you to figure out how to
get your message on SAIL's BBOARD without duplicating it on the
others. Then I'll post my brilliant reply.
∂10-Aug-85 1338 JMC re: diophantine equations with sums of squares
To: DEK
[In reply to message rcvd 10-Aug-85 12:45-PT.]
Thanks for the suggestions. The Hasse-Minkowski theorem seems not
to be relevant, but I'll look in Geometry of Numbers books. I checked
Diophantine Geometry which is also not relevant. Don't bother with the
following unless you are curious.
The original problem was whether it was possible to have an
equilateral triangles whose vertices are lattice points. It isn't even
possible to have a 60 degree angle, because all angles between lattice
segments have rational tangents. It turns out that a figure made of lines
and points is similar to a "lattice figure" iff all angles that can be
formed have rational tangents.
The next problem was to find the smallest similar figure. This is
readily obtained from any lattice figure by putting one vertex at the
origin and regarding the other vertices as Gaussian integers and then
dividing all vertices by their gcd considered as Gaussian integers.
Then we come to ask about plane figures in 3-space, stimulated by
noting that the points (1,0,0), (0,1,0) and (0,0,1) form an equilateral
triangle. I have got this far. Consider a lattice angle in the plane
lx+my+nz=0, where l, m and n are integers. Any such angle has tangent a
rational multiple of sqrt(l↑2+m↑2+n↑2); I don't yet know whether all such
angles are possible.
In four dimensions one is tempted to try to use the gcd of
quaternions, but this doesn't give the general similarity transformation.
∂10-Aug-85 1725 JMC re: wednesday
To: CLT
[In reply to message rcvd 10-Aug-85 17:23-PT.]
OK, I may say something about proof checking, but I think I'd like to say
more about commitment and related ideas.
∂10-Aug-85 2252 JMC re: news from Nicaragua (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - MRC, you've really put your foot in it this time with your
"I haven't heard of any response from Reagan about this incident, but
it should be to sack this loony.", which goes far beyond the loony
in question in impropriety and indiscretion.
Let us imagine, if we can, that MRC is a political appointee of the
Reagan Administration concerned with personnel and with the public
relations of the Administration. Some high official, my imagination
doesn't go so far as to suppose it to be the President, gets a memo
from MRC ending, "Let's sack this loony". His first reaction is, "Hasn't
this guy learned after four and a half years of Reagan Administration
that you don't put `Let's sack this loony' in a memo, because it shows
prejudice and will be used by the lawyer for the "loony" or by someone
else who obtains it under the Freedom of Information Act. Doesn't he
know that we can't just sack loonies, we have to initiate dismissal
proceedings".
What MRC should have said is, "We must investgate whether this action by
Mr. Sundseth goes beyond what is permissible for a Government employee,
and we may be obliged to file charges leading to his dismissal". Note
that prejudgment about whether Sundseth should be fired is not expressed.
Now the question is what charges under what rules? I offer the following
challenge to MRC.
1. State what rule Sundseth has violated. Since it is probably too much
trouble to look up dozens of Government personnel regulations, for this
purpose we'll allow MRC to invent what he thinks the rule should be.
2. State what the personnel regulation should be that requires
Sundseth to be sacked. However, the regulation must be stated in a way
that doesn't mention the particular case. It should, of course, handle
Sundseth's claim that he was acting on his own and not as a Government
employee. Note that 2 is different from 1. Not every rule violation by
an employee permits or requires him to be dismissed.
3. The third exercise is to give the specific charges against Sundseth
that would lead to his dismissal under the rule.
∂11-Aug-85 1443 JMC re: Your bboard message on smoking
To: KARNICKY@SU-SCORE.ARPA
[In reply to message sent Fri 9 Aug 85 23:09:20-PDT.]
Thanks for the message.
∂12-Aug-85 0957 JMC re: [LAUBSCH%hplabs.csnet@csnet-relay.arpa: jhl@sail account]
To: Reuling@SU-SCORE.ARPA
[In reply to message sent Mon 12 Aug 85 08:13:33-PDT.]
Yes, close the jlh account.
∂12-Aug-85 1125 JMC re: Wiktor Marek
To: VAL
[In reply to message rcvd 12-Aug-85 10:30-PT.]
Sure,invite him.
∂12-Aug-85 1424 JMC re: Request to cancel 2 pm appt for today
To: SUBRAMANIAN@SUMEX-AIM.ARPA
[In reply to message sent Mon 12 Aug 85 12:27:51-PDT.]
How about Thursday 2pm?
∂13-Aug-85 1521 JMC re: [John McCarthy <JMC@SU-AI.ARPA>: re: Forum on Risks to the Public in Computer Systems (from SAIL's BBOARD) ]
To: Neumann@SRI-CSL.ARPA, G.GEOFF@SU-SCORE.ARPA
[In reply to message sent Tue 13 Aug 85 14:19:14-PDT.]
I was taking my model Petr Beckmann's book "The Health Hazards of not
Going Nuclear" in which he contrasts the slight risks of nuclear energy
with the very large number of deaths resulting from conventional energy
sources from, e.g. mining and air pollution. It seemed to me that your
announcement was similarly one sided in its consideration in risks of
on-line systems and ignoring the possibility of risks from their non-use.
I won't be specific at present, but if you or anyone else wants to make
the claim that there are no such risks, I'm willing to place a substantial
bet.
∂13-Aug-85 1540 JMC re: 29 year old job announcement (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - I don't understand Randall Neff's motivation in posting this ancient
advertisement for people to work on aircraft nuclear propulsion. I fear
it was that the advertisement seems ridiculous from a 1980s point of view.
My own reaction was that those were the good old days when the country was
technologically more adventurous than it has been recently. I recall that
a six years ago the Adminstrator of NASA answered a question as follows:
A: "There
is no way the U.S. could go back to the moon in this century"?
Q: You mean Congress wouldn't put up the money?
A: No, not even if Congress put up the money.
Q: How come we can't do in 20 years now what we did in ten years starting in
1960?
A: It's not the same country.
∂13-Aug-85 1558 JMC re: Central America and IJCAI (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - I disagree with Chris Stuart about U.S. Central American policy. I
also note that as a new visitor, he is probably ignorant of the extensive
BBOARD discussions of the topic and imagines he is bringing a challenge
that his audience hasn't seen before. However, I don't complain (with ME)
his using the net for this purpose. The ARPAnet is bound to be used
for all sorts of purposes unrelated to defense research as long as it
is the main means of inter-computer communication. DARPA should get
out of the network business; Dialnet protocols that use the ordinary
telephone should become the main means of communications, and institutions
should devise whatever means suit them to reduce their phone bills by
making people pay for their own non-business telephone use.
∂13-Aug-85 2220 JMC reply to message
To: DCO
[In reply to message rcvd 13-Aug-85 22:13-PT.]
Certainly not for breakfast. Lunch some time after IJCAI would be
fine.
∂14-Aug-85 0923 JMC re: IJCAI
To: KACZMAREK@USC-ISIF.ARPA
[In reply to message sent 14 Aug 1985 07:35:57 PDT.]
I think Thursday would suit me best, but I'll have to think further about
the topic. I'll be at the Hilton.
∂14-Aug-85 1140 JMC re: Planning
To: PEDNAULT@SRI-AI.ARPA
[In reply to message sent Wed 14 Aug 85 11:36:26-PDT.]
Thanks for the papers. I'll be in touch after IJCAI.
∂14-Aug-85 1635 JMC IJCAI as a forum
To: su-bboards@SU-AI.ARPA
Like Chris Stuart, I have also contemplated using IJCAI as a
forum. My issue concerns the computer scientists who have claimed, in one
case "for fundamental computer science reasons" that the computer programs
required for the Strategic Defense Initiative (Star Wars) are impossible
to write and verify without having a series of nuclear wars for practice. Much of
the press (both Science magazine and the New York Times) have assumed (in
my opinion correctly) that these people are speaking, not merely as
individuals, but in the name of computer science itself. The phrase
"for fundamental computer science reasons" was used by one of the
computer scientist opponents.
In my opinion these people are claiming an authority they do not
possess. There is no accepted body of computer science principles that
permits concluding that some particular program that is mathematically
possible cannot be written and debugged. To put it more strongly, I don't
believe that there is even one published paper purporting to establish
such principles. However, I am not familiar with the literature on
software engineering.
I think they have allowed themselves to be tempted into
exaggerating their authority in order to support the anti-SDI cause,
which they support for other reasons.
I have two opportunities to counter them. First, I'm giving
a speech in connection with an award I'm receiving. Since I didn't
have to submit a paper, I was given carte blanche. Second, I have
been asked by the local arrangements people to hold a press conference.
I ask for advice on whether I should use either of these opportunities.
I can probably even arrange for some journalist to ask my opinion on
the Star Wars debugging issue, so I wouldn't have to raise the issue
myself. Indeed since my position is increasingly public, I might
be asked anyway.
To make things clear, I have no position on the feasibility
of SDI, although I hope it can be made to work. Since even the
physical principles that will be proposed for the SDI system haven't
been determined, it isn't possible to determine what kind of programs
will be required and to assess how hard they will be to write
and verify. Moreover, it may be possible to develop new techniques
involving both simulation and theorem proving relevant to verifying
such a program. My sole present point is that no-one can claim
the authority of computer science for asserting that the task
is impossible or impractical.
There is even potential relevance to AI, since some of the
opponents of SDI, and very likely some of the proponents, have suggested
that AI techniques might be used.
I look forward to the advice of BBOARD contributors.
∂14-Aug-85 1837 JMC re: IJCAI talk
To: VAL
[In reply to message rcvd 14-Aug-85 17:31-PT.]
Good.
∂14-Aug-85 2155 JMC re: Prof. Shapiro's visit
To: ashok@SU-SUSHI.ARPA
[In reply to message sent Wed 14 Aug 85 19:30:51-PDT.]
Any time is ok with me. If no-one else prefers it, just after his talk
or just before lunch. I'll either take him to lunch or there can be
a collective lunch or if someone else can only meet then, I'll defer.
∂14-Aug-85 2352 JMC
To: RA
Did I ever reply to letter about French honorary degree?
∂15-Aug-85 0949 JMC re: IJCAI as a forum
To: Geoff@SRI-CSL.ARPA
[In reply to message sent 15 Aug 1985 08:56-PDT.]
It will be next Wednesday morning at IJCAI in L.A.
∂15-Aug-85 1543 JMC
To: subramanian@SUMEX-AIM.ARPA
COL1.[E81,JMC] prolog coloring program
color.ax[e81,jmc] discussion of axioms for coloring maps
FRED.PR[E81,JMC] Prolog coloring program using Fred's idea
MAPS.BIG[E81,JMC] map coloring and kowalski with notes
maps.pr[e81,jmc] prolog programs for coloring maps
maps[e81,jmc] Map coloring and the Kowalski doctrine
∂15-Aug-85 1554 JMC
To: subramanian@SUMEX-AIM.ARPA
∂18-Mar-85 1055 RESTIVO@SU-SCORE.ARPA
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 18 Mar 85 10:54:17 PST
Date: Mon 18 Mar 85 10:41:15-PST
From: Chuck Restivo <RESTIVO@SU-SCORE.ARPA>
To: JMC@SU-AI.ARPA
[cwr] the introspective code is under [1,CWR] as:
INTERP
MAP
KEMPE
sorry it took so long but I've been working on my chess stuff
too much.
-------
∂15-Aug-85 1643 JMC
To: smith@MCC.ARPA
The laser printer output is for Doug Lenat if he doesn't have it.
∂15-Aug-85 1648 JMC re: Joel Shirkin, News and Publications
To: RA
[In reply to message rcvd 15-Aug-85 16:47-PT.]
Tell him no.
∂15-Aug-85 1711 JMC
To: RA
Please send my two circumscription papers to Prof. Kenneth Arrow, Stanford.
∂16-Aug-85 1028 JMC re: Appointment of Carolyn
To: BSCOTT@SU-SCORE.ARPA
[In reply to message sent Fri 16 Aug 85 10:21:28-PDT.]
That's ok.
∂16-Aug-85 1309 JMC Centre Mondial
To: reddy@CMU-CS-C.ARPA
Because of their continued failure to pay what was agreed for past
work, I think I will have to resign. Now that JJSS is gone from it,
his letter is probably worthless, either morally or in court.
∂16-Aug-85 1640 JMC
To: RA
perrot.1
∂16-Aug-85 1641 JMC
To: RA
The address is in my out box.
∂16-Aug-85 2109 JMC re: sand and SDI (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - One of the SDI ideas is to zap the Russian missiles in some similar
way.
∂16-Aug-85 2117 JMC re: sand and SDI (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - As to the legality of zapping other people's satellites, lawyers have
been writing about "space law" since the 1950s, and the U.N. has negotiated
treaties about it. I believe treaties exist barring damaging other countries'
satellites. Some "space lawyers" consider that there is an implicit space
law in analogy to the traditional law of the sea. The traditions under
which pirates were tried and hanged are a kind of international common law.
However, I'll bet that if an American deliberately damages an American
launched satellite, he would be tried under the ordinary laws concerning
damaging property, whether Government or private.
∂16-Aug-85 2130 JMC arrogant unix programmers
To: su-bboards@SU-AI.ARPA
I would like a mail command that would allow sending to bboards with the
exception of unixes. Whoever programs them behaves in an arrogant and
uncivilized way returning messages for no good reason. Sprite and Helens
are particularly villainous.
∂16-Aug-85 2200 JMC sdi
To: su-bboards@SU-AI.ARPA
I thank those who advised me on whether to say something about the
SDI controversy in my lecture or at the press conference. I don't
presently intend to say anything about it in my lecture. Mainly
this is because thinking about what to say about a public issue
would interfere with thinking about AI. I may say something or
distribute a statement at the press conference.
I am not sure I understand the views of those who claim the computer
part of SDI is infeasible. Namely, do they hope it won't work? If
so, why? My reactionary mind thinks up hypotheses like the following.
It's really just partisanship. They have been against U.S. policy
in many areas including defense, that they automatically oppose any
initiative and then look for arguments.
∂17-Aug-85 0641 JMC re: Planning Workshop/Conference
To: georgeff@SRI-AI.ARPA
[In reply to message sent Fri 16 Aug 85 08:13:52-PDT.]
Yes, all that's true. Normal grants are up to $5,000, but we can go
as high as $10,000.
∂25-Aug-85 1503 Mailer failed mail returned
To: JMC@SU-AI.ARPA
In processing the following command:
MAIL
The command was aborted because these Host Name(s) are Unknown:
SUMEX.ARPA; MC.ARPA
------- Begin undelivered message: -------
∂25-Aug-85 1503 JMC re: Analytical Learning Workshop
To: Jaime.Carbonell@CMU-CS-A.ARPA
CC: mitchell@RUTGERS.ARPA, mostow@RUTGERS.ARPA,
minton@CMU-CS-CAD.ARPA, doyle@CMU-CS-C.ARPA
[In reply to message from Jaime.Carbonell@CMU-CS-A.ARPA sent Sun, 25 Aug 85 12:31 EDT.]
Exclusiveness is not a disqualification, and I'm happy to approve the
Workshop for a budget of up to $6K. Please arrange details with Claudia
Mazzetti.
------- End undelivered message -------
∂25-Aug-85 1523 JMC My continuing as workshop-approver
To: phw@MIT-MC.ARPA
Pat,
As you probably know, I have been serving as workshop-approver
at Woody's request. I started approving them when I was President. I
am agreeable to continuing, but I consider myself serving "at the pleasure
of the President". Would you like me to continue?
I have been considering the requests entirely informally and
have been authorized to approve up to $10K but typically $5K. When I
approve I copy Claudia and the current President. So far all the proposals
for genuine workshops have come from qualified people and I have approved
all of them. There have been some proposals for supporting large conferences
or other non-workshops on which I have passed the buck to the President.
Also I have done a certain amount of pump-priming and suggested that certain
people organize workshops, since the number of proposals is far smaller
than it ought to be.
I have imposed no specific rules on the form of workshop. The
organizer may establish whatever procedure he chooses for selecting
papers and participants and may provide a proceedings if he wishes.
There follows a proposal that I will approve forthwith
if you authorize me to continue the program.
John
∂25-Aug-85 0934 JC50@CMU-CS-A.ARPA Analytical Learning Workshop
Received: from CMU-CS-A.ARPA by SU-AI.ARPA with TCP; 25 Aug 85 09:34:02 PDT
Date: Sun, 25 Aug 85 12:31 EDT
From: Jaime.Carbonell@CMU-CS-A.ARPA
To: jmc@SU-AI.ARPA
Subject: Analytical Learning Workshop
CC: mitchell@RUTGERS.ARPA, mostow@RUTGERS.ARPA, minton@CMU-CS-CAD.ARPA,
doyle@CMU-CS-C.ARPA
Message-Id: <25Aug85.123133.JC50@CMU-CS-A.ARPA>
John,
I'm starting to organize a small, informal workshop on analytical
methods for machine learning (e.g. the application of constraint-back
propagation, derivation-based generalization supported by a TMS, etc.)
The idea would be to have 2 intensive days and a small number of
participants (less than 20) who are actively working on this new
approach -- about 1/2 faculty & 1/2 graduate students.
My question is whether you think this would be appropriate for AAAI
sponsorship. The budget would be modest, say 6K, as it would be held
at University premises. But, it definitely would not be open to the
public, or even all AAAI members. So, this might disqualify it.
--Jaime
∂26-Aug-85 1516 JMC re: Next Visit
To: ai.boyer@MCC.ARPA
[In reply to message sent Monday, 26 August 1985, 15:54-CDT.]
I hate to deny you an opportunity to practice negotiation, but the
dates you propose are fine with me. Sept. 9 and 10 it is.
∂26-Aug-85 1708 JMC ijcai expenses and award
To: scrocker@AEROSPACE.ARPA
Claudia reminds me that you are writing the checks. Anyway the
letter from Allan Mackworth said that the Research Excellence Award
included $1,000 plus expense. You already paid my hotel bill, so
all that remains is $70 taxis and $100 meals. Airfare was paid
from other sources. I know you are going on vacation, but I don't
remember how long. If it seems to long before I get a reply from
you, I'll try Mackworth, but I'm in no hurry, not being broke.
∂30-Aug-85 1353 JMC re: arrogant unix programmers
To: SATZ@SU-SIERRA.ARPA
[In reply to message sent Wed 28 Aug 85 10:30:10-PDT.]
It seems I owe a partial apology. It seems that "message refused" was some
kind of error.
∂30-Aug-85 1408 JMC re: Industrial lecturers
To: RA
[In reply to message rcvd 30-Aug-85 14:02-PT.]
Fernando Pereira of SRI, John Williams of IBM, and Dan Bobrow, et. al. of Xerox
∂30-Aug-85 1420 JMC re: leave early
To: RA
[In reply to message rcvd 30-Aug-85 14:16-PT.]
It would be good if you could schedule such appointments for the morning
as much as possible.
∂30-Aug-85 1453 JMC re: Would you like to be on the distribution list, or just BBOARDed?
To: Neumann@SRI-CSL.ARPA
[In reply to message sent Wed 28 Aug 85 11:41:15-PDT.]
Please put JMC-LISTS@SU-AI on the list. That's where I receive mailing list
mail.
∂30-Aug-85 1651 JMC
To: RA
∂30-Aug-85 1545 SJG circumscription paper
Dear John:
Glenn Shafer asked me to get a copy of your latest for him. If you'll
give it to me, I'll send it along.
Matt
∂30-Aug-85 2308 JMC
To: mazzetti@SUMEX-AIM.ARPA
scrocker@aerospace doesn't work.
∂30-Aug-85 2309 JMC
To: bscott@SU-SCORE.ARPA
I wish to apologize for this morning.
∂30-Aug-85 2343 JMC
To: nilsson@SU-SCORE.ARPA
brag[s85,jmc] Self-praise for various purposes
John McCarthy was born in Boston in 1927. His father was
born in Ireland, immigrated in 1917, and worked at various trades
including carpentry but spent the middle part of his career as
a trade union organizer. His mother was born in Lithuania of
Jewish parents, immigrated at the age of six, graduated from
Goucher College in Baltimore and worked as a journalist and
as a social worker.
McCarthy attended high school in Los Angeles, graduated
from Caltech in mathematics in 1948 and received the PhD in
mathematics in 1951 from Princeton University.
His thesis was on partial differential
equations, and his adviser was Solomon Lefschetz.
McCarthy remained at Princeton as an instructor for two years
after his PhD and then went to Stanford University as an acting assistant
professor. He moved to Dartmouth College in 1955, to M.I.T. in
1957 and back to Stanford in 1962 where he remains.
While still at Caltech in 1949, McCarthy was influenced
by attending the Hixon Symposium on Cerebral Mechanisms in Behavior
to begin thinking about how machines could be made to think. He
found this a difficult problem and first published on it in 1956
in Automata Studies, which he co-edited with Claude Shannon, then
of Bell Telephone Laboratories.
While at Dartmouth his attention shifted from pure mathematics to
artificial intelligence and computer science. He coined the term
artificial intelligence in connection with the Dartmouth Summer Research
Project on Artificial Intelligence which he organized for the summer of
1956. This meeting was the first occasion that united most of the
individuals thinking about how to make computers behave intelligently.
Taking ideas from the IPL language of Allen Newell and Herbert
Simon and from Fortran developed by John Backus at IBM, McCarthy began work
on an algebraic programming language for computation with symbolic
expressions. The first result was the Fortran List Processing Language of
Gelernter and Gerberich, but the main outcome was the development of LISP
in 1958.
LISP was not just a synthesis of its antecedents but
introduced many new ideas to computer science. These
include recursive use of conditional expressions as a form of
programming, the use of list structure representing S-expressions
for both program and data permitting convenient programs that
operate on programs, the use of Church's lambda notation for
representing functions and the first meta-cyclic interpreter.
Within a few years, LISP became the main programming language
used for artificial intelligence and computation with algebraic
expressions. It is now the second oldest (after Fortran) programming
language in general use. In the last few years several companies
(Xerox, Symbolics, LMI and Texas Instruments) have marketed computers
specialized for LISP. Numerous companies specialize in producing
LISP compilers for standard machines, and the Defense Department
has made Common LISP a standard along with ADA for programs
for computers imbedded in military equipment.
Besides using LISP as a practical programming language, McCarthy
based on it the first ``mathematical theory of computation''. This theory
permits the writing of specifications for programs in languages of
mathematical logic and permits formal computer-checked proofs that
programs meet their specifications. McCarthy's 1961 approach to this
problem has been developed by him and others into one of the two main
approaches to proving programs correct. In the hands of Robert Boyer and
J Moore of the University of Texas it has led to powerful interactive
theorem provers that are on the verge of real usefulness to computer
programmers and to mathematicians. (The other approach was proposed by
Robert Floyd in 1965, although it had been used in unpublished special
cases by both Allan Turing and John von Neumann). McCarthy developed the
mathematical theory of computation in series of papers in the 1960s and
1970s.
While at M.I.T. starting in 1957, McCarthy proposed
a new way of using computers, time-sharing. Rather than preparing
programs using key punches and then submitting decks of punched
cards, time-sharing involves each user having a terminal in his
office directly connected to the computer. To the user a
time-sharing computer behaves as if he had a computer of his
own. He prepares programs and data using editor programs and
keeps the information permanently in the computer's disk storage.
In order to make time-sharing practical, it was necessary to
add to the computers of that day features that have since become
standard. These include the following. Memory protection
prevents an erroneous program from damaging other programs.
Memory relocation allows a program to be located anywhere
the operating system finds space for it on its successive runs.
Replacing all instructions that can stop the computer by interrupts
to the operating system is also required to prevent erroneous
programs from affecting other users. Direct input-output
instructions in ``user mode'' need to be inhibited and replaced
by calls to the operating system. All input and output needs to
result in interrupts, so that the operating system can handle
input and output asynchronously with the users' computations.
McCarthy was the first to propose that an integrated package of such
features was required for time-sharing.
McCarthy began proposing that M.I.T. switch to this way
of using computers in 1957, but the first presently available
document is an internal memo dated January 1, 1959. In it
McCarthy proposed that M.I.T. enlist IBM's co-operation in
developing a time-sharing system for the forthcoming IBM 7090.
McCarthy's proposals were further developed by others and
contributed to the CTSS time-sharing system for the IBM 7090
built under the direction of Fernando J. Corbato.
However, the first time-sharing system to actually operate was
developed under McCarthy's direction for the Digital Equipment Corporation
PDP-1 computer at Bolt Beranek and Newman. The project was sponsored by
NIH, and McCarthy's co-authors were J.C.R. Licklider, Sheldon Boilen and
Edward Fredkin. The system began working in the summer of 1962.
McCarthy has always regarded his work on LISP, mathematical
theory of computation, and time-sharing as secondary to his
main scientific interest --- artificial intelligence.
His first approach was through automata theory and information
theory starting in 1949. It involved a measure of the amount of
information about its environment and automaton could collect.
Since it didn't seem to relate much to actual common sense knowledge
and behavior, it was abandoned without publication. The second
approach led to the 1956 paper ``Inversion of functions defined
by Turing machines'' but was abandoned for the same reason. In
1957 McCarthy started on a chess program and invented the alpha-beta
and killer heuristics, which are still used today in chess programs.
However, his talents were more suited to theory than to making large
programs, and he didn't continue his work on games except to direct
student work.
McCarthy's greatest scientific contribution is the ``reasoning
program'' approach to artificial intelligence pioneered in his 1960 paper
(given at a 1958 conference) ``Programs with common sense''. This
approach involves representing facts about the world and the problem the
program is to solve as sentences of mathematical logic. The program then
must use directed search procedures in order to deduce that a certain
course of action will achieve the goal.
This logic-based approach to artificial intelligence has
proved very difficult to implement, but so have all the others.
Others, in fact, have turned out to be special cases of the logic-based
approach. At present logic based approaches to AI are increasingly
popular.
In the 1960s McCarthy developed the ``situation calculus''
formalism for representing the effects of actions. Much of the
work in making AI programs that plan sequences of actions to achieve
goals is based on situation calculus or its variants.
In the 1970s McCarthy developed the ``circumscription'' method
of non-monotonic reasoning as an supplement to logical deduction
in reasoning programs. It has proved the most widely accepted of the
three methods proposed in that decade, and many papers have followed
McCarthy's 1980 paper. (1980, 1985).
In the 1960s and 70s McCarthy developed formalisms for
representing facts about people's and programs' knowledge. This field has
recently become popular.
∂31-Aug-85 1748 JMC your "Meinongian semantics"
To: rapaport%buffalo@CSNET-RELAY.ARPA
Thanks for the paper. I am still somewhat puzzled about the relation
between its approach and the one I took in my 1979. However, when
and if I return to the subject, I plan to look more closely at
Church's approach, because I think there should be concepts of
concepts.
∂01-Sep-85 0055 JMC re: Holy Communion, Transsubstantiation, and Vegetarian Roman Catholics (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - I believe that most of the corn produced in the U.S. is fed
to animals. This includes most of the corn we sell to foreign
countries and includes the corn we sell to the Soviet Union.
Except for Africa, the world food situation has considerably eased
in the last ten years. In particular, India has become self-sufficient.
∂01-Sep-85 1332 JMC Qlisp
To: RWW
CC: CLT
The bowdlerized edition of the proposal (ask Rutie) contains an updated
paper by RPG. The original joint paper is in the Proceedings of the 1984 Lisp
Conference.
∂01-Sep-85 2125 JMC Chudnovsky proposal
To: kahn@USC-ISI.ARPA
They don't have direct access to the ARPAnet and asked me to ask you
the status of their proposal. In my opinion they are excellent
mathematicians and have a good idea about how to improve algebraic
computation and increase its availability.
∂02-Sep-85 1251 JMC Re: Chudnovsky proposal
To: DAC
∂02-Sep-85 1238 KAHN@USC-ISI.ARPA Re: Chudnovsky proposal
Received: from USC-ISI.ARPA by SU-AI.ARPA with TCP; 2 Sep 85 12:38:02 PDT
Date: 2 Sep 1985 15:38-EDT
Sender: KAHN@USC-ISI.ARPA
Subject: Re: Chudnovsky proposal
From: KAHN@USC-ISI.ARPA
To: JMC@SU-AI.ARPA
Cc: kahn@USC-ISI.ARPA
Message-ID: <[USC-ISI.ARPA] 2-Sep-85 15:38:30.KAHN>
In-Reply-To: The message of 01 Sep 85 2125 PDT from John McCarthy <JMC@SU-AI.ARPA>
John,
I did read their proposal on the trip back from the
west coast and i agree with your assessment
based entirely on what they wrote. Im inclined to
fund it, but in any event, it wont get out of here before
amarel comes on board in a few weeks so the action will
probably fall on his watch. Ill be supportive.
bob
∂03-Sep-85 1654 JMC reply to message
To: MAZZETTI@SUMEX-AIM.ARPA
[In reply to message sent Tue 3 Sep 85 14:33:10-PDT.]
Thanks for Crocker address, but I've now heard from Allan Mackworth.
∂03-Sep-85 1709 JMC IJCAI Expenses
To: RA
∂03-Sep-85 1140 mack%ubc.csnet@csnet-relay.arpa IJCAI Expenses
Received: from CSNET-RELAY.ARPA by SU-AI.ARPA with TCP; 3 Sep 85 11:40:31 PDT
Received: from ubc by csnet-relay.csnet id a023372; 3 Sep 85 14:30 EDT
Date: Tue, 3 Sep 85 11:28:51 pdt
Received: by ubc.csnet id AA19516; Tue, 3 Sep 85 11:28:51 pdt
From: Alan Mackworth <mack%ubc.csnet@csnet-relay.arpa>
To: John McCarthy <jmc@SU-AI.ARPA>
Message-Id: <1992:mack@vision.ubc.cdn>
Subject: IJCAI Expenses
John,
Please itemize your travel and living expenses for IJCAI and send the
accounting to me. I'll then arrange for reimbursement. The award cheque
for $1000 will arrive separately.
I enjoyed your talk. The map coloring example is nice. Although map
colouring is technically "hard" your little algorithm makes it easy
on many (most?) real maps. It inspired me to think again about
the constraint satisfaction paradigm as it applies to maps -
a nice thesis in there somewhere. It also connected in with
Hector's talk about resource limits etc.
Alan
∂04-Sep-85 0918 JMC re: DAI workshop
To: SJG
[In reply to message rcvd 04-Sep-85 09:06-PT.]
Just one question. How much money do you need, and what do you need it for?
∂04-Sep-85 1020 JMC
To: CLT
Featherbed called to say your pillows are ready.
∂04-Sep-85 1148 JMC Workshop in distributed AI
To: aaai-office@SUMEX-AIM.ARPA
CC: SJG@SU-AI.ARPA, genesereth@SUMEX-AIM.ARPA,
phw%oz@MIT-MC.ARPA
I have approved $5K for a workshop in distributed AI proposed by
Mike Genesereth and Matt Ginsberg. They will be in touch directly.
∂04-Sep-85 1200 JMC test message
To: chard.yktvmv.ibm@CSNET-RELAY.ARPA
This is a test message to the Chudnovskys to see if mail works.
∂04-Sep-85 1229 JMC re: typing your notes
To: RA
[In reply to message rcvd 04-Sep-85 12:25-PT.]
Please move the files to SAIL, same names, in e85,jmc.
You can then delete them at SCORE.
∂04-Sep-85 2046 JMC message to mc
To: phw%oz@MIT-MC.ARPA
Did you get my message which I assumed would be forwarded from MC
about whether you want me to continue approving workshops?
∂05-Sep-85 0034 JMC re: From Wolfgang Bibel
To: tum%germany.csnet@CSNET-RELAY.ARPA
[In reply to message sent Wed, 4 Sep 85 17:16:21 -0100.]
Thanks for the invitation. Sept. 30 is the best date for me.
∂05-Sep-85 1006 JMC
To: CLT
We have $15,211.20 in the account + $5,000 overdraft protection.
∂05-Sep-85 1008 JMC
To: CLT
In Thomson-McKinnon there is a little over $1,000.
∂05-Sep-85 1145 JMC re: September Visit
To: AI.ELLIE@MCC.ARPA
[In reply to message sent Tue 3 Sep 85 10:11:53-CDT.]
Those reservations should be for Sunday and Monday nights, since I
will be at MCC on Monday and Tuesday.
∂06-Sep-85 1151 JMC
To: CLT
Stanford paycheck of $3201 was deposited today.
∂06-Sep-85 1211 JMC Centre Mondial
To: shortliffe@SUMEX-AIM.ARPA
I just talked to Dan Resnick, and the meeting of the Conseil Scientifique
scheduled for the 22nd has been cancelled. The Conseil is to be
reorganized and many of the foreign members to be dropped, most likely
including me; I didn't ask about you. The notification of all this
is to be mailed shortly!
∂06-Sep-85 1422 JMC parking
To: CLT
Betty called the Department of Public Safety. You can go there and buy
a sticker; they'll believe you. However, they have to give you a temporary
permit till you get the license plates.
∂06-Sep-85 1539 JMC test
To: spad.yktvmv.ibm@CSNET-RELAY.ARPA
This is a test. Reply to jmc@su-ai.arpa if you can.
∂06-Sep-85 1541 JMC
To: RPG
What are lisps for IBM-PC-AT?
∂06-Sep-85 1608 JMC
To: HST
We need to talk on the phone. Please MAIL telephone numbers.
∂06-Sep-85 1940 JMC
To: SMC
Susan Leder called. Call her. She doesn't have a spare horse.
∂06-Sep-85 2239 JMC please get copies of the reviews mentioned in the following
To: RA
∂09-Jul-85 2344 LIBRARY@SU-SCORE.ARPA Reviews of "Not in Our Genes."
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 9 Jul 85 23:44:28 PDT
Date: Tue 9 Jul 85 23:43:08-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: Reviews of "Not in Our Genes."
To: jmc@SU-AI.ARPA
Message-ID: <12125806251.30.LIBRARY@SU-SCORE.ARPA>
Dear John,
A reference librarian at the Green Library checked "Book Review Digest"
and "Book Review Index" for me. She did not check the New York Times database
as she thought that these indexes would cover the same ground. I also checked
the on-line version of "Book Review Index" (DIALOG file #137). These sources
list the following reviews for "Not in Our Genes":
America. v.151 (August 18, 1984), p.82
Booklist. v.80 (May 15, 1984), p.1279
Best Sellers. v.44 (August 1984), p.195
Choice. v.22 (September 1984), p.196
Humanist. v.44 (November 1984), p.39
Instructor. v.94 (September 1984), p.158
Kirkus Reviews. v.52 (March 1, 1984), p.242
Natural History. v.93 (August 1984), p.66
Nature. v.310 (July 19, 1984), p.255
Newsweek. v.103 (May 28, 1984), p.84
New York Review of Books. v.31 (August 16, 1984), p.30
New York Times Book Review. v.89 (May 20, 1984), p.9
Psychology Today. v.18 (August 1984), p.77
Publishers Weekly. v.225 (March 9, 1984), p.102
Wilson Quarterly. v.8 (August 1984), p.152
Women's Review of Books. v.2 (January 1985), p.7
None of these appears to be a Marxist publication. I could probably come up
with more non-Marxist reviews, but I believe it was particularly a "communist"
review that you were interested in. I will see if one of the librarians at
Hoover can help me and also check with our Russian and East European Curator.
- Richard
-------
∂07-Sep-85 1329 JMC The risks of not using some technology
To: risks@SRI-CSL.ARPA
The problem with a forum on the risks of technology is that
while the risks of not using some technology, e.g. computers, are
real, it takes imagination to think of them. A further problem
with newspaper, magazine and TV discussion of technology is that
journalists and free-lance writers tend to run in intellectual
mobs. This biases the discussion for everyone, especially when
the same journalists read each others writings and call it public
opinion. Here are some illustrations.
1. Suppose some organization manages to delay interconnecting
police data systems on some specious civil liberty grounds.
Suppose some wanted murderer is stopped for a traffic offense
but not arrested, because he is wanted in a jurisdiction not
connected to the computer system used by the police officer.
He later kills several more people. The non-use of computers
will not be considered as a cause, and no-one will sue the
police for not interconnecting the computers - nor will anyone
sue the ACLU. The connection will not even be mentioned in
the news stories.
2. No relative of someone killed on U.S. 101 during the 10 years
the Sierra Club delayed making it a freeway sued the Sierra Club.
3. No non-smoker who dies of lung cancer in an area newly polluted by
wood smoke will sue the makers of "Split wood not atoms" bumper
stickers.
***
Based on past experience, I expect this question to be ignored, but here's
one for the risk-of-computers collectors. Is a risk-of-computers
organization that successfully sues to delay a use of computers either
MORALLY or LEGALLY LIABLE if the delay causes someone's death? Is there
any moral or legal requirement that such an organization prove that they
have formally investigated whether their lawsuit will result in killing
people? As the above examples indicate, the present legal situation
and the present publicity situation are entirely unsymmetric.
***
Here's another issue of the social responsibility of computer
professionals that has been ignored every time I have raised it.
The harm caused by tape-to-tape batch processing as opposed to on-line
systems.
From the earliest days of commercial computing people have complained
about seemingly uncorrectable errors in their bills. The writers
don't know enough to connect this with the use of tape-to-tape
batch processing. Under such a system when a customer complains,
the person who takes the complaint fills out a form. A key puncher
punches the form on a card. At the next file-update, this card
goes to tape, and a tape-to-tape operation makes the correction.
If there is any error in the form or in the key punching, the
correction is rejected, and the customer gets the wrong bill again.
On-line systems permit the person who takes the complaint to make
the correction immediately. Any errors in making the correction
show up immediately, and the person can keep trying until he gets
it right or ask for help from a supervisor. Not only is the customer
better off, but the complaint-taker has a less frustrating job.
My own experience with the difference occurred in 1979 when my
wallet was stolen, and I had to tell American Express and Visa.
American Express had an on-line system, and the person who took
the call was even able to give me a new card number on the spot.
The Visa complaint-taker had to look it up on a micro-fiche file
and call back, and still they got it wrong. They gave me a new
account number without cancelling the old one.
Perhaps this issue is moot now, but I suspect there are still
many tape-to-tape systems or systems using modern equipment that
still emulate the old systems. Shouldn't computer professionals
who pretend to social responsibility take an interest in an
area where their knowledge might actually be relevant?
Once upon a time, beginning perhaps in the middle nineteenth
century, scientific organizations were active in pressuring
government and business to use new technology capable of
reducing risk and promoting the general welfare. I have in
mind the campaigns for safe water supplies and proper sewage
disposal. Here's a new one that involves computer technology.
Theft can be reduced by introducing the notion of registered
property. When you buy a television, say, you have the option
of buying a registered model, and the fact that it is registered
is stamped on it. Whenever someone buys a piece of used registered
property he has the obligation of telephoning the registry to
check whether the property with that serial number has been reported
stolen and recording his ownership. Repairmen are also obliged
to telephone either by voice or by keyboard.
Unfortunately, too many computer people imagine their social
responsibility to consist solely of imagining risks.
∂07-Sep-85 1629 JMC re: Files
To: BSCOTT@SU-SCORE.ARPA
[In reply to message sent Sat 7 Sep 85 16:00:27-PDT.]
Yes, it will be ok to bring some boxes down - a little at a time.
∂07-Sep-85 1646 JMC remarks on your report
To: pednault@SRI-AI.ARPA
CC: VAL@SU-AI.ARPA
I like your approach via logic, but it isn't as general as the
situation calculus, because the languages you use do not permit
sentences involving more than one situation. For example, you
cannot say that one situation is better than another. Nor can
you express the chess rule that it is not legal to castle if the
rook or king has moved. The latter can be gimmicked by introducing
additional no-argument predicates in your system, but this is
unintuitive. I don't see how you will deal with the former at all,
and comparison of situations is important for AI.
I think your system would be improved by introducing situation variables
and situation arguments to the predicates. It would permit expressing
some of your rules as sentences of the language. For example, you could
directly prove the theorem that block B must be moved onto C before A is
moved onto B.
I will be away Monday and Tuesday and again Friday, so the following
week will be more convenient for getting together than next week.
∂08-Sep-85 1108 JMC Hot rodding you AT and the weather
To: risks@SRI-CSL.ARPA
Assuming, contrary to fact, that putting a faster clock
in your AT would cause it to catch fire, the RISKS contributors are
quite wrong about who would be sued. According to the new legal
doctrine of bursae profundae, it isn't the poor BBOARD operator
that would have to pay but rich IBM. It wouldn't take much of a lawyer
to figure out that IBM should have anticipated that someone might
do this and warned against it or even somehow made it impossible.
Changing the subject, consider the effect of the court decision that
NOAA was liable for not fixing the buoy. Up to now weather predictions
have been offered as a non-guaranteed service with the user taking
responsibility for the extent to which he relies on it. The court
has said that this is no longer possible. Any institution with
"deep pockets" cannot safely offer information on a user responsibility
basis. What if Stanford University has negligently failed to
replace a stolen book from its medical library, and someone dies
who would have been saved had his doctor found the book? Stanford's
lawyers should advise Stanford to deny access to its medical library
to practicing physicians.
∂08-Sep-85 1115 JMC re: Risks of omissions
To: Nicholas.Spies@CMU-CS-H.ARPA
[In reply to message sent 8 Sep 1985 12:00-EST.]
I'll answer this to RISKS after yours comes out. For now it seems to
me that that the poster manufacturer, the ACLU, and the Sierra Club
could be guilty of negligence if it could be shown that they had
gone off half-cocked and hadn't made a good faith effort to determine
whether their efforts would have harmful side-effects. Something like
an internally used impact statement is at least morally required.
∂08-Sep-85 1306 JMC re: The risks of not using some technology
To: LIN@MIT-MC.ARPA
[In reply to message sent Sun, 8 Sep 85 15:51:44 EDT.]
I will reply at length after RISKS publishes your message and its fellows.
However, you miss the point of the passage you cite. That passage doesn't
deal with the moral and legal responsibility of a user of technology to
use modern technology. Rather it asks about the moral and legal responsibility
of those who sue to delay technology. In my opinion, any organization
that intervenes legally has at least the moral responsibility to prepare
a statement about the impact of their intervention, including the impact
of delays they may cause. Specifically, the Sierra Club had a responsibility
to attempt to estimate the likely effect on traffic fatalities of success
in their attempt to delay making U.S. 101 south of San Jose into a freeway.
They were successful, and I think the ten year delay killed some tens
of people. They might argue that it was worth it for some other reason,
but I don't believe they ever provided such an argument, because their
ideology holds only Government and corporations responsible for the
probable consequences of their acts. I believe that none of the "handbooks"
on how to sue "the system" mentions any such considerations.
∂08-Sep-85 1319 JMC re: The risks of not using some technology
To: LIN@MIT-MC.ARPA
[In reply to message sent Sun, 8 Sep 85 16:08:47 EDT.]
I rely in a simple-minded way on published statistics on the number killed
per hundred million miles of traveel on freeways and ordinary highways.
Presumably there are specific statistics on accidents that have occurred
on that highway before and after it was finally made a freeway. Back
in the days when the media were enthusiastic about freeways, these
statistics were published in the newspapers. My recollection is that
the freeways had about 2 fatalities per hundred million miles and ordinary
highways about 10. The literature of traffic engineering contains
doctrine about how to save the most lives per dollar. Before the
environmental movement became powerful, these considerations played
an important role, along with the politics of spending money in
legislators' districts, in determining what improvements were made.
Presumably, it still plays some role.
∂08-Sep-85 1343 JMC abstract
To: VAL
Here's another draft. While I am at MCC, I'll be checking and
answering my mail here.
halper.abs[e85,jmc] Draft abstract for Halpern conference on knowledge
Non-monotonic reasoning may require reifying states of knowledge.
The use of circumscription to solve the frame problem
is discussed in (McCarthy 1984) and? (Lifschitz?). The formalism
given in that paper isn't quite right, and trying to fix its difficulties
has led us to some interesting phenomena, which we will discuss in
this paper.
Here's the difficulty. In the axiomatization of that paper,
it is considered abnormal for one object to be on another. This results
in the desirable consequence that when abnormality is circumscribed
only those on(x,y,s) relations that can be inferred are considered
true. Now suppose we consider the new situation result(move(A, top(B)),s0).
If A and B inferred to be clear in the situtation s0, then we wish to
infer the action move(A,top(B)) succeeds. However, this makes
allows us to infer on(A,B,result(move(A,top(B)),s0)), which implies
an abnormality that can be avoided if the action is unsuccessful.
Therefore, circumscription of abnormality results in a disjunction,
whereas we prefer to conclude definitely that the action is successful.
There are several approaches.
We can minimize on(x,y,S0) with higher priority than
on(x,y,result(move(A,top(B)),S0)). This is what we want to happen,
but it isn't immediately clear how to formulate the axioms generally
so that it will.
It is tempting to give normality in an early situation
priority over normality in a later situation. This will probably
work in planning problems, but it isn't generally correct. Sometimes
we want to infer facts about the past from knowledge of the present.
In that case giving priority to early situations is wrong. The
answer seems to be for the system to give priority to the situations it
knows about over those whose properties it must infer from them.
Our formalism must then take into account not merely what it
believes but also how the different assertions came to be believed, e.g.
which were inferred from which others. We are then may be led
to reifying and formalizing some of the data taken into account
by ``reason maintenance systems''.
For this purpose we are considering using a ``mental situation
calculus'' involving mental states and mental actions. The full
paper will cover our results on this and other approaches.
∂09-Sep-85 1445 JMC
To: RA
I do indeed expect to be in Germany Sep 24 to 26. So tell him sorry.
∂10-Sep-85 0736 JMC
To: pednault@SRI-AI.ARPA
How about 16th after planlunch?
∂10-Sep-85 2114 JMC re: Qlisp funding
To: LES@SU-AI.ARPA, CLT@SU-AI.ARPA, RPG@SU-AI.ARPA,
Fateman@UCB-VAX.ARPA
[In reply to message from LES rcvd 10-Sep-85 15:01-PT.]
This looks like a reasonable way to proceed. So proceed.
∂10-Sep-85 2115 JMC re: monadic predicate logic
To: VAL
[In reply to message rcvd 10-Sep-85 15:00-PT.]
I'll have to talk to Richard, but I think I already told him about mine.
∂11-Sep-85 1559 JMC
To: RA
AUTHOR: Scarr, Sandra.
TITLE: Race, social class, and individual differences in I.Q. / Sandra Scarr
; commentaries by Leon J. Kamin, Arthur R. Jensen.
IMPRINT: Hillsdale, N.J. : Lawrence Erlbaum Associates, 1981.
xii, 545 p. : ill. ; 24 cm.
LOCATION: BF432.A1S3: Education;
∂11-Sep-85 1608 JMC re: siglunch
To: SJG
[In reply to message rcvd 11-Sep-85 15:57-PT.]
How about 11/15 its own self?
∂12-Sep-85 0057 JMC SDI
To: risks@SRI-CSL.ARPA
Some remarks of mine about SDI on Stanford BBOARD have been
referred to. For the benefit of non-readers of that BBOARD, they
mainly concerned whether I, like Chris Stuart, should use the IJCAI
platform to say something about it. I said nothing in my lecture,
but in my press conference, added to my remarks on AI, the remark
that there was no principle of computer science that says that
programs of any particular task cannot be written and debugged.
Not much interest was shown by the assembled press; there was
exactly one question on that point.
At the suggestion of Robert Jastrow, who is one of the main
scientific defenders of SDI, I made the same point in letters to
three Congressmen, said to be influential in the matter of SDI
appropriations.
Now I shall say my opinion about SDI.
1. If it can be done, it should. If it affords complete
protect, that's great, and if it affords partial protection, that's
good. The balance of terror is a bad thing. Here are answers to
some counter arguments to its desirability. (a) Joe Weizenbaum
says that it attempts a technological solution to a problem that
should be solved morally. Alas, moral progress has been so slow
that almost the only moral problems to be even partially solved
are those that can at least partially been turned into technological
problems. For example, the technology of contraception has greatly
reduced human unhappiness. (b) It is argued that the Soviets would
have to attack at the first sign of deployment. Every past imminent
advance by either side has in principle given the other side some
temptation to strike before it can be deployed. So far as we know,
neither side has even come close to giving in to such temptation.
One reason is that the effect of any advance is always subject to
a probabilistic estimate, so temporizing has always looked better
than attacking. Even if SDI works very well, it may be that no-one
will be able to be sure that it is that good.
However, most likely the main reason has been is
that neither side ascribes the very worst intentions to the other
with certainty. Each side has always said, "Perhaps they don't
actually mean to attack us. Why have a nuclear war for sure instead
of only a certain probability?" Anyway the Soviets have experienced
a period in which we had complete nuclear superiority and didn't
attack them.
2. My opinion is that if the physics of the problem permits a
good anti-missile defense the programs can be written and verified.
However, it will be quite difficult and will require dedicated work.
It won't be done by people who are against the whole project.
Computer checked proofs of program correctness will probably play
some role. So will anticipating what kind of bugs would be most
serious and putting the biggest effort into avoiding them. Having
many people go over and discuss all the critical parts of the
program will also be important.
∂12-Sep-85 0855 JMC abstract
To: VAL
I didn't get as far as I hoped. However, mental[s85,jmc] contains some notes,
and mental[e85,jmc] begins an abstract. More later today.
∂12-Sep-85 1312 JMC re: [LANSKY@SRI-AI.ARPA: Next week's PLANLUNCH -- notice change in date, place]
To: PEDNAULT@SRI-AI.ARPA
[In reply to message sent Thu 12 Sep 85 13:09:22-PDT.]
How about Monday at 11am here?
∂12-Sep-85 2327 JMC message from Stoyan
To: RA
I expect either a net message or a phone call from Stoyan in Germany.
If he phones, ask him to call me at Inference in L.A. 213 417-7997.
If there is a net message for me from him, please phone me there.
If he asks what reservations I have, they are in my calendar file.
∂12-Sep-85 2340 JMC re: Halpern's conference
To: VAL
[In reply to message rcvd 12-Sep-85 16:29-PT.]
You're probably right about the frame problem per se not being appropriate
for Halpern's conference. I'm confused about all these matters myself.
I don't see why expression evaluation is the problem; does this mean
determining the value of an expression? As I mentioned in my notes,
my present idea is to see if we can develop a notion of one mental
situation being "better" than another, so that a system could hill
climb in mental space to a solution of some problem, e.g. a block
stacking problem. The idea is that the mental space is richer than
physical space. For example, in mental space we can decide that it
is an advance to decide that a certain block might as well be moved
to the table immediately or (a la Pednault) that block A must be
put on block B after block B has been put on block C. These
decisions don't correspond to actions in physical space.
∂13-Sep-85 0730 JMC re: gwai@germany
To: GOERZ@SUMEX-AIM.ARPA, HST@SU-AI.ARPA
[In reply to message sent Fri 13 Sep 85 02:34:30-PDT.]
Your proposed schedule is ok. I will arrive in Frankfurt on
Wednesday on Panam 150 at 12:35pm and take the train to
Stuttgart. Better MAIL me a Stuttgart telephone number
in case the flight is late. I will get the ticket myself, and I'll
let you know the price soon, probably today. I guess being
reimbursed in dollars will be simplest. My present return flight
Panam 151 leaves Frankfurt at 2:30pm on Tuesday; is this a feasible
a schedule?
∂13-Sep-85 0740 JMC germany
To: RA
I have received one message from Stoyan. Please find out from Franklin
Hirsch the price of the round trip ticket and MAIL this to HST and to
me.
The reservations he made are ok.
∂13-Sep-85 2023 JMC re: Parallel Lisp
To: jefferso@LOCUS.UCLA.EDU
[In reply to message sent Fri, 13 Sep 85 17:05:39 PDT.]
The conclusion to which you jumped was correct. The language Qlisp, formerly
Qlambda, is the same as ordinary Lisp in data structures. It is described
in a paper by Gabriel and McCarthy given at the 1984 Lisp conference and
published by ACM in the proceedings of that meeting. I will send you a
report version of the paper.
∂13-Sep-85 2024 JMC please send Gabriel and McCarthy to
To: RA
David Jefferson
3531 Boelter Hall
UCLA
Los Angeles, Cal. 90024
∂13-Sep-85 2026 JMC phw
To: aaai-office@SU-SCORE.ARPA
Finger says that he hasn't logged in to MC since January, but I note
that he was addressed there in a recent executive committee message.
Is mail sent to him at MC forwarded to where he is? I asked him if
I should continue to approve workshops and have not received a reply.
∂13-Sep-85 2038 JMC phw
To: aaai-office@SUMEX-AIM.ARPA
Finger says that he hasn't logged in to MC since January, but I note
that he was addressed there in a recent executive committee message.
Is mail sent to him at MC forwarded to where he is? I asked him if
I should continue to approve workshops and have not received a reply.
∂13-Sep-85 2120 JMC re: Where to find an OCR?
To: CASBS@SU-CSLI.ARPA
[In reply to message sent Fri 13 Sep 85 10:05:35-PDT.]
There is a DEST OCR reader for typewritten documents at the Stanford Business
School attached to one of their DEC-20s. It works pretty well. Xerox PARC
has a Kurzweil reader that can in principle capable of book fonts. The time
I saw it, it wasn't working very well.
∂13-Sep-85 2122 JMC re: Need to Copy Slides FAST.
To: ALTMAN@SUMEX-AIM.ARPA
[In reply to message sent Fri 13 Sep 85 18:09:10-PDT.]
Try
Stanford Design Associates, 327-9398, make 35mm slides
Lois Thornhill
P.O. Box 60451, Palo Alto
∂13-Sep-85 2328 JMC cs306 TA
To: reges@SU-SCORE.ARPA
I need one and have no obvious candidate.
∂14-Sep-85 0917 JMC re: Your talk in Munich
To: freksa%germany.csnet@CSNET-RELAY.ARPA
[In reply to message sent Fri, 13 Sep 85 11:49:38 -0100.]
Unfortunately, I will have to delay my Munich talk until my next
visit.
∂14-Sep-85 1054 JMC ticket costs
To: HST
CC: RA
These prices aren't right, because, as agreed, I'm going Business
Class. My intent was to find out what is the price of the tickets
my travel agent will actually deliver on Monday, but there was
apparently a misunderstanding with my secretary. We'll find out
Monday. There will also be some incidental expense, e.g transportation
to the airport, but they won't amount to much.
∂14-Sep-85 1902 JMC clarification
To: risks@SRI-CSL.ARPA
It appears that in referring to Weizenbaum's views I was too concise in
characterizing them. I refer to the following statement of his:
This is not the place to air political and ideological positions. For
clarity's sake, I just want to add to the above that I believe it to be
necessary to the survival of us all that we come to some social, political,
cultural accommodation with the rest of the peoples of the world even when,
or especially when, they organize their societies differently than we do
ours. SDI is in the tradition of the great technological fixes which
appear to their authors to relieve them of the responsibility to confront
underlying human problems.
I think President Reagan is quite clear in his views of the "underlying
human problem" of relations with the Soviet Union. He regards it as an
"evil empire" that cannot be removed by force. To one who
takes this position, SDI is a good idea if it can be made to work.
The phrase "social, political, cultural accomodation with the rest of the
peoples of the world ... when they organize there societies differently ... .
would seem naive to him in two senses. In the first place, he doesn't
consider the Soviet and other communist Government as representative of
their peoples for the simple-minded reason that the people can't vote on
the policies of their countries, and therefore coming to an accomodation
with the Government isn't the same as coming to an accomodation with the
peoples. In the second place, he would consider the terms "social,
political, cultural accomodation" vague and subject to anti-democratic
interpretations. For example, Finland has come to an "accomodation"
that involves returning anyone who escapes from the Soviet Union
and not publishing news embarrassing to the Soviet Union in newspapers.
The Administration criticized the Book publishers' inclusion of only
political works critical of the U.S. in their Moscow exhibit and the
omission of all works of a conservative character. This is certainly
a failure to accomodate.
Reagan would also agree with Weizenbaum that more is required than
more military equipment. As an example, we have Radio Marti intended
to combat Castro's lies. In the direction of "accomodation", there
seem to be renewed negotiations on reviving exchanges.
I agree with the views I have ascribed to Reagan.
∂16-Sep-85 1007 JMC re: clarification
To: Neumann@SRI-CSL.ARPA
[In reply to message sent Mon 16 Sep 85 09:58:41-PDT.]
OK, although a remark by you in the digest saying that the topics had
gotten too political (perhaps even including the authors of rejected
messages so that the curious might inquire directly) might be appropriate.
∂16-Sep-85 1035 JMC re: cs306 TA
To: REGES@SU-SCORE.ARPA
[In reply to message sent Mon 16 Sep 85 09:18:47-PDT.]
Actually I had two or perhaps one and a half last year.
∂16-Sep-85 1405 JMC re: tickets to Germany
To: RA
[In reply to message rcvd 16-Sep-85 10:52-PT.]
No changes.
∂16-Sep-85 1426 JMC reply to message
To: CLT
[In reply to message rcvd 16-Sep-85 12:17-PT.]
One cad you probably can't help taking, but let's take the other also.
I take it you don't want to leave early and eat in SF. Do you have to
go home? If not, since I do, I'll pick you up in the Cadillac at
6:45. If yes, we can leave from home.
∂16-Sep-85 1428 JMC re: cs306 TA
To: REGES@SU-SCORE.ARPA
[In reply to message sent Mon 16 Sep 85 11:52:05-PDT.]
ok, I'll be away from tomorrow until next Wednesday. Carolyn Talcott
will be able to tell candidates what's involved.
∂16-Sep-85 1438 JMC re: trip to Germany
To: RA
[In reply to message rcvd 16-Sep-85 14:24-PT.]
I plan to be back next Tuesday. I might stay over till Wednesday
if there is a good reason but certainly not longer.
∂17-Sep-85 0055 JMC two students
To: VAL
will talk with you about research assistantship. They are Stephanie Singer
and Paul xxx. I have talked with them. If you think they're likely to
to be useful, you can offer them RA positions. I'll be back next
Wednesday.
∂17-Sep-85 0104 JMC abstract for knowledge conference
To: halpern.ibm-sj@CSNET-RELAY.ARPA
CC: VAL@SU-AI.ARPA
I don't know whether you are still planning to have me as an invited
speaker. But anyway here is an abstract of a submitted paper, which
is far from ready, but I think it will be interesting and new.
The paper may turn out to be joint with Vladimir Lifschitz. I'll
be away for a week.
Mental Situation Calculus
The situation calculus of (McCarthy and Hayes 1969) has mainly
been used to reason about states of the physical world, taking into
account the locations and physical properties of objects and admitting
such events as moving them. Analogously we can consider a {\it mental
situation calculus} (MSC) in which the situations include beliefs,
goals, intentions and other mental qualities, and the events include
inferring, observing, establishing goals and discharging them.
MSC has several motivations.
1. MSC involves reifying beliefs, and one of its basic
forms will be believes(<proposition>,ss) standing for the assertion
that the proposition is believed in mental situation ss. The formalism
allows for belief not to be closed under inference. In fact one of
the possible mental actions is to make an inference. Therefore, we
can describe in detail the circumstances under which we want our
system to make inferences.
2. Non-monotonic reasoning requires closer control over
inference than deduction, because of its tentative character.
Some problems that have recently arisen with blocks world axiomatizations
may require that circumscription be controlled in accordance with
with the pedigree of the system's objective beliefs and not merely
determined by what the beliefs are.
3. It looks like several useful methods of control of reasoning
can be accomplished by hill-climbing in mental situation space.
Besides reifying beliefs, MSC involves reifying goals and
partial plans for achieving them.
Depending on progress the paper to be presented will include
both general discussion of MSC and specific formalizations.
∂17-Sep-85 1013 JMC
To: VAL
CC: RA
I left your Stanford ID on your terminal; it was in my box.
∂17-Sep-85 1013 JMC
To: VAL
Now I remember; it's Paul Roberts.
∂17-Sep-85 1031 JMC
To: RA
Please maintain a supply of "Applications of Circumscription".
∂17-Sep-85 1039 JMC
To: bosack@SU-SCORE.ARPA
The esp, rover, and boise are all flakey.
∂20-Sep-85 0120 JMC (→17401 24-Sep-85)
To: "#___JMC.PLN[2,2]"
I will return from Germany on Tuesday, Sept. 24.
∂20-Sep-85 0128 JMC martha coulton
To: jrobinson@SRI-AI.ARPA
My phone file has Martha Coulton, 321-3707, 151 Seale,work 734-2244x6731.
∂20-Sep-85 0135 JMC
To: samuel@SU-SCORE.ARPA
No response from IBM, which I expected to support work.
∂20-Sep-85 0140 JMC histories
To: spar!hayes@DECWRL.ARPA
I haven't thought about histories in some time, but as I recall it, 3 would
come come closest to my opinion. I'll be back from Germany next Wednesday, and
I think it would be best to discuss it in person.
What are your telephone numbers?
∂24-Sep-85 0000 JMC Expired plan
To: JMC
Your plan has just expired. You might want to make a new one.
Here is the text of the old plan:
I will return from Germany on Tuesday, Sept. 24.
∂25-Sep-85 1132 JMC cs206
To: givan@SU-SUSHI.ARPA, laube@SU-SUSHI.ARPA
Let's get together either this afternoon, I suggest after 3, or tomorrow
at 11am concerning your TAing 306.
∂25-Sep-85 1210 JMC re: maps to new student lunch
To: MDIXON@SU-SUSHI.ARPA
[In reply to message sent Wed 25 Sep 85 11:57:56-PDT.]
I received the above but not the original message. When is this lunch?
∂25-Sep-85 1236 JMC re: e visitorsLa Fiesta
To: RA
[In reply to message rcvd 25-Sep-85 12:35-PT.]
17
∂25-Sep-85 1531 JMC
To: JMC
b.bgivan@lotsa,b.bowie@lotsa
∂25-Sep-85 1706 JMC reply to message
To: RA
[In reply to message rcvd 25-Sep-85 17:04-PT.]
As I understand it, Yoram will be at M.I.T. in the Fall and paid by them.
So far as I know, we don't give RA's to people who aren't working here.
∂26-Sep-85 0553 JMC re: Want to talk?
To: Rich.Thomason@C.CS.CMU.EDU
[In reply to message sent Thu 26 Sep 85 08:37:51-EDT.]
I have no commitments at any time that weekend. For definiteness
let me suggest Saturday the 19th at 11am followed by lunch. We
could meet at my office 356 Margaret Jacks Hall where I would be
from 10:30. There is a Stanford telephone outside downstairs
from which you could dial 7-4430, short for 497-4430. It is
50 feet logical south of the entrance in the opening through
which you can see the church. Home phone 857-0672. MAIL me
some phone numbers in case I should need to call.
I look forward to it, as I can think of several questions I would
like to discuss.
∂26-Sep-85 1143 JMC re: re-keying
To: HANRAHAN@SU-SCORE.ARPA, LES@SU-AI.ARPA
[In reply to message from HANRAHAN@SU-SCORE.ARPA sent Thu 26 Sep 85 11:21:39-PDT.]
The offices that have SM-2 keys now should continue to have the same keys.
Les will handle any questions that may arise.
∂26-Sep-85 2056 JMC This may be worthwhile, but I have a class.
To: LES, CLT
∂26-Sep-85 1048 @SU-SUSHI.ARPA:MAYR@SU-SCORE.ARPA talk by Marc Snir
Received: from SU-SUSHI.ARPA by SU-AI.ARPA with TCP; 26 Sep 85 10:44:56 PDT
Received: from SU-SCORE.ARPA by SU-SUSHI.ARPA with TCP; Thu 26 Sep 85 10:38:14-PDT
Date: Thu 26 Sep 85 10:18:47-PDT
From: Ernst W. Mayr <MAYR@SU-SCORE.ARPA>
Subject: talk by Marc Snir
To: su-bboards@SU-SCORE.ARPA, aflb.su@SU-SCORE.ARPA
Message-ID: <12146369201.9.MAYR@SU-SCORE.ARPA>
Seminar Announcement
Software Issues in the Design of Shared Memory MIMD Software
Speaker: Marc Snir
Location: MJH352
Time: Tuesday, Oct. 1, at 1:00pm
Abstract: The NYU Ultracomputer and the IBM RP3 machines are MIMD shared
memory parallel machines. Processors are connected to global memory by a
global packet switched inter-connection network; the network is actively
involved in processing.
These machines pay the cost of a complex interconnnction network in order
to avoid the burden of explicit resource allocation by the user. They are
intended to provide a close approximation to an ideal model of computation,
where accesses to shared memory are atomic, and can be concurrently done by
all processors.
In reality memory accesses are not atomic, and resource allocation overheads
cannot be ignored. The gap between the ideal model and the machine reality
can be narrowed by a mixture of compile time and run time optimization
techniques, and by new hardware structures.
We shall survey some specific problems and some of the techniques that can
be used to soften their impact. How to schedule memory accesses so that
resulting code is serializable.
How to provide atomic access to structures.
How to cache or prefetch shared data.
How to avoid excessive process fragmentation.
How to synchronize.
-------
∂27-Sep-85 1458 JMC
To: RA
Please prepare a bill to MAD for one hour consulting.
∂28-Sep-85 0143 JMC Carolyn's ID
To: bscott@SU-SCORE.ARPA, LES@SU-AI.ARPA
It is now more than a month since the effective date of her
employment. Tonight she was harassed in her office by Stanford
policemen demanding staff ID. Can't you move the system faster?
∂28-Sep-85 1231 JMC re: Offices
To: CLT
[In reply to message rcvd 28-Sep-85 12:27-PT.]
Seems reasonable. Talk to Les - also to VAL about his move.
∂28-Sep-85 1626 JMC Re: Carolyn's ID
To: CLT
∂28-Sep-85 1418 BSCOTT@SU-SCORE.ARPA Re: Carolyn's ID
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 28 Sep 85 14:18:11 PDT
Date: Sat 28 Sep 85 14:15:40-PDT
From: Betty Scott <BSCOTT@SU-SCORE.ARPA>
Subject: Re: Carolyn's ID
To: JMC@SU-AI.ARPA, LES@SU-AI.ARPA
cc: BSCOTT@SU-SCORE.ARPA
In-Reply-To: Message from "John McCarthy <JMC@SU-AI.ARPA>" of Sat 28 Sep 85 01:43:00-PDT
Message-ID: <12146936613.10.BSCOTT@SU-SCORE.ARPA>
I'm very sorry about this, John. We will process a separate request for
an ID for Carolyn on Monday.
Betty
-------
∂28-Sep-85 2001 JMC
To: LES
Skywood is closed Sun. & Mon. contrary to Yumyum.
∂29-Sep-85 1920 JMC re: working schedule
To: VAL
[In reply to message rcvd 27-Sep-85 16:09-PT.]
OK. Let me suggest that you set up and announce the seminar - let's
call it "Common sense and non-monotonic reasoning" before you leave
with the first meeting right after you return. Les has terminal you
could use while yours is being fixed, but since you're going on vacation,
maybe you'll prefer to wait till you return - unless it is convenient
to take it for a week.
∂30-Sep-85 1149 JMC re: take the afternoon off
To: RA
[In reply to message rcvd 30-Sep-85 11:04-PT.]
Yes, try to get Sarah.
∂30-Sep-85 1828 JMC
To: g.gorin@LOTS-A
tomorrow at 2:40?
∂01-Oct-85 0041 JMC re: CSD's file server (from SAIL's BBOARD)
To: OTHER-SU-BBOARDS@SU-AI.ARPA
jmc - I notice a large number of BBOARD inquiries for information about
CSD systems that ought to be addressed specifically to the CSD CF
administration, specifically to Len Bosack, the manager of these
facilities. This includes Fogelsong's inquiry about Labrea.
∂01-Oct-85 0043 JMC re: CSD's file server
To: FOGELSONG@SU-SUSHI.ARPA
[In reply to message sent Tue 1 Oct 85 00:27:43-PDT.]
jmc - I notice a large number of BBOARD inquiries for information about
CSD systems that ought to be addressed specifically to the CSD CF
administration, specifically to Len Bosack, the manager of these
facilities. This includes Fogelsong's inquiry about Labrea.