perm filename MAR.ME[LET,JMC] blob
sn#095077 filedate 1974-03-29 generic text, type T, neo UTF8
∂29-MAR-74 1734 1,YAW
further to below---the mans name was Bill Simpson.
∂29-MAR-74 1731 1,YAW
(1)you called me a couple of weeks ago about a potential grad student Id recommended---------do happen to remember off hand if he was admitted?
(2)could you give me a frank estimate on whats likely to happen to that ROK issue
of the AI journal-----and, in particular, to my paper? I ask because Ive got a much
improved version [!] and Id like if possible to get it published----i.e.elsewhere
if that AI issue is in fact being qiuietly abandoned.
∂22-MAR-74 1055 R,AJT
My `opus' is not intended for publication at all. It was just something
I scribbled in response to seeing your draft. I should appreciate a later
version if you've written one. a.
∂18-MAR-74 1706 AP,ME
I got the following inquiry. What are your comments?
Is WUTHERING HEIGHTS in the public domain? --Martin
∂17-MAR-74 0840 1,JBS
I HAVE BEEN LOOKING AT YOUR TEXT FOR "WUTHERING HEIGHTS".
IT SEEMS TO BE ACCESSABLE OVER THE NET
("TYPE WUTHER[LIB,DOC]") AND YET
IT CONTAINS NO COPYRIGHT STATEMENT.
GENERALLY IT IS FORBIDDEN TO DUPLICATE
A COPYRIGHTED WORK WITHOUT THE PERMISSION
OF THE COPYRIGHT HOLDER, EVEN INTO AN
AUTOMATED INFORMATION RETRIEVAL SYSTEM.
EVEN WHEN PERMISSION IS GIVEN THE HOLDER
USUALLY INSISTS THAT THE COPYRIGHT
STATEMENT BE INCLUDED IN THE REPRODUCTION
SO AS TO PROTECT THE COPYRIGHT, SINCE
THE COPYRIGHT LAWS REQUIRE THAT EVERY
COPY OF THE COPYRIGHTED MATERIAL INCLUDE
THE COPYRIGHT STATEMENT.
JOHN SAUTER
∂18-MAR-74 0851 S,WD
if bill fletcher
calls you
about a hand held
terminal,
give him an audience. it is a little marvel.
∂17-MAR-74 1551 1,TK
HI,
THE MAGIC INCANTATION IS:
CONNECT TO DM, LOG IN BY TYPING
LOG <RANDOM NAME>
THEN TYPE:
NETWRK<CR>
IT WILL TYPE INFINITEE GARBAGE
THEN SIT STILL FOR A MINUTE.
THEN TYPE
SUR<ALT MODE>
FOR SURVEY DAA
OR
SUM<ALT MODE>
FOR A SUMMARY OF TODAY'S DATA
OR
LONG<ALT>SUM⎇<ALT>
FOR THE LONGTERM SUMMARY OF SURVEYS.
IF YOU HAVE AY TROUBLE, JEFF CAN PROBABLY HELP YOU.
∂17-MAR-74 0602 network site CMUA
***** FTP mail from [A310AN02] (NEWELL)
To: John McCarthy
From: Newell
Date: 17 Mar 74 0905 EDT
John: Marvin sent a message which I assume you
received. Steve was here for a speech meeting 13-14 Mar. I am sure
that we have slid off the April date and don't know why he hasn't sent
a messge to that effect.
A.N.
∂16-MAR-74 1254 AP,ME
READ command should work now. Runs new E. Documentation is
available from the NEWS program, subject "BOOK".
∂15-MAR-74 1400 APG,DCL (reminder)
HOW ARE THINGS PROGRESSING ON THE SRA?-DAVID
∂15-MAR-74 1326 network site AI
DO YOU KNOW IF ANYONE HAS EXTENDED FLOYD'S COMPLETENESS RESULT, FOR
ASSIGNMENTS, TO THE CASE WHERE ASSIGNMENT TO ARRAYS IS PERMITTED?
(REF. FLOYD'S 1967 JACM ARTICLE, WHERE HE GIVES THE MOST GENERAL POSSIBLE
CONSEQUENT (WITH COMPLETENESS RELATIVE TO DEDUCTIVE SYSTEM D CONSTITUTING
THE CRITERION FOR GENERALITY) GIVEN AN ANTECEDENT AND AN ASSIGNMENT
TO A VARIABLE.)
β
∂13-MAR-74 1027 TST,TOB
I think we can get some money for hand/eye
and cart-related things. Is that ok?
When can we talk about it?
Tom
∂13-MAR-74 1007 FOO,DBA
LIGHT.COM[2,JMC] is LIGHT.RE3[2,JMC] with DBA's comments in the text.
∂12-MAR-74 1554 1,MLM
O.K. I contacted Minsky about this blunder. By way of explanation,
I have been completely unsuccessful in clarifying for myself lines of authority and
modes of protocol in this environment. I certainlydidn't think '
I should bother Minsky about this. Enough people had asked me about
this article, knowing I was working with Terry that I thought it would be
best made available publicly. That this was O.K. I inferred from the existence
here of Xeroxed copies of the article from Minsky himself with the indication
on them that the article was more or less ready for informal distribution. Apparently
I guessed the wrong file (with the help of some other people here) and obtained an
obsolete version. Apologies to all concerned, and I guess I'll have to
try harder on figuring out how to go about this kind of thing!
---- Mitch Model
∂12-MAR-74 1552 1,KNM
what do you want me to do with these linguistic people's folders. terry has
read them and said that he will talk to you about them.
carolyn taynai wants to know when you will be able to meet with S. Ward next
week in order to interview him for faculty opening. do you want to see him
here or could you meet him on campus for lunch or at erl. see needs to know
right away so that she can make bookings for him with knuth and with floyd.
∂12-MAR-74 1550 1,KNM
what audio visual equipment will you need for your talk march 20th at
santa clara? you are to be there guest for dinner at 6:15 at u of
santa clara faculty club before your presentaion.--kasee
∂11-MAR-74 1049 network site SRAI
Date: 11 MAR 1974 1051-PDT
From: NILSSON at SRI-AI
Subject: REVIEW
- - - -
YOUR LIGHTHILL REVIEW FOR AI JOURNAL LOOKS OK TO ME.
NILS
-------
∂11-MAR-74 0756 1,MM
aha. disregard my last comment, for the copy i read had ended at the mention of
PLANNER and CONNIVER.
∂10-MAR-74 2221 1,MM
The frame-paper advertised in sail's login message is unauthorized.
I would not mind circulating the draft, but it is too bad he did
not ask me, for the sail version is much older than MLM thinks,
and is not labelled DRAFT. I have MAIL-ed him the file
name of a more recent version, but he should
realize that one should ask permission to do such a thing, and
there may be some bad statements in his version that
I will regret having been circulated.
∂10-MAR-74 1914 network site AI
- - - - - - - - - - - -
03/10/74 22:09:48
From: MINSKY@MIT-AI
- - - - - - - - - - - -
Review pointed but perhaps too brief. I think one should go on to argue
that robot research contributed to psychology in many ways in Vision,
and is beginning to appear in cognitive psychology
texts, etc. Furthermore,
perple are beginning to go deep into the representation of knowledge, etc.
So I would claim ver| strong effects on psychology already.
Generally, the question is whether the review should try to dismiss lighthill
as too superficial and worthless to consider
for a detailed review, or to try to argue for the importance
of robotics in its own right. I'm not sure
what impression your review is supposed to leave
in the reader's mind.
∂6-MAR-74 2153 network site AI
- - - - - - - - - - - -
03/07/74 00:49:11
From: MINSKY@MIT-AI
- - - - - - - - - - - -
∂6-MAR-74 2152 network site AI
- - - - - - - - - - - -
03/07/74 00:47:53
From: MINSKY@MIT-AI
- - - - - - - - - - - -
have formulated hasty ill-considered objection to participating in
arpa aicommittee in near future.
Read minsky; to newell
∂04-MAR-74 1206 1,KNM
2-MAR-74 10:40:16,495
-------
Date: 2-MAR-74 1040-PDT
From: UK
Re: W.G.2.2
- - - -
ARE YOU GOING TO BE AT STUTTGART? I AM MEETING AUUDREY AT STRASGOURG ON
THE FRIDAY NIGHT, THEN WE ARE GOING TO HAVE A SUPER LUNCH. IT
WOULD BE SO VERY, VERY NICE FOR BOTH OF US TO SEE YOU AGAIN. I HAVE
A MAILBOX AT CMU (N980JL02). AND, PERHAPS TOO, WRITE TO 18
EBURY ST SO WE CAN CONFIRM NUMBERS. I SHALL HAVE MY CAR WITH ME FOR
TRANSPORT - LOVE OR SOMETHING - JOHN LASKI (AND AUDREY TOO BUT SHE
ISNT HERE)
-------
∂2-MAR-74 2023 network site CMUA
00100 TO: STEVE CROCKER
00200 FROM: A. NEWELL
00300 DATE: 2 MAR 74 2230 EDT
00400 RE: PROPOSED AI COMMITTEE & SITE VISITS
00500 COPIES: HART @ SRI-AI, MINSKY @ MIT-AI, PHW @ MIT-AI,
00600 JMC @ SU-AI
00700
00800 I have several reactions to your proposal of 27 Feb 74,
00900 including a mild counter proposal.
01000
01100 (1) I believe that an occasional or regular intensive review of
01200 the total research program of an AI Lab is a good thing. Your
01300 proposal to conduct this in an open and public fashion seems to
01400 me an improvement over exclusively private negotiations with a
01500 project manager, though it can not replace them. I like the
01600 inclusion of outside people. I especially like the inclusion of
01700 people from the other Labs, since this promotes a general state
01800 of awareness of what the Labs are doing. Finally, I like the
01900 notion of the total AI program being reviewed as well, and not
02000 just a Lab in isolation.
02100
02200 (2) I do believe (as you mentioned in a conversation to me) that
02300 one should be careful not to make too sharp a distinction
02400 between ARPA AI efforts that are "labs" and those that are not.
02500 The benefits of being forced to present a coherent account is
02600 not limited to the larger efforts -- though it does have special
02700 application to them.
02800
02900 (3) I believe there are severe problems in asking such a
03000 committee to evaluate a labortory (in the traditional sense of a
03100 site visit) and make recommendations about the future course of
03200 a laboratory's work. The main one is that you (ARPA) are not
03300 prepared to really make this a proposal review. Hence, double
03400 jeopardy and double organization occur -- in which a group is
03500 reviewed in this public fashion, then reviewed again at proposal
03600 time. This is not an idle problem. It has come up in Speech,
03700 where the SU Steering Committee evaluates, then ARPA evaluates
03800 (either just before or just after). In the present context, the
03900 recent intensive reviews and reorganizations of SRI-AI and
04000 MIT-AI imply that the proposed April review has a strong
04100 double-jeopardy aspect. (And if there are no implications of the
04200 April review, because the course of these labs is already set,
04300 then what game is being played?)
04400
04500 (4) You have charged the committee with two functions: to
04600 evaluate the Labs in their own terms; and to identify the
04700 potential places for more organized effort. Behind this latter
04800 are several (much discussed) issues. ARPA feels (and some of us
04900 share this feeling) that various amounts of organization add
05000 substantially to progress in research under the right
05100 conditions. It can serve to pose goals, it can serve to make
05200 efforts at various Labs add together, etc. A second issue is
05300 that ARPA sees itself as project oriented. A third issue is that
05400 the AI program area must acquire a "new look" of some kind,
05500 given its age within the ARPA community.
05600
05700 There is something odd about the proposed committee
05800 "discovering" areas ripe for organized effort. For one thing,
05900 committees never can perform this discovery function. They can
06000 assess assertions by others (ie, pass judgement). They can
06100 even explore nascent ideas (as in summer study groups -- though
06200 this committee will be much less than that). But they can't, on
06300 predetermined dates, sense that areas unnoted by the field
06400 (including themselves, priorly) are ripe.
06500
06600 In fact, one reason the proposal doesn't seem outrageous (and it
06700 didn't to me on first reading) is that we all pretty much know
06800 what the alternatives might be. Speech is already organized.
06900 Automatic Programming is in process of such organization
07000 (whether it is ripe or not). Vision-Robotics is ripe, in the
07100 sense that a coordinated program among the Labs, who have been
07200 carrying on a well developed program or research for some years,
07300 might yield dividends. And possibly Natural Language is ripe.
07400 There isn't much else, right at the moment. One of the reasons
07500 why not is that an area, to be ripe in the sense needed, must
07600 exist is good form in several Labs. Research that is
07700 idiosyncratic to a Lab (eg, the cognitive work at CMU) has not
07800 recruited a broad enough base to yield to organized effort. One
07900 certainly does not recruit that base by having a committee
08000 observe the potentialities, at least not among a research
08100 community that is creative enough so that most members have
08200 their own research directions (which reorient themselves
08300 continually, but not under direction). On the other hand, an
08400 area like Knowledge Acquisition, which is certainly hot within
08500 all the Labs, is probably not well-formed enough to gain much
08600 from organization (especially after Natural Language is factored
08700 out). Some might want to disagree on the particular example.
08800
08900 (5) You are trying to arrange a happening -- in which the AI
09000 field is to delineate its goals and set its sights for several
09100 years ahead (25?). I do not seem to be able to get a firm grip
09200 at the moment about whether I think that is a good thing to be
09300 tried. I do have some reactions to your mechanisms. You
09400 subscribe to the view that outsiders (your committee) are the
09500 appropriate vehicle to make the prognosis and take the critical
09600 look. I don't believe that is possible -- I take the standard
09700 scientist's view that no one except those immersed in the field
09800 can sense the field's pulse (while agreeing that these, too,
09900 might be biassed). The guild and elitist character of science
10000 has its vicious aspects, but it has its elements of truth. No
10100 one can speak for us.
10200
10300 (6) I am also concerned about the preparation of papers. You
10400 have done one very neat thing -- made the papers on a per-Lab
10500 basis, thus skirting the problem that we do not collectively
10600 seem to be capable of producing a reasonable overview. (On
10700 this, however, I think Feigenbaum's recent piece to ARPA is
10800 pretty good. In fact, his suggestions for action are a
10900 variation somewhere between yours and mine. Also, Nils Nilsson's
11000 coming piece to IFIPS looks like it will be very nice.) You are
11100 asking all of us to spend the next months writing overview
11200 papers and not doing other science. There is a view in ARPA,
11300 I know, that you-all do not place very much in the way of
11400 administrative burden on us (eg, see Lick's comment on the
11500 Fri-Sun demand for a list of research accomplishments). But
11600 since about Oct last year I have had ARPA-managment problems on
11700 the top of my stack continuously and with the present plan it
11800 will be true that for essentially 10 months I will have played
11900 project manager and science administrator, and not scientist.
12000
12100 (7) As you undoubtedly understand, the review mechanism is not
12200 simply a measurement device. It operates best when it
12300 encourages the organization to so shape its effort that the
12400 visit and review itself is a foregone conclusion -- that the Lab
12500 is proceeding to produce excellent and exciting science. In
12600 this respect your proposal to make the review occur in less than
12700 two months is distinctly suboptimal -- more a measurement than a
12800 shaper. Three to four months is a more suitable time, given
12900 that we are moving into an initial cycle.
13000
13100 (8) With the above background let me propose a revised version
13200 of your scheme. It is not totally responsive to all your
13300 demands, but it perhaps is a start.
13400
13500 (A) There will be an annual review of the ARPA AI program.
13600 This will occur by means of site visits at each
13700 laboratory. The charge to each laboratory is to
13800 present its scientific program in substantial technical
13900 depth. (Thus each visit might be two days long.) It
14000 would not be a total review of management, facilities,
14100 etc., these being left to review by project management.
14200 In particular, the line would not be drawn closely on
14300 ARPA-supported versus non-ARPA supported, but on AI
14400 versus non-AI science.
14500
14600 (B) The smaller AI efforts within ARPA could either be
14700 folded into visits at a site, or an additional general
14800 session could be held.
14900
15000 (C) An ad hoc Committee would be created each year to make
15100 the review. This consists of people from the non-ARPA AI
15200 community, from the rest of computer science, and from
15300 other sciences and technologies. There could be a fair
15400 number of these people, say 15. The Committee is to
15500 operate as consultants to ARPA-IPT. Their charge is to
15600 present ARPA a report, both on the state of the program
15700 as a whole and on the science being done at the
15800 individual laboratories. Their charge is not to
15900 approve a proposal for the future -- they are not a
16000 proposal review committee and the contractors do not
16100 prepare their proposals for this group. I favor their
16200 report consisting of a day (or half day) of private
16300 interaction with ARPA IPT management plus each member
16400 submitting an individual report. I see no need for
16500 consensus being reached by the Committee, so a single
16600 report seems wasted effort -- though perhaps I am wrong.
16700 Maybe there should be a brief joint communique. What
16800 the Committee members get for their pains (which are
16900 considerable) is a first rate scientific meeting on
17000 artificial intelligence.
17100
17200 (D) Members of the Labs would also go to the visits of each
17300 other's Labs -- one of the functions of the meetings
17400 being to produce communication. Also, the interchange
17500 amoung members of the Labs would be an important vehicle
17600 for revealing the nature of the approaches and the
17700 quality of the science at each place. I would figure on
17800 2-3 from each Lab -- there is no concern about exact
17900 numbers. I would not make these people part of the
18000 Committee.
18100
18200 (E) Each group being reviewed would be responsible for
18300 preparing whatever outlines and overviews seem suitable
18400 to help the Committee see the group's research whole.
18500 But no position paper is required, nor are the
18600 presentations to be written as papers (scientific papers
18700 belong to the scientific literature). One can establish
18800 high standards for the communicative efficiency to
18900 Committee, however.
19000
19100 (F) There is no phase locking of the budget and proposal
19200 cycle to this meeting, except that it is understood that
19300 the view of a Lab's operation provided at this review
19400 forms the basis for action, whenever the proposal is
19500 processed. The review probably should be located
19600 at an appropriate time relative to ARPA-IPT's planning
19700 cycle, since ideas for future directions are likely to
19800 emerge from these meetings.
19900
20000 (G) The first round should occur sometime about four months
20100 from agreement time, for my feeling is that all the Labs
20200 will take this review very seriously and a good deal of
20300 serious reflection and self-review will be generated by
20400 it. I know we will.
20500
20600 (H) In general the Labs agree that various substantive areas
20700 of research should be coordinated by having Steering
20800 Committees. Whether a specific substantive area (however
20900 identified) is ripe for such coordinated organization is
21000 clearly a question of joint consensus in the field and
21100 with ARPA-IPT. (If it is desired to promote such
21200 organization, one might think of inducements, such as
21300 that Committees have budgets, which can be spent for
21400 research to help the area out.) The nature of these
21500 committees and their charters is a function of the area,
21600 and is negotiated at initiation time (on the basis of a
21700 study or meeting of more-or-less depth). The Speech
21800 Understanding Research Group and its Steering Committee
21900 is not necessarily a model. What is agreed is that a
22000 Steering Committee develops and maintains an overview of
22100 the research in its area, that it deals with both the
22200 need for and the mechanisms to enforce cooperation and
22300 coordination among the several researches, and that it
22400 is responsive to assessing the state of progress and
22500 setting research goals and concrete expectations for its
22600 area. It is also agreed that a major purpose of so
22700 organizing areas is to formulate, launch and carry
22800 through coordinated assaults on selected scientific
22900 targets with relatively well specified goals and
23000 schedules.
23100
23200 (I) We (the AI Labs) agree that Vision and Natural Language
23300 are both at the stage where Steering Committees might
23400 prove valuable. We agree to set up committees to
23500 explore these possibilities. It is reasonable to expect
23600 an initial report on such organizational efforts by the
23700 time of the Review, given that it is four months off.
23800 This is probably not enough time for a Study Group,
23900 however, especially with the general Review pending.
24000 Whether other areas are ripe for any organization should
24100 be left until these two areas have been explored.
∂02-MAR-74 2004 APG,DCL
JOHN-DO YOU HAVE ENOUGH GOOD LETTERS ABOUT ME TO PROCEED
WITH THE SRA REQUEST?-DAVID
∂1-MAR-74 2055 network site AI
- - - - - - - - - - - -
03/01/74 23:56:31
From: PHW@MIT-AI
- - - - - - - - - - - -
I AGREE ON BOTH POINTS:
BLEDSOE WOULD BE GOOD
IT WOULD BE GOOD TO PUT IT OFF
∂1-MAR-74 2007 network site CMUA
***** FTP mail from [A310AN02] (NEWELL)
To: John McCarthy
From: Newell
Date: 1 Mar 74 2310 EDT
John: I am writing a reaction to Steve's proposal basically
accepting, but with a mild counter proposal. I hope to get it around
to you-all this weekend. Bledsoe is fine, as far as that goes.
A.N.
∂01-MAR-74 1540 1,KNM
1-MAR-74 08:01:23,363
Net mail from site BBN-TENEX rcvd at 1-MAR-74 08:01:20
Date: 1 MAR 1974 1102-EDT
From: FIELDS at BBN-TENEX
Subject: FRONT END MEETING
cc: Principal-Investigators
- - - -
THE FRONT END MEETING (MARCH 14-15 AT ARPA) WILL BEGIN AT
10:00 A.M. FOR THE CONVENIENCE OF TRAVELERS. POSITION
PAPERS SHOULD ARRIVE BY MARCH 7.
BEST
CRAIG FIELDS
-------
1-MAR-74 11:26:46,200
-------
Date: 1-MAR-74 1126-PDT
From: PIPES
Re: ACCOUNTING.FEB-1-28
- - - -
2-1-74 TO 2-28-74 ACCOUNTING DATA IS IN YOUR DIR AS FILENAME
ACCOUNTING.FEB-1-28. QUESTIONS TO <PIPES>.
-------
∂1-MAR-74 0649 network site AI
- - - - - - - - - - - -
03/01/74 09:44:36
From: MINSKY@MIT-AI
- - - - - - - - - - - -
bledsoe is fine. suggest feynman on committee if he would.