perm filename W83.OUT[LET,JMC]1 blob
sn#705067 filedate 1983-04-01 generic text, type C, neo UTF8
COMMENT ⊗ VALID 00213 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00018 00002
C00019 00003 ∂03-Jan-83 1557 JMC
C00020 00004 ∂03-Jan-83 1657 JMC industrial professorship
C00021 00005 ∂03-Jan-83 2330 JMC
C00022 00006 ∂03-Jan-83 2331 JMC probably redundant message
C00023 00007 ∂04-Jan-83 1123 JMC
C00024 00008 ∂04-Jan-83 1204 JMC
C00025 00009 ∂04-Jan-83 1323 JMC Santa Cruz course
C00026 00010 ∂04-Jan-83 1426 JMC Santa Cruz course
C00027 00011 ∂06-Jan-83 2212 JMC
C00028 00012 ∂07-Jan-83 1143 JMC name spellings
C00029 00013 ∂07-Jan-83 1612 JMC
C00030 00014 ∂07-Jan-83 1649 JMC
C00031 00015 ∂07-Jan-83 1721 JMC change to turing.tex[w83,jmc]
C00032 00016 ∂09-Jan-83 2359 JMC
C00033 00017 ∂10-Jan-83 0839 JMC
C00034 00018 ∂10-Jan-83 1325 JMC
C00035 00019 ∂11-Jan-83 1036 JMC
C00036 00020 ∂11-Jan-83 2358 JMC
C00037 00021 ∂12-Jan-83 0040 JMC
C00038 00022 ∂12-Jan-83 0106 JMC phil-sci file
C00039 00023 ∂12-Jan-83 1451 JMC
C00040 00024 ∂12-Jan-83 1722 JMC shuttle
C00041 00025 ∂13-Jan-83 2303 JMC
C00044 00026 ∂14-Jan-83 0112 JMC
C00047 00027 ∂14-Jan-83 0943 JMC on TTY1 0943 can't open message
C00048 00028 ∂14-Jan-83 0954 JMC consensus theory of truth
C00050 00029 ∂14-Jan-83 1331 JMC Can't open
C00051 00030 ∂14-Jan-83 1456 JMC heat and temperature
C00052 00031 ∂14-Jan-83 1509 JMC industrial lectureship
C00053 00032 ∂14-Jan-83 1549 JMC Consensus theory of truth
C00056 00033 ∂15-Jan-83 0002 JMC
C00057 00034 ∂15-Jan-83 1426 JMC correspondence theory of truth
C00066 00035 ∂15-Jan-83 1430 JMC
C00067 00036 ∂15-Jan-83 1723 JMC correspondence model of truth
C00074 00037 ∂15-Jan-83 1729 JMC
C00075 00038 ∂15-Jan-83 1912 JMC consensus theory of truth and Solomonoff et al.
C00079 00039 ∂15-Jan-83 2312 JMC philosophy of science discussion
C00080 00040 ∂15-Jan-83 2332 JMC "Obstacles-and-Roofs" Worlds
C00081 00041 ∂15-Jan-83 2333 JMC "Obstacles-and-Roofs" Worlds
C00082 00042 ∂16-Jan-83 1154 JMC correspondence theory of truth
C00092 00043 ∂16-Jan-83 1203 JMC
C00093 00044 ∂16-Jan-83 1657 JMC theories of meaning
C00095 00045 ∂16-Jan-83 1703 JMC mail
C00096 00046 ∂16-Jan-83 1814 JMC Message of 15-Jan-83 20:26:24
C00098 00047 ∂16-Jan-83 2122 JMC keeping file of reports
C00099 00048 ∂17-Jan-83 0051 JMC verificationism
C00102 00049 ∂17-Jan-83 1144 JMC verificationism and correspondence
C00103 00050 ∂17-Jan-83 1352 JMC
C00104 00051 ∂17-Jan-83 1423 JMC Lakatos and Solomonoff
C00106 00052 ∂17-Jan-83 1643 JMC Sten-Ake
C00107 00053 ∂17-Jan-83 2157 JMC
C00108 00054 ∂17-Jan-83 2234 JMC Correspondence theory of truth and meta-epistemology
C00113 00055 ∂18-Jan-83 1204 JMC Correspondence theory of truth
C00115 00056 ∂18-Jan-83 1500 JMC correspondence theory
C00117 00057 ∂18-Jan-83 2156 JMC previous committment
C00118 00058 ∂18-Jan-83 2231 JMC Correspondence theory
C00121 00059 ∂19-Jan-83 0828 JMC
C00122 00060 ∂19-Jan-83 1509 JMC
C00123 00061 ∂19-Jan-83 1548 JMC
C00125 00062 ∂19-Jan-83 1551 JMC
C00126 00063 ∂19-Jan-83 1753 JMC
C00127 00064 ∂19-Jan-83 2321 JMC Lakatos review, Putnam, and Solomonoff (or even Solomonov)
C00135 00065 ∂19-Jan-83 2351 JMC
C00136 00066 ∂20-Jan-83 1123 JMC
C00139 00067 ∂21-Jan-83 0107 JMC printing mathfile equations
C00140 00068 ∂21-Jan-83 0111 JMC
C00141 00069 ∂21-Jan-83 1848 JMC correspondence theory
C00142 00070 ∂21-Jan-83 2249 JMC
C00148 00071 ∂21-Jan-83 2258 JMC
C00149 00072 ∂22-Jan-83 1239 JMC
C00153 00073 ∂22-Jan-83 1241 JMC
C00154 00074 ∂22-Jan-83 1258 JMC arpa
C00155 00075 ∂22-Jan-83 1301 JMC New ARPA proposal.
C00157 00076 ∂22-Jan-83 1502 JMC Marconi
C00163 00077 ∂23-Jan-83 0014 JMC
C00165 00078 ∂23-Jan-83 0114 JMC
C00169 00079 ∂24-Jan-83 0111 JMC objective physical and mathematical worlds
C00177 00080 ∂24-Jan-83 1415 JMC "your version of reality"
C00181 00081 ∂24-Jan-83 1426 JMC patience with GAVAN
C00183 00082 ∂24-Jan-83 1508 JMC correspondence theory
C00186 00083 ∂25-Jan-83 1131 JMC
C00187 00084 ∂25-Jan-83 1509 JMC correspondence theory, misunderstanding thereof
C00189 00085 ∂25-Jan-83 1817 JMC Ad nauseum
C00190 00086 ∂25-Jan-83 1858 JMC afghanistan
C00192 00087 ∂26-Jan-83 0037 JMC letters
C00193 00088 ∂26-Jan-83 1116 JMC isaacson%isi,phil-sci%oz/cc
C00194 00089 ∂26-Jan-83 1121 JMC intuitionism
C00195 00090 ∂26-Jan-83 1156 JMC
C00196 00091 ∂26-Jan-83 1248 JMC Common Lisp proposal
C00197 00092 ∂26-Jan-83 1351 JMC
C00198 00093 ∂26-Jan-83 1824 JMC
C00199 00094 ∂27-Jan-83 1309 JMC
C00200 00095 ∂27-Jan-83 1528 JMC
C00201 00096 ∂27-Jan-83 1557 JMC
C00202 00097 ∂27-Jan-83 1756 JMC world library
C00203 00098 ∂27-Jan-83 1929 JMC
C00204 00099 ∂28-Jan-83 2333 JMC sentences
C00206 00100 ∂29-Jan-83 1405 JMC typology
C00208 00101 ∂29-Jan-83 1523 JMC
C00209 00102 ∂29-Jan-83 1627 JMC innateness, sentences, etc.
C00217 00103 ∂29-Jan-83 1649 JMC proposal
C00218 00104 ∂29-Jan-83 2111 JMC protection?
C00220 00105 ∂30-Jan-83 1233 JMC
C00226 00106 ∂30-Jan-83 1240 JMC my error
C00227 00107 ∂30-Jan-83 1515 JMC
C00228 00108 ∂30-Jan-83 1657 JMC innateness
C00230 00109 ∂30-Jan-83 2216 JMC There you go again, Gavan.
C00231 00110 ∂31-Jan-83 1818 JMC CORRESPONDENCE, etc. and meta-epistemology again
C00259 00111 ∂31-Jan-83 2030 JMC narrowness
C00260 00112 ∂31-Jan-83 2350 JMC criticism of coherence and consensus
C00263 00113 ∂01-Feb-83 0040 JMC
C00264 00114 ∂01-Feb-83 1040 JMC Paper
C00265 00115 ∂01-Feb-83 1254 JMC Ascribing ...
C00266 00116 ∂01-Feb-83 1615 JMC
C00267 00117 ∂01-Feb-83 1845 JMC
C00268 00118 ∂01-Feb-83 2213 JMC checking circuit
C00269 00119 ∂01-Feb-83 2252 JMC library
C00270 00120 ∂01-Feb-83 2341 JMC loneliness
C00271 00121 ∂02-Feb-83 0135 JMC
C00272 00122 ∂02-Feb-83 1440 JMC messages
C00274 00123 ∂02-Feb-83 1453 JMC
C00275 00124 ∂02-Feb-83 1455 JMC
C00276 00125 ∂02-Feb-83 1505 JMC
C00277 00126 ∂02-Feb-83 1607 JMC
C00278 00127 ∂02-Feb-83 1732 JMC Thanks.
C00280 00128 ∂02-Feb-83 1828 JMC
C00281 00129 ∂03-Feb-83 0051 JMC Heat and temperature
C00283 00130 ∂03-Feb-83 0125 JMC
C00284 00131 ∂03-Feb-83 1147 JMC phil-sci
C00285 00132 ∂03-Feb-83 1450 JMC
C00286 00133 ∂03-Feb-83 1908 JMC Does Robby need the predicate true?
C00291 00134 ∂03-Feb-83 2301 JMC
C00292 00135 ∂03-Feb-83 2307 JMC
C00293 00136 ∂04-Feb-83 1441 JMC Job available
C00294 00137 ∂04-Feb-83 2105 JMC heat
C00296 00138 ∂05-Feb-83 1719 JMC unique factorization
C00299 00139 ∂05-Feb-83 1725 JMC "knowledge" is useful
C00307 00140 ∂05-Feb-83 1933 JMC
C00308 00141 ∂05-Feb-83 2039 JMC
C00310 00142 ∂05-Feb-83 2039 JMC this is a test
C00313 00143 ∂05-Feb-83 2109 JMC
C00318 00144 ∂06-Feb-83 0208 JMC Tarski, si! Tarsky, no!
C00322 00145 ∂06-Feb-83 0217 JMC
C00323 00146 ∂06-Feb-83 0256 JMC fixing garbles
C00325 00147 ∂06-Feb-83 2346 JMC fixing garbles
C00327 00148 ∂07-Feb-83 0013 JMC Philosophy's debt to science
C00332 00149 ∂07-Feb-83 0131 JMC
C00333 00150 ∂07-Feb-83 1801 JMC I also once enumerated the 2 by 2 Turing machines, though not as
C00337 00151 ∂07-Feb-83 2251 JMC possible worlds and elaboration tolerance
C00344 00152 ∂08-Feb-83 1230 JMC
C00345 00153 ∂08-Feb-83 1231 JMC
C00346 00154 ∂08-Feb-83 2300 JMC
C00353 00155 ∂08-Feb-83 2312 JMC Wessels
C00354 00156 ∂08-Feb-83 2316 JMC
C00355 00157 ∂08-Feb-83 2317 JMC
C00356 00158 ∂08-Feb-83 2320 JMC
C00363 00159 ∂09-Feb-83 1511 JMC JMC si! JCM no!
C00364 00160 ∂09-Feb-83 2330 JMC "God's eye view"
C00365 00161 ∂10-Feb-83 0030 JMC
C00366 00162 ∂10-Feb-83 0105 JMC Book Report
C00384 00163 ∂10-Feb-83 0212 JMC Addendum to book report
C00386 00164 ∂10-Feb-83 1852 JMC
C00387 00165 ∂11-Feb-83 1114 JMC Course for IBM
C00388 00166 ∂12-Feb-83 1214 JMC
C00389 00167 ∂13-Feb-83 1939 JMC
C00390 00168 ∂14-Feb-83 1103 JMC
C00391 00169 ∂14-Feb-83 1452 JMC
C00392 00170 ∂14-Feb-83 1607 JMC
C00393 00171 ∂16-Feb-83 1654 JMC
C00394 00172 ∂16-Feb-83 1911 JMC Computers & Standards--new journal
C00398 00173 ∂16-Feb-83 1913 JMC
C00399 00174 ∂17-Feb-83 1656 JMC interested in your new paper.
C00401 00175 ∂17-Feb-83 1819 JMC
C00402 00176 ∂18-Feb-83 1443 JMC tentative declination of invitation
C00404 00177 ∂18-Feb-83 1515 JMC
C00405 00178 ∂18-Feb-83 1700 JMC mailing list
C00406 00179 ∂18-Feb-83 1706 JMC
C00407 00180 ∂19-Feb-83 1557 JMC
C00408 00181 ∂26-Feb-83 1740 JMC
C00409 00182 ∂27-Feb-83 1502 JMC
C00410 00183 ∂27-Feb-83 2311 JMC Lowell return
C00411 00184 ∂27-Feb-83 2350 JMC
C00412 00185 ∂28-Feb-83 1448 JMC
C00413 00186 ∂28-Feb-83 2123 JMC
C00415 00187 ∂01-Mar-83 0126 JMC texts
C00416 00188 ∂02-Mar-83 1638 JMC Arden House conf.
C00417 00189 ∂02-Mar-83 1759 JMC
C00418 00190 ∂04-Mar-83 0059 JMC
C00420 00191 ∂04-Mar-83 1534 JMC
C00421 00192 ∂04-Mar-83 1535 JMC
C00422 00193 ∂07-Mar-83 1350 JMC time-sharing history
C00425 00194 ∂07-Mar-83 1429 JMC phone call
C00426 00195 ∂07-Mar-83 1435 JMC
C00427 00196 ∂07-Mar-83 1448 JMC
C00428 00197 ∂07-Mar-83 1455 JMC New York abstract
C00430 00198 ∂07-Mar-83 2200 JMC
C00431 00199 ∂08-Mar-83 0044 JMC
C00432 00200 ∂08-Mar-83 1107 JMC 3330 disk drives available
C00433 00201 ∂08-Mar-83 1820 JMC
C00438 00202 ∂08-Mar-83 1822 JMC
C00462 00203 ∂08-Mar-83 1824 JMC
C00481 00204 ∂09-Mar-83 1027 JMC
C00482 00205 ∂10-Mar-83 0222 JMC lunch on Tuesday
C00483 00206 ∂11-Mar-83 1021 JMC
C00484 00207 ∂11-Mar-83 1136 JMC
C00485 00208 ∂11-Mar-83 1150 JMC
C00486 00209 ∂11-Mar-83 1529 JMC
C00487 00210 ∂11-Mar-83 2127 JMC
C00488 00211 ∂13-Mar-83 1733 JMC LOTS
C00494 00212 ∂15-Mar-83 1105 JMC
C00495 00213 ∂15-Mar-83 1328 JMC
C00497 ENDMK
C⊗;
∂03-Jan-83 1557 JMC
To: DFH
Tom Bower, The perceptual world of the child., Fontana
Development in infancy
article: The object in the world of the infant, Scientific American 1971
∂03-Jan-83 1657 JMC industrial professorship
To: bmoore at SRI-AI
CC: GHG at SU-AI
Let me remind you that I need a course description by Feb. 1.
∂03-Jan-83 2330 JMC
To: mccarthy at SRI-AI
test
∂03-Jan-83 2331 JMC probably redundant message
To: bmoore%sri-ai at SU-SCORE
I need a course description by Feb. 1.
∂04-Jan-83 1123 JMC
To: RAC
See bboard for the room number of the first meeting on Jan. 13 at 1:15.
∂04-Jan-83 1204 JMC
To: DFH
I need a second phone directory.
∂04-Jan-83 1323 JMC Santa Cruz course
To: DFH
Please phone Steve Mandel and tell him that Mike Genesereth and
I have agreed to teach his course in AI next summer. Give him
Mike's phone number.
∂04-Jan-83 1426 JMC Santa Cruz course
To: csd.genesereth at SU-SCORE
I have transmitted our agreement and given Robert Stevens 408 425-1503
your number.
∂06-Jan-83 2212 JMC
To: DFH
Diana, that's spelled Elliott Bloom for future reference.
∂07-Jan-83 1143 JMC name spellings
To: DFH
Also it's Tom Connolly, not Connelly. I mention the spellings only, because
I communicate with both Connolly and Bloom regularly, so we might as well
go along with how they spell their names.
∂07-Jan-83 1612 JMC
To: csd.armer at SU-SCORE
Both Feb 10 and Feb 15 are ok. Do you have any information yet on
what our volume of report material is?
∂07-Jan-83 1649 JMC
To: YOM
It's now turing.tex[w83,jmc]. It looks fine; print it.
∂07-Jan-83 1721 JMC change to turing.tex[w83,jmc]
To: YOM
I have included after the title a note that it was reprinted from
Automata Studies. Before you print it, could you TEXorate it so
that the note prints reasonably?
∂09-Jan-83 2359 JMC
To: llw at S1-A
I will endeavor to give satisfaction.
∂10-Jan-83 0839 JMC
To: DFH
I will be free for app't with people from Livermore tomorrow by 2:45.
∂10-Jan-83 1325 JMC
To: nilsson at SRI-AI
The following message in its entirety was received:
∂10-Jan-83 1303 Nilsson@SRI-AI (SuNet) test
Mail-from: ARPANET site SRI-AI rcvd at 10-Jan-83 1103-PST
Date: 3 Jan 1983 1503-PST
From: Nilsson at SRI-AI
Subject: test
To: csd.mccarthy at SCORE
cc: Nilsson
-------
∂11-Jan-83 1036 JMC
To: csd.armer at SU-SCORE
At 50 lines of 80 bytes per page this comes to about 100 megabytes,
which we can certainly afford - even immediately. I suggest the creation
of directory CSDREPORTS at SCORE maintained by the reports staff. There
should b a notic inviting all writrs of rports to tell how to get
their reports to the area. It looks like the letter E broke on
my Datamedia for about a minute. We'll worry about uniformizing the
format in committee.
∂11-Jan-83 2358 JMC
To: gavan at MIT-MC
Thanks for sending me the remarks by Batali and your comments. I'm
inclined to react to them, but I would like to know a bit more about
the background. Is there a regular phiosophy of science computer
mailing list in which such discussions are carried out? If so,
I'd like to be on it, at least for a while, with the name JMC-LISTS@SU-AI.
Also, how big is it, and approximately who is on it?
∂12-Jan-83 0040 JMC
To: gavan at MIT-MC
1. In principle, willing. 2. (human-name 'jcma)?
∂12-Jan-83 0106 JMC phil-sci file
To: gavan at MIT-MC
I would be glad to have it ftped to the file
phil[jnk,jmc].
∂12-Jan-83 1451 JMC
To: DFH
Print JANLER.COM[W83,JMC] and include it with Janlert paper.
∂12-Jan-83 1722 JMC shuttle
To: pourne at MIT-MC
Dear Jerry:
I sent L-5 a $30 membership or something like that, and they
sent me a print of a painting of the Shuttle rising above the clouds
by K. Poor. What worries me is the painting in which the Shuttle seems
to be at an angle of about 50 degrees while at an altitude of 50,000
feet. Shouldn't it still be vertical at that altitude? I fear that
the range safety officer would have to destroy it if it weren't
still going straight up.
By the way, is the Citizen's Council on Space now defunct?
Best Regards,
John
∂13-Jan-83 2303 JMC
To: gavan at MIT-MC, phil-sci%mit-oz at MIT-MC
"I'm tempted to ask just what TRUTH is and what makes you think there's
any such thing, but that's a little off the subject. Maybe not. To
me, there is no truth but only consensus. What we call "truth" is
only what we have agreed upon, given certain conventions which we agree
are "rational." It seems to me that the notion of coming to a consensus
brings us back to the problem which motivated the discussion. How do we
in society and mental agents in a society-of-mind consensually validate
our beliefs and theories?"
The above quote from Gavan strikes me as muddled and scientifically
unpromising. It is similar to the Vienna circle ideas of the 1920s.
A young graduate student named Kurt Godel attended the Circle meetings
and had a different idea. His idea was that truth was one thing conceptually
and what you could prove was another. For his PhD thesis he proved that
in the case of first order logic the two coincided. Later he was able
to show that in the case of the arithmetic of Principia Mathematica and
related systems they could not coincide. Still later he was able to
show that the continuum hypothesis could not be disproved from the
Godel-Bernays axioms of set theory while maintaining his belief that
the continuum hypothesis is false. Another young man named Alfred Tarski
was able to show around 1930 that truth in arithmetic was not arithmetically
definable.
In my opinion, a person who makes a clear distinction between
truth and what is "consensually validated" will have a better chance of
advancing philosophy and/or artificial intelligence than someone who
muddles them. He might, for example, be able to show that the notions
coincide in some cases and differ in others.
∂14-Jan-83 0112 JMC
To: phil-sci%mit-oz at MIT-MC
Here is an example of a possible payoff from distinguishing truth
from consensus. Suppose we try to develop a formal "meta-epistemology"
based on a dynamical system (system evolving in time according to your
favorite formalism) called the "world" and a distinguished subsystem
called the "scientist". We suppose that certain functions of the
"scientist" are to be interpreted assertions about the "world".
We can study the effects of various "scientific strategies" for
finding information about the world. Some "worlds" are more knowable
than others. Some strategies are more effective than others. For
example, a realist might hope to prove that a strategy that confined
itself to relations among sense data would never learn certain facts
about the world that more liberal ontology could discover. We might
even be able to prove that a "scientists" using consensual notions
of "truth" would be unable to formulate certain truths. Probably,
before such a formal meta-epistemology can be developed, it will
be necessary to find a simpler yet relevant system to study.
To summarize, in order to model scientific and other knowledge seeking
activity, it will be necessary to distinguish what is true in
the model world from what the model scientist believes.
My previous message was imprecise. Godel proved that in first
order logic, validity (truth in all models) coincides with provability.
Even to formulate the result required keeping the concepts distinct. I
should mention, however, that van Heijenoort informed me that Hilbert and
Ackermann formulated the problem in their book on logic even though
Hilbert's philosophy was nominally formalistic.
∂14-Jan-83 0943 JMC on TTY1 0943 can't open message
To: BUG-E
The "can't open file" message gives no indication of what one's options
are. Thus after getting this message after trying to write \BBOARD, I
wanted to take my valuable comments to another file to try again later.
There was no hint of how this might be done and I had to quit. The
writeup in E of the improvement gives no hint either.
∂14-Jan-83 0954 JMC consensus theory of truth
To: gavan at MIT-MC, dam at MIT-MC,
phil-sci%mit-oz at MIT-MC
(If our mailer permits a REPLY-TO field, I don't yet know how to use it).
I am, as I suppose you suspect, an adherent of the correspondence theory of
truth, both within mathematics and outside it. Certainly there are differences
between mathematics and the common sense world, and I expect to address these.
DAM seems not to have understood that my meta-epistemology proposal
involved the correspondence theory. While the "scientist" in that proposal
can only learn about the "world" through his senses, we mathematicians
can study the correspondence between what he believes and what is true
of that "world". We can study what correspondences are possible and
what strategies achieve them.
∂14-Jan-83 1331 JMC Can't open
To: ME
Perhaps message should be
"Can't open file now. Try ⊗XSEND to other users or ⊗XCANCEL before moving your
changes".
∂14-Jan-83 1456 JMC heat and temperature
To: kdf at MIT-MC
Can you give me a complete reference to "When heat and temperature were one"?
I am interested in developing "ambiguity tolerant" formalisms, so I would
expect that some correct reasoning can be done even while confused
about basic matters, e.g. "If you add enough (heat or temperature) any
substance will become gas". If the paper is local to M.I.T., could you
transmit a request to the authors?
∂14-Jan-83 1509 JMC industrial lectureship
To: csd.golub at SU-SCORE
I have three written course proposals from SRI and an oral
proposal from IBM San Jose.
∂14-Jan-83 1549 JMC Consensus theory of truth
To: dam at MIT-MC
CC: phil-sci%mit-oz at MIT-MC
Subject: Consensus theory of truth
Replying to: DAM message of 1983 Jan 14, 18:17-EST
I am not very familiar with the present courses on philosophy, so perhaps
I was mistaken in supposing that "correspondence theory of truth"
has a generally accepted informal meaning. However, I now recall that
Tarski explains it by saying that "Snow is white" is true provided
snow is white. Thus it is assuming that there is a real world with
certain properties and sentences are true provided the propositions
to which they refer hold in the world. Of course, this is circular,
so a theory of true propositions has to be like, for example, a theory
of electrons. Naturally, this doesn't assume any particular set of
symbols. This now leads me to suppose that perhaps you are not as
familiar as you think with Tarski's ideas about truth, since his work
included in the collection "Logic, semantics and metamathematics"
contains informal philosophy of truth as well as notions applicable
to first order theories.
In my view, a theory of truth need not begin with a definition
of truth. As in science, generally other parts of the theory are often
more stable than the definitions of the basic concepts. This is even
true of mathematical definitions such as those of natural number. A
theory of truth must live with the necessity of treating the truth
of sentences that aren't defined in terms of the basic physics of the
world, and this will be complicated. Nevertheless, besides particular
variants of the correspondence theory of truth, there is a stable
general idea that has many adherents besides the afore-mentioned Godel.
∂15-Jan-83 0002 JMC
To: shapiro at SRI-AI
The talk will be Wednesday noon, so reserve that. I'll be back to you on
the other matters.
∂15-Jan-83 1426 JMC correspondence theory of truth
To: DAM at MIT-MC, phil-sci%mit-oz at MIT-MC
Subject: correspondence theory of truth
Replying to: DAM message of 1983 January 15, 16:17
David Mc, I think I agree with what I regard as your main points,
but let me reformulate them. A correspondence theory of what truth is,
whether formal or informal, doesn't say how truth is to be obtained.
Indeed what we obtain are beliefs, some of which are true. The
correspondence theory of truth deliberately
does not satisfy the positivist or pragmatic criterion
that truth should be defined in terms of how it is obtained.
Thus my meta-epistemology judges truth of the "scientist"'s
statements their correspondence with the facts of the "world"
part of our meta-epistemological model. It would be a theorem
of meta-epistemology that truth cannot be obtained without
experiment in certain epistemological systems.
The biggest part of epistemology indeed concerns how
truth is obtained from sense data. I also agree that Occam's razor
is essentially involved, and my proposed circumscription method
of non-monotonic reasoning can be used for formalizing Occam's
razor arguments. (There are to common spellings: Occam and Ockham.
This discussion is the first place I've seen Occum, and I tentatively
regard it as a spelling error.)
I share your doubts about the utility
of the Solomonoff, Kolmogorov, Chaitin approach, the neatest version
of which seems to be Chaitin's. Asymptotically, the approach is correct,
in that the length of the shortest program for describing the facts
is independent of the initial programming language apart from an
additive constant. However, I believe that all common sense facts
and all scientific theories produced up to the present are too short
for the asymptotic virtues of the Solomonoff approach to dominate.
I'll call it the Solomonoff approach, although I haven't read his
papers, since Minsky correctly complains that his presentation precedes
those of Kolmogorov and Chaitin, and was regrettably neglected.
Example: Suppose we have a sequence of 0's and 1's produced by a family
of rules which I will shortly give, and we wish to describe them compactly.
Solomonoff correctly points out (at least Chaitin did) that we can
start with whatever programming language L we like and use the length
of the program as a measure of the complexity of a particular rule of
the family. The asymptotic behavior won't depend on the language L, because
if another programming language L' would give shorter programs, we
have only to define an interpreter for L' in L and then use L' for
our programs. Since the interpreter is of fixed length, starting
with the wrong language only adds a constant to the length of the
descriptions of the family of rules.
Suppose the sequences to be formed in the following way. There
is a rectangular area in which a particle moves. The particle moves
with constant velocity but when it hits a wall or an obstruction it
bounces off with the angle of reflection equal to the angle of
incidence. The area in which the particle moves contains rectangular
obstructions and rectangular roofs that the particle can go under
without its motion being affected. However, when the particle is
under a roof, it is invisible. The observable sequence of 0's and 1's is formed
by sampling every second and outputting 1 if the particle is visible
and 0 if it is invisible. The different rules for determining
sequences are determined by different initial positions and velocities
and different collections of obstacles and roofs. We suppose that
the rectangle, the particle, and the obstacles and roofs are not
directly observable; all that the scientist can see directly is the
sequence of 0's and 1's.
I use this example for a variety of purposes but mainly to
argue that the heuristic or AI
essence of science cannot be summarized as the extrapolation
of the sequence of sense data. For the present purpose the point is
this. The initial programming language may be one well suited to
describing rules giving sequences of 0's and 1's. However, it
probably won't be convenient to give these sequences directly except
for very simple collections of obstacles and roofs. Instead it
will be necessary to build an interpreter for "obstacle and roof language"
and use that to go back to sequences. As scientific domains go,
the obstacle-and-roof world is very simple, but someone confronted
with the sequences and 0's and 1's would probably have to invent
obstacles and roofs in order to explain them. In fact, I'll almost
bet that if an important physics phenomenon produced such sequences,
and there were no a priori reasons to suggest the obstacle-and-roof
model, the problem might persist for years, and the ultimate
inventor of the obstacle-and-roof theory would rate a Nobel prize.
The fact that the Solomonoff model gives theories of the
obstacle-and-roof world that are asymptotically optimal as the
number of obstacles and roofs goes to infinity won't be mentioned
in the Nobel citation, although in the acceptance speech, the
inventor will thank the programmer's who generated the sequences
he compared with observation.
This is already too long, and I'll discuss the
use of circumscription to formalize Occam's razor later.
∂15-Jan-83 1430 JMC
To: kdf at MIT-MC
Is the book as whole worth buying? If it won't cost you money, and
the chapter exists in memo form, I'd like a copy. Also any new naive
physics material.
∂15-Jan-83 1723 JMC correspondence model of truth
To: dam at MIT-MC, phil-sci%mit-oz at MIT-MC
Subject: correspondence model of truth
In reply to: DAM of 1983-jan-15 19:14
I wonder if you would make an effort to find the reference to
mu-calculus. As you have described it, it is more general than the
circumscription of my paper in one respect - allowing more general
Q's - and less general in another. I do not restrict the way P
appears in Q. On the one hand, I cannot guarantee that a model
exists, but that circumscription allows non-unique minimal models
as when I circumscribe the predicate isblock in the sentence
"isblock(a) or isblock(b)". Such models are important in some of
the proposed AI applicaions. The new version of circumscription
allows minimizing an arbitrary formula, and therefore may equal
mu-calculus in this respect. I'm still writing the paper but let
me offer the following formula (modifying the notation so as not
to make presumptions about the reader's character set).
Q'(P) iff Q(P) and (P')(Q(P') and (x)(E(P',x) implies E(P,x))
implies (x)(E(P,x) implies E(P',x))
Here Q is a defined second order predicate and E is a formula
in a predicate and a variable. Q'(P) then requires that P
minimize E(P,x) subject to the condition Q. The ordering in
which the minimization occurs is then given by
R1 lesseq R2 iff (x)(R1(x) implies R2(x).
Writing the two formulas with more characters (for those with reasonable
displays) gives
Q'(P) ≡ Q(P) ∧ (∀P'.Q(P') ∧ ∀x.(E(P',x)⊃E(P,x)) ⊃ ∀x.(E(P,x) ⊃ E(P',x))
and
λx.R1(x) ≤ λx.R2(x) ≡ ∀x.(R1(x) ⊃ R2(x)).
From your hints, I would imagine that there is an important
difference of motivation, since I believe that for AI, the finite
cases are much more important than the use of the formalism to
define concepts like well ordering.
The new circumscription doesn't handle "ontological development"
by itself, but I am trying to use it to design what I have been
calling "elaboration tolerant" formalisms. Here is the example
problem I am now working on. I have written down the problem,
but I haven't yet written down my idea of a solution. If anyone
else regards this as an interesting problem, i.e. the problem of
generalizing a predicate to take an increased number of arguments
without losing information not refuted, I'd be glad to see what
they come up with.
Consider at(Stanford, California) in view of the fact that,
although it is unlikely, the trustees could decide to move Stanford
to New Jersey. In a sufficiently wide context, we might therefore
write at(Stanford,California,s). Our objectives are the following:
1. We want to include at(Stanford,California) in a database without
even imagining that it might be movable.
2. We want to be able to generalize to wider contexts. In these
such a generalization, it should be conceivable that Stanford is
movable.
3. When such a generalization is made, it is a non-monotonic conclusion,
that at(Stanford,California) is still the appropriate expression -
unless the movability of Stanford is considered.
4. Merely considering the possible movability of Stanford doesn't
prevent at(Stanford,California) from being said. However, we can
also say something like at(Stanford,California,s).
5. When we are forced to at(Stanford,California,s), the usual
properties of Stanford go along with it by suitable non-monotonic
reasoning.
6. The reasoning may force the splitting of the concept into
several. Some refer to the University, which may move and some
refer to purely geographical features like Lake Lagunita which
continues immovable. There is also the post office.
In the above we have used s as a situation but perhaps
also as a context. Pat Hayes and Bob Moore do things this way,
but I have always been dubious though without convincing objections.
We'll see whether we need distinct concepts.
Here's a try at solving the problem:
1. We reify at(Stanford,California) so the alternate
formulations are now holds(at(Stanford,California)) and
holds(at(Stanford,California),s).
...
∂15-Jan-83 1729 JMC
To: minsky%mit-oz at MIT-MC
I agree with what you think I would agree with. Which Levin is this?
∂15-Jan-83 1912 JMC consensus theory of truth and Solomonoff et al.
To: phil-sci%mit-oz at MIT-MC
Subject: consensus theory of truth and Solomonoff et al.
In reply to: the discussion
I think Lakatos's theory, based on the example of Euler's theorem,
is wrong. Euler's theorem is an extreme case, because it concerns an
area in which mathematicians were content to follow a very intuitive
tradition until very recently. Euler's theorem was considered to be built
on Euclidean geometry which had a rigorous superstructure and foundations
of sand until Hilbert cleaned it up. If this were a world in which
DARPA would pay for such things, we could, with a few man years of
work, provide a proof-checking system for an area of geometry that
would admit the statement of Euler's theorem and very much more.
We could also safely offer theorem insurance at very attractive
premiums to mathematicians who would use our system.
The Chaitin version of the theory is very powerful in the sense that
if only we knew the first few thousand decimal places of his big omega,
we could make a program that would decide Fermat's last theorem, the
Riemann hypothesis and many other open problems. It is also the only
present piece of "strongly Hardian mathematics". G. H. Hardy once
expressed satisfaction that number theory had no practical applications
- in which he was mistaken. The recent procedure for finding very
large primes for use in the RSA cipher uses higher order reciprocity
laws, which are quite difficult number theory, though I don't know
whether it uses Hardy's own work. The theory of big omega is strongly Hardian
mathematics in that it contains a proof of its own lack of applications.
You can't get the first thousand places of big omega, and if you
had it the computations to settle Fermat's last theorem wouldn't
finish before the heat death. I fear this applies to the whole
Solomonoff-Kolmogorov-Chaitin theory, though I wouldn't bet money
on it.
∂15-Jan-83 2312 JMC philosophy of science discussion
To: RWW
The M.I.T. people are have a netmail discussion on philosophy of
science that might interest you. If you look in msg.msg[jnk,jmc]
you will see, intermixed with random seminar announcements and
mail to faculty, the discussion. If you want to participate, you
mail to the person to whom you are replaying with a copy to
"phil-sci%mit-oz"%mc. If you like we can separate the phil-sci
from my junk mail file and have them put one file on their mailing
list.
∂15-Jan-83 2332 JMC "Obstacles-and-Roofs" Worlds
To: jdi at USC-ISI, phil-sci%mit-oz at MIT-MC
Subject: "Obstacles-and-Roofs" Worlds
In response to: Isaacson of 1983 jan 15 10pmPST
It would be interesting if the "obstacles-and-roof world" were an
example of your "fantomark" patterns. It would be even more
interesting if your "actual machiner" could infer obstacles-and-roofs
systems from the binary strings they produce.
∂15-Jan-83 2333 JMC "Obstacles-and-Roofs" Worlds
To: jdi%isi at SU-SCORE
Subject: "Obstacles-and-Roofs" Worlds
In response to: Isaacson of 1983 jan 15 10pmPST
It would be interesting if the "obstacles-and-roof world" were an
example of your "fantomark" patterns. It would be even more
interesting if your "actual machiner" could infer obstacles-and-roofs
systems from the binary strings they produce.
∂16-Jan-83 1154 JMC correspondence theory of truth
To: gavan%mit-oz at MIT-MC
CC: phil-sci%mit-oz at MIT-MC
Subject: correspondence theory of truth
In reply to: GAVAN of 1983 jan 16 0827EST
"Are you saying there's only one true model of the world, that you have
access to it, and that you seek to assess different scientific
truth-claims against this base-line? Am I missing something here?
What's the story? Is what is true of the world independent of what is
believed by someone? How do you know?"
A correspondence theory of what truth is can't reasonably
be based on a claim to have guaranteed access to it. Thus it may
be true that Napoleon died of arsenic poisoning, but perhaps no-one
will ever know it. Moreover, independently of whether it is true,
there may or may not develop a consensus on the question. This
view is a commonplace of folk psychology, and my opinion is that
this is another of the matters in which folk psychology is right,
and most attempts to be more sophisticated are unsuccessful.
Godel's merit was to systematically apply this idea to mathematics.
He thought that the continuum hypothesis was most likely false,
proved that ZF was inadequate to prove it false, hoped that someone
would find additional intuitively acceptable axioms that from which
it could be proved false, and wasn't sure that this would ever happen.
In my view, this is an eminently reasonable attitude except that
I never understood very well the anomalous examples that led Godel
to think the continuum hypothesis is very likely false.
Before Godel, it was possible to believe in a correspondence
theory and to hope that the truth about every question would
eventually be determined. If you don't separate the notion of
truth from the procedures for determining it, then you find yourself
unable to accept a proposition as meaningful unless you have
some advance assurance that it is decidable. This is contrary to
common sense practice, which makes conjectures without prejudice
to being able to settle them.
(Alas for people inclined to constructivist wishful thinking,
I fear it is going to be necessary to have theories
that allow discussing propositions without prejudice to whether
they are meaningful - let alone decidable. A hint: If we remove the
restriction in the comprehension axiom of ZF that the selection
is a subset of an existing set, we get an inconsistent naive set
theory. However, the ZF restriction on comprehension is more
restrictive than is necessary to get something believed consistent.
I believe it is possible to play a Godel-like trick to construct
from a set theory a stronger set theory that is consistent if the
first one is and to iterate this process through constructive
ordinals as Feferman does with theories of arithmetic. The iteration
process involves arbitrary choices that can't be specified systematically
so the limits of the iteration do not have r.e. sets of axioms.
The consequence of all this, I conjecture, will be that the set
of meaningful proppositions of set theory will not be r.e., i.e.
cannot be specified in a single theory. This is not what one
would like to be true, but when one accepts a correspondence
theory of truth, one cannot expect that everything will turn
out in accordance with one's preferences). As some famous scientist
put it: "The universe is not only queerer than we know; it is
queerer than we can know".
The meta-epistemology I propose is indeed a "God's eye view",
but it doesn't presuppose or conclude that I have a God's eye view
of this world. It proposes that we begin by studying the problem
of developing science within worlds of known structure. For example,
it has been shown that Conway's life universe admits universal
computers that can reproduce: (all M.I.T. rumor; I don't have a
reference). We can imagine a physicist program in the life world,
and can ask what epistemological mechanisms (i.e. research strategies)
if any would permit a life world physicist to determine that the
fundamental physics of his world was that of a cellular automaton and
Conway's automaton in particular. Besides doing examples of varying
complexity, we can develop mathematical theories relating properties
of the world and the subsystems regarded as scientists to what facts
about the world these scientists can determine. After developing
such theories, we can try to apply them to our own world. My
expectation would be that it would be discovered that consensus based
strategies would either be impossible to define rigorously or would
rarely work or would be unnecessarily complicated and simplify down
to correspondence based strategies.
Finally, my remarks about "muddled" and "scientifically unpromising"
were intended to suggest that it is possible and necessary to do
a lot better than this discussion has been doing. Perhaps I'm
mistaken in this opinion of much of the discussion.
There is an eight page article entitled "The correspondence theory
of truth" in Edwards's "Encyclopedia of Philosophy". It begins with
Plato, includes the Stoics, and in modern times includes G. E. Moore,
Russell (who coined the term) and Wittgenstein. It ends with Tarski.
I have to confess I didn't get much out of the article, which is by
A. N. Prior, the developer of tense logic. To put it in hacker terms:
"The program has so many bugs and such bad comments, that it
seems better to chuck it and start over" - except of course for Tarski's
technical results. Others may have better luck with it, and there is
a bibliography. The Encyclopedia is generally excellent and is sometimes
available as Book-of-the-Month premium for joining - worth it if you
are strong minded enough to subsequently buy the bare minimum of four
books.
∂16-Jan-83 1203 JMC
To: jdi%isi at SU-SCORE
CC: phil-sci%mit-oz at MIT-MC
The statement, which may be wrong, was based on the presumption that
the machinery had no a priori reason for assuming any kind of
"obstacles-and-roofs" model. The more specialized the machinery,
the less one would be impressed. A program specialized to
obstacles-and-roofs systems might or might not be impressive depending
on what was in it. Is it that you are of a different opinion or that
you hope to impress the world? If the latter, do the research and
publish.
∂16-Jan-83 1657 JMC theories of meaning
To: hewitt%mit-xx at MIT-MC, phil-sci%mit-oz at MIT-MC
Subject: theories of meaning
Replying to: Hewitt's of 1983 Jan 16 19:41est
"I don't believe that your argument quite settles the debate. Suppose
that the meaning of a sentence is taken to be the (partial) procedures
for establishing or refuting the sentence. Then there is no requirement
that the sentence be decidable. Does this theory of meaning have a
standard name? Any good citations?"
I don't see what the partial procedures would be in either the
case of the continuum hypothesis in mathematics or the question of
whether Napoleon died of arsenic poisoning. I believe 18th or early
19th century references can be found claiming that the geography of
the far side of the moon and the composition of the sun are both
meaningless, because there were no procedures for determining them.
So I suppose the question of whether another question is meaningful
may be meaningful, but again it may not be clear whether there is
a procedure for deciding whether a question is meaningful.
I don't know about citations. Try the Encyclopedia of Philosophy.
∂16-Jan-83 1703 JMC mail
To: ME
Is anyone taking steps to have the mailer know about routings?
Thus the mailer should be able to translate MAIL FOO%MIT-OZ into
MAIL "FOO%MIT-OZ"%MC.
∂16-Jan-83 1814 JMC Message of 15-Jan-83 20:26:24
To: dam at MIT-MC
I received the following message. However, the message evidently got
through, since you replied to it, and it came back to me through phil-sci.
If you feel like it, please forward it to whoever is concerned with
mailer anomalies.
∂16-Jan-83 1755 The Mailer Daemon <Mailer at MIT-OZ at MIT-MC> Message of 15-Jan-83 20:26:24
Date: 16 Jan 1983 2047-EST
From: The Mailer Daemon <Mailer at MIT-OZ at MIT-MC>
To: JMC at SU-AI
Subject: Message of 15-Jan-83 20:26:24
Message undelivered after 1 day -- will try for another 2 days:
*PS:<hewitt>mail-philosophy-of-science.txt.1 at MIT-OZ: Append access required
------------
Date: 15 Jan 83 1723
From: John McCarthy <JMC@SU-AI>
Subject: correspondence model of truth
To: dam at MIT-MC, phil-sci%mit-oz at MIT-MC
-------
∂16-Jan-83 2122 JMC keeping file of reports
To: admin.library at SU-SCORE
CC: csd.armer at SU-SCORE
Now that SCORE has plenty of disk, I would advocate keeping the
list of reports indefinitely, or, better yet, re-organizing the
old lists into a suitably accessible grand list of reports.
The cost should be estimated, but I'm sure it's easily affordable.
∂17-Jan-83 0051 JMC verificationism
To: phil-sci at MIT-OZ
I think verificationism is no good for AI for the same reason as it
is no good for science. Taken seriously, it limits thought. Consider
the hypothesis that matter is composed of atoms. During most of the
nineteenth century, no-one could think of any way of verifying it,
and towards the end of the century, some famous chemist of a positivist
frame of mind, Ostwald perhaps, emphasized that it was just a means
of keeping track of some phenomena, e.g. the law of combining proportions.
Shortly thereafter, it was verified and Avogadro's number was computed
in a variety of ways. The most spectacular way was that individual
scintillations from radioactive decay were observed, a rate of decay
in atoms per second computed and compared with a rate of decay in
(say) micrograms per year of a large sample.
If AI systems were programmed to consider only propositions
when the could generate a means of verification, they would be
extremely unimaginative.
I must confess, however, that my real objection is different.
If a universe exists and evolves intelligence, why should it only
evolve intelligences that can observe every feature of the universe?
I can't imagine a mechanism of evolution that would guarantee this.
∂17-Jan-83 1144 JMC verificationism and correspondence
To: gavan at MIT-MC, phil-sci at MIT-OZ
Subject: verificationism and correspondence
In-reply-to: GAVAN of 1983 jan 17 0805EST
"Isn't this also an argument against the correspondence theory?"
You'll have to elaborate this a bit before I can respond.
∂17-Jan-83 1352 JMC
To: kdf@MIT-OZ
Thanks.
∂17-Jan-83 1423 JMC Lakatos and Solomonoff
To: gavan@MIT-OZ
CC: dam@MIT-OZ, phil-sci@MIT-OZ
Lakatos, if Proofs and Refutation is the book in question is concerned
with the social process whereby the mathematical community comes to
accept a theory. Perhaps it also supposes that the meaning of a theorem
and its truth is also socially determined.
What the Chaitin version (the one I have read about in Scientific
American) concerns is a notion of simplicity of sequences and functions
using the shortest program that generates the sequence. As far as I
can see neither theory is best explained as a variant of the other
or in contrast with the other. Each is best explained starting from
scratch. The surprising and impressive thing about the Solomonoff,
et al theory is that so many of the results are independent of the
programming language chosen, and, in fact, Chaitin doesn't bother
to use a specific language. The theory isn't very difficult, and
the Scientific American article, a few years back, and the references
it gives are an excellent source.
∂17-Jan-83 1643 JMC Sten-Ake
To: CLT
He's in town and is agreeable to giving a seminar on the 26th, and
I think that's the only possible time. If you like, I'll extract
an abstract and have him mail it to you.
∂17-Jan-83 2157 JMC
To: DCL
Feb 2 is ok with me.
∂17-Jan-83 2234 JMC Correspondence theory of truth and meta-epistemology
To: gavan@MIT-OZ
CC: phil-sci@MIT-OZ
Subject: Correspondence theory of truth and meta-epistemology
In reply to: GAVAN of 1983 jan 17 1736
jmc- The meta-epistemology I propose is indeed a "God's eye view",
but it doesn't presuppose or conclude that I have a God's eye view
of this world. It proposes that we begin by studying the problem
of developing science within worlds of known structure.
GAVAN - Whose known structure? This, by the way, is part of what Lakatos is
doing. See his comments on generating and degenerating research
programs.
I evidently not made one thing perfectly clear. "Whose known structure"
refers to mathematical structures concocted for studying meta-epistemology
which will not be anyone's conjecture of what the real world is like.
For example, the Conway life universe with its life physicists is such
a structure. The question for mathematical study is what if any strategies
by this program will discover that it "lives" in the Conway universe.
The advantage is precisely that we can study the what strategies are
effective in what kinds of universe without presumptions about our
own universe. Besides studying particular strategies in particular
hypothetical universes, we can try to prove general theorems about
what kinds of strategies will work in what kinds of universes. My
conjecture is that various rigidly defined operationalist strategies
can be proved not to work.
Remember also the point, admitted by almost everyone when
not arguing about scientific method, that we evolved very recently
in a world which shows no evidence of having been designed for our
convenience - intellectual convenience as well as physical
convenience. There is no guarantee whatsoever that every aspect
of the world will ever be observable. Presumably, meta-epistemology
will have theorems about when a world is completely knowable, and
I conjecture that the condition will be quite special and not very
plausible with regard to the real world.
Mathematical platonists, of which I am one hold that there
is objective mathematical truth which mathematicians try to discover.
This truth does not depend on the physical world, and this fact
gives rise to difficulties in figuring just what kind of a beast it
is. While many mathematicians hold formally other points of view,
formalist, logicist or constructivist of some flavor, lots will admit
that they informally act as Platonists. Godel argued, and I tried
to paraphrase some of his arguments, that much of his success was
due to his Platonism. Van Heijenoort, a historian of logic, doesn't
agree, so it's not unanimous about whether Godel's philosophy helped
his mathematics.
∂18-Jan-83 1204 JMC Correspondence theory of truth
To: gavan@MIT-OZ, batali@MIT-OZ
CC: phil-sci@MIT-OZ
Subject: Correspondence theory of truth
In reply to: GAVAN of 1983 jan 18 0622EST
The Tarski theory ascribes truth to sentences, so the correspondence
is between these sentences and the world. How these sentences are
represented, whether in a physical structure or (as might be preferred
by a dualist if dualists exist) in a mind is, strictly speaking,
not part of the theory. In my proposed meta-epistemology, the
sentences (or more abstract objects called propositions) are represented
in the "memory" of the "scientist" part of the system. Thus in the
Conway life world physicist, the question is whether the fact that
their fundamental physics is life, will become represented in the
language we are talking about in the memory of the "physicist" or
in their "Physical review". Both the memory of the physicist and
the Physical review are encoded in complicated configurations of
life cells being on or off. Thus we are talking about a correspondence
between two physical structures.
∂18-Jan-83 1500 JMC correspondence theory
To: gavan@MIT-OZ, batali@MIT-OZ
CC: phil-sci@MIT-OZ
Subject: correspondence theory
In reply to: GAVAN of 1983 jan 18
I don't have a design for a life-world physicist. The point is the
definition of truth. Suppose we have such a physicist program
in the life-world, and it generates certain sentences in the
configuration of life cells we call its memory, and we have
an interpretation of these sentences ask making assertions about
the life world. Then, we say that a sentences is true provided
what it asserts about the life world is true. For example, the
life physicist may produce sentences interpretable as a theory
of cellular automata. There may be another sentences in the
same language asserting that the particular world is a certain
cellular automaton. If it asserts that its world is the life
automaton, we have the desired CORRESPONDENCE between the sentence
and the world.
To recapitulate: we are talking about a correspondence definition
of truth, not about how to design a life world physicist. The latter
would be difficult in the present state of AI.
∂18-Jan-83 2156 JMC previous committment
To: llw@S1-A
I forgot that I have to be at a seminar at noon and so couldn't get
to LLL before 2:30 but can stay into the evening. If I hear from you
that we could still interact, I'll come; otherwise I'll phone and
arrange another time.
∂18-Jan-83 2231 JMC Correspondence theory
To: gavan@MIT-OZ, batali@MIT-OZ
CC: phil-sci@MIT-OZ
Subject: Correspondence theory
In reply to: GAVAN of 1983 Jan 18 1549
A last try: If the life automaton writes in its memory a sentence
asserting in the language we suppose it to be using that the physics
of its world is life, i.e. writes it in a sense similar to that in which
our physicists write that our universe satisfies the genral theory of
relativity, then we
say that the sentence is true, because what it says corresponds to
the structure of its world. This is the sense in which Russell,
who invented the term "correspondence theory" and the other advocates
of the theory, going back to the ancient Greeks meant to define truth
by correspondence. Defining truth by such correspondences is distinct
from defining it by coherence or by consensus.
Volume 1 of Putnam's Philosophical Papers doesn't include one called
"The Meaning of Meaning", but Putnam generally characterizes himself
as a realist, both in physics and mathematics, e.g. on page 60 in an
article entitled "What is mathematical truth?", he begins, " In this
paper I argue that mathematics should be interpreted realistically
- that is, that mathematics makes assertions that are objectively true
or false, independently of the human mind, and that SOMETHING answers
to such mathematical notions as 'set' and 'function'."
I can't find Volume 2 at the moment.
∂19-Jan-83 0828 JMC
To: LLW@SU-AI
I call you, and we'll make it another time.
∂19-Jan-83 1509 JMC
To: llw@S1-A
Would next Wednesday be suitable?
∂19-Jan-83 1548 JMC
To: csd.genesereth@SU-SCORE
John McCarthy is Professor of Computer of Science at Stanford University
and one of the founders of artificial intelligence research. He invented
the LISP, the programming language most used in AI research and also
first proposed the general purpose time-sharing mode of using computers.
The emphasis of his AI research has been in identifying the common
sense rules that determine the consequences of actions and other events,
the expression of such rules and other common sense information as
sentences in logical languages in the data bases of artificial intelligent
programs. He has also worked on formalizing common sense reasoning,
and his recent work has concerned non-monotonic common sense reasoning.
He is President-Elect of the American Association for Artificial Intelligence
and a member of the Editorial Board of the journal Artificial Intelligence.
∂19-Jan-83 1551 JMC
To: llw@S1-A
How about Monday Jan 31 or Friday Feb 4?
∂19-Jan-83 1753 JMC
To: DFH
Please decorate greens.1.
∂19-Jan-83 2321 JMC Lakatos review, Putnam, and Solomonoff (or even Solomonov)
To: gavan@MIT-OZ, minsky@MIT-OZ
CC: phil-sci@MIT-OZ
Subject: Lakatos review, Putnam, and Solomonoff (or even Solomonov)
In reply to: various
I have FTPed it, so you can delete it if you want. Its flashy rhetorical
style and many references make it difficult for me to determine what its
points amount to. It seems to me that the Popper, Kuhn and Lakatos
are in disagreement, not so much about what scientists do, but about
how to describe it. Morevover, none of them seems to be advocating
a change in actual scientific practice.
In principle this should be great for AI, since a big part of our
problem is to describe scientific method to a computer. However,
I can't find anything at all usable in the review. It seems like
we have to go back to simple models in which theories of relatively
simple phenomena can be proposed and refuted or tentatively
confirmed.
Putnam is indeed not consistently realist, and therefore I'm inclined
to disagree with him. One rhetorical question of his on p. 19 of
volume 1 illustrates one place of disagreement. He is discussing
Russell believing in the genuine existence of the set of all
predicates on natural numbers. Cantor, believing in this set,
showed that it is non-denumerable which means that its members
can't all individually be named. Moreover, whatever properties
you attempt to characterize it by in ZF, it turns out that there
are models (denumerable even) that consist of only some of the
subsets that have all the same properties. (My exposition of
the mathematical situation). All this leads Putnam to ask the
rhetorical question: "Surely it is reasonable in science to ask
that new technical terms should EVENTUALLY be explained?"
The meta-epistemology viewpoint answers Putnam as follows: It
would indeed be nice if all technical terms could be explained.
However, it seems that if certain kinds of worlds evolve intelligent
beings, they won't be able to EXPLAIN all the technical terms they
can profitably use - neither in physics nor in mathematics. From
this point of view Putnam's remark is wishful thinking. Both science
and mathematics seem to be capable of obtaining only partial
results. Moreover, an attempt to syntactically limit the language
to those terms that can be EXPLAINED also fails. As soon as you
give me a rule defining what you regard as acceptable, I will use
this very rule to construct an intuitively meaningful concept
that falls outside your rule.
A remark on Solomonoff and simplicity: I fear that the simplest
theories are not obtainable and can't be verified to be simplest
if obtained except in trivial cases. From what Marvin says,
Solomonoff is careful about making his criteria often prefer
ensembles of theories. We can partially confirm his worries about
the most compact theories by recalling Shannon's minimal relay
machines which were utterly incomprehensible, because they relied
on noticing accidental overlaps of function between that permitted
hardware to be shared in unintuitive ways. I believe that the
ensemble methods may produce theoretically interesting results
and may even suggest practically useful methods, but I don't
believe that an attack at such a general level will produce
useful results. We need methods that allow us to tell the
machine what we know about the common sense world, e.g. about
space and objects in space. As an aside: are the Solomonoff
methods directly applicable to the sequences produced by
"obstacles-and-roofs", and if so, what results would Marvin
expect?
Since Solomonoff is an American, his name, though of Russian
derivation, must be spelled as he spells it, especially if you
want to find him in the phone book or his papers in the library.
As for Kolmogorov,
that is the most common American transliteration of the Russian
and is used by the Library of Congress and Mathematical Reviews,
so if you want to find his works, it's what you'd better use.
Other languages have other systems of transliteration, and both
Kolmogoroff and Kolmogorow are sometimes seen. If Solomonoff
were to move to Russia, God forbid, and publish there, the
American librarians would then call him Solomonov unless they
knew about his previous publications, and then ... . My friend
Andrei Ershov, spells his name that way when he publishes in
English, but when he publishes in Russian, it is often transliterated
here as Yershov. The former looks like the Russian, but the
latter is in a system that permits unambiguous (usually) transliteration
back to the Russian alphabet.
∂19-Jan-83 2351 JMC
To: llw@S1-A
The fourth it shall be.
∂20-Jan-83 1123 JMC
To: gavan@MIT-OZ
CC: phil-sci@MIT-OZ
With regard to giving emergent research programs sufficient time,
my group is expecting a visit on February 2 from Colonel Adams of DARPA
and John Machado of the Office of Naval Research. I don't know whether
they are adherents of Popper, Kuhn or Lakatos in determining how much
time they should give emergent research programs. More seriously, I
am not aware, perhaps I should be, of research programs that have
retained or lost adherents as a direct result of Popper's advocacy.
Hmm, perhaps the mistaken (in my opinion) abandonment in the 1960s
of much language translation research was a result of Popperian ideology.
In reply to my arguments that the world may well be such that the
truth cannot always be ascertained, you ask why I still uphold the
correspondence theory. My point was that whether a statement is
true depends on the world and not on the methods available to the
truth seeker for guessing the truth or confirming his guesses. To
put the matter sharply, if a British museum monkey types "The earth
is round", then it has typed a true sentence of English, while if it
types "The earth is flat" it has typed a false sentence of English.
The monkey's state of mind, if any, is irrelevant.
∂21-Jan-83 0107 JMC printing mathfile equations
To: DEK
Today I tried using Mathfile by Dial out of SAIL, and it works fine including
the ability to get the abstracts themselves into a file. However, the formulas
are in some kind of text formatting notation. Do you know anything about
this notation, e.g. is it related to TEX, and is there some way of printing
it here as formulas?
∂21-Jan-83 0111 JMC
To: DEK
foo1[1,jmc] is the result of a search on "rotation number".
∂21-Jan-83 1848 JMC correspondence theory
To: gavan@MIT-OZ
CC: phil-sci@MIT-OZ
Subject: correspondence theory
In reply to: Gavan of 1983 jan 21
Gavan: Indeed! Even if there were no English interpreters, the statement
"The world is round" would still be a true sentence of English. As far
as I can tell, this would be the position of all the supporters of the
correspondence theory including Tarski. A sentence in a language
is an abstract object existing mathematically independent of whether
anyone ever interprets it or even exists to interpret it.
∂21-Jan-83 2249 JMC
To: minsky@MIT-OZ, gavan@MIT-OZ
CC: phil-sci@MIT-OZ
Marvin: Indeed! My reaction to Tarski's exposition was the same,
except for reading it in English. The correspondence theory of truth
seems obvious, because it agrees with the common sense notion. It is
only when someone proposes some other theory or denies that there is
any such thing as a true sentence that Tarski's statements seem other
than tautologous, i.e. that there seems to be a need for a theory of
truth. Notice, however, that the correspondence theory requires that there be
something objective to correspond to - a physical world that either agrees or
not with the sentences or mathematical objects such as sets that either
do or don't have the properties asserted. This also is the common
sense view, and hence IS obvious unless challenged. As I understand it,
coherence theories don't require that there be an objective reality, since
they purport only to relate experiences. However, we correspondence
theorists consider that the coherence theorists have been unsuccessful
in relating experiences to one another except in so far as they have
allowed external reality to sneak back into their theories.
In the hopes of eliciting some reaction from someone besides GAVAN, who
seems not to believe in objective reality, I will again
advocate meta-epistemology. We try to get a mathematical
theory of the relation between the strategy of a knowledge
seeker in a world and its success in discovering facts about
the world. This theory doesn't directly involve conjectures
about the real world, because the worlds studied are abstract
mathematical objects. The theory would relate the following
things:
1. The structure of the world. Since we get to postulate the world
or give it any properties we want, it's no mystery to us.
2. The imbedding of the knowledge seeker in the world. He could be
outside it as in Ed Moore's "Gedanken Experiments with Sequential
Machines" in Automata Studies. However, it is more interesting
and gives theories more transferable to ourselves if he is built
as part of it, as in the Conway life world. The imbedding also
includes his input-output relations to the rest of the world.
3. The language used by the knowledge seeker to express his
conjectures about the world. Let's assume that we have an
interpretation of at least part of this language as expressing
assertions about the world.
4. His philosophy of science - what assertions he considers
meaningful and what he regards as evidence.
5. Finally, what he succeeds in discovering about the world, i.e.
the true sentences he generates in the language we interpret as
expressing assertions about the world. Some knowledge seekers,
of course, might generate expressions that could be regarded
as assertions about the world in some other language, but we
won't count them.
The issues that arise include the following:
1. If its language only includes input-output relations, will it
even discover as many input-output relations as someone with
a more liberal philosophy of science?
2. Can we make a good one using the Solomonoff strategies, or rather,
in what kinds of worlds will the Solomonoff strategies be effective?
3. What about mathematics? Lenat's AM worked with numbers and
properties of numbers, but I don't think it had any richer
ontology - if it was formal enough to be said to have an ontology.
∂21-Jan-83 2258 JMC
To: minsky@MIT-OZ
CC: phil-sci@MIT-OZ
Marvin: It's to harsh to say that a joke should not be told more than
once. As Lloyd Shapley said to me when I told him I was taking a
trip to Novosibirsk, "If you haven't got a new lecture, get a new
audience". T. H. Huxley once said, "Mr. Herbert Spencer's idea of a tragedy
is when a beautiful theory is slain by cruel fact". When I tried to
track down the precise occasion, I discovered that both he and Spencer
liked the joke so well that they both said it several times.
∂22-Jan-83 1239 JMC
To: minsky@MIT-OZ
CC: phil-sci@MIT-OZ
I agree with your theory of sparseness, which resembles the sign posted
in the Durgin-Park restaurant: "There isn't any place anything like this
place anywhere near this place, so this must be the place". I have called
the theory "the argument from cryptography" in my "Ascribing Mental
Qualities to Machines" paper. In principle, a cryptogram could have
multiple solutions, but this doesn't happen. When someone gets an
English text resembling from a cryptogram, it always turns out to be
the text the writer of the cryptogram started with. There is a slight
counterexample in French where a cryptogram has the two solutions both
making sense, "Le prisonnier est fort, il n'a rien dit" and "Le prisonnier
est fort, il n'a rien dit" which translate to "The prisoner is strong,
he has said nothing" and the "The prisoner is dead, he has said nothing".
The difference is based on a letter occurring once in the cryptogram which
could either be an "f" or an "m". Philosophers often pose the
possibility that a novel in one language could be a cookbook in
another and therefore suggest that a child needs some a priori built-in
knowledge that leads him to human language and not some other. However,
the Shannon information theory indicates that a book length text having
two completely different interpretations in two languages is as
improbable as the molecules of air all rushing to one side of the room.
Russian and English share 20 glyphs in their alphabets, but the longest
texts anyone has come up with that could be either Russian or English
have three letters. Thus the inscription POT means pot in English and
company (military) (pronounced like the English "rote") in Russian.
However, I don't believe this sparseness argument makes obsolete
traditional treatments of truth, etc. It's merely another fact to
be taken into account.
∂22-Jan-83 1241 JMC
To: minsky@MIT-OZ
Yes, I prefer not to receive two copies of notes to phil-sci. If this
is also your preference, I won't again send copies of phil-sci messages
to you.
∂22-Jan-83 1258 JMC arpa
To: RPG
Your proposal wasn't lost and seems ok. I'll need to look at it further.
I'm not sure that there shouldn't be some more prose justifying ARPA
support of Commmon Lisp, i.e. how money will be saved and innovation
in the right areas encouraged. We should discuss it.
The time for the new ARPA proposal approaches fast, and we need an abstract
for Adams and Machado February 2.
∂22-Jan-83 1301 JMC New ARPA proposal.
To: JK, LGC, RPG, CG, CLT, JJW, YOM, DFH
Our current ARPA funding runs out Oct 1, but they need a LONG time to process
a proposal. We are included with Wiederhold, Luckham, Binford and Manna
in a grand proposal. Col. Duane Adams (ARPA) and John Machado (ONR) will
be here Feb. 2 for a 9-11 meeting covering the whole group. Therefore, there
will not be time for any real show-and-tell. However, they want an abstract
of a proposal by then. Of course, much of our work is supported by NSF, but
nevertheless, I would like to put everything into the ARPA abstract. Therefore,
I would like messages from each of you with what you propose to do in the
next two years. In the case of Joe and Yoram, this amounts to preliminary
thesis proposals. I will ask for help later in putting the whole thing
together into a proposal.
∂22-Jan-83 1502 JMC Marconi
To: pourne@MIT-MC
Dear Jerry:
I am very pleased and grateful that you would consider proposing
me for a Marconi prize, but it may be more work than you are prepared
to undertake. This is because the work I have done that might possibly
merit it isn't nicely encapsulated in three or four books. This is my
fault, and it might well be taken as a reason for my not meriting such
an award. There are five lines of work that might be taken into account.
These are (1) initiating the formalization of common sense as a line
of work in artificial intelligence. This has was started in "Programs
with Common Sense" in 1958 and continued in the papers marked (1)
in the bibliography I will send you. Patrick Hayes of the Computer
Science Department at the University of Rochester and Nils Nilsson of
SRI International in Menlo Park, CA are best equipped to evaluate this.
(2) The Lisp programming language. The research was started in 1956,
and the first version of LISP was working early in 1959. It took about
20 years to be fully recognized as "the artificial intelligence language",
although its extensive use began much earlier. Douglas Hofstadter of
Indiana University is writing about it in Scientific American now and
might contribute some good words. Also Dan Friedman also of Indiana.
(3) Mathematical theory of computation - proving facts about computer
programs. My approach is one of two main approaches, and the adherents
of one do not always recognize the value of the other. Prof. Rod Burstall
of Edinburgh University and Prof. Woodrow Bledsoe of the University of
Texas at Austin and Prof. Robert Boyer of the University of Texas might
write about that.
(4) The first proposals for general purpose time-sharing. I believe my
work did initiate this work, but there isn't really a single major paper.
Prof. Joseph C. R. Licklider of M.I.T., Prof. Edward Fredkin of M.I.T.
and possibly Dr. Gordon Bell of Digital Equipment Corporation might
be able to say something about that.
(5) Within the last 5 years, I proposed a system of non-monotonic reasoning
called circumscription which is intended to allow a computer to jump
to conclusions and perform reasoning involving Occam's razor. This has
not (yet) won out decisively over competing proposals for non-monotonic
reasoning but it demonstrates continuing significant contributions and
ought therefore to be relevant.
Besides these personal scientific contributions, people might be inclined
to give some weight to my having initiated the Dartmouth Summer Research
Project on Artificial Intelligence in 1956, which was the first attempt
to get researchers in this field together and my roles in initiating
and directing artificial intelligence work first at M.I.T. (together
with Marvin Minsky) and later at Stanford. Minsky, Prof. Raj Reddy
of Carnegie-Mellon University, and Professor Edward Feigenbaum of Stanford
might say something about these activities. Both Minsky and Feigenbaum
might say something about my research ideas in AI, but both of them
probably disagree with some of my opinions about what is likely to
work.
I will send you the biographies and bibliographies, but I will
understand if you conclude that the amount of work involved in getting
a proper case before the Committee is more than is warranted by either
the probability of success or the worthiness of the enterprise.
I'll tell you about my part in another, so far unsuccessful,
attempt to propose someone for the Marconi award when we can talk
personally.
∂23-Jan-83 0014 JMC
To: hewitt@MIT-OZ
Carl: I keep getting these messages, although I think my messages get
through to phil-sci. Could you pass the problem to the responsible actor.
∂22-Jan-83 2312 The Mailer Daemon <Mailer at MIT-OZ at MIT-MC> Message of 22-Jan-83 01:50:32
Date: 23 Jan 1983 0207-EST
From: The Mailer Daemon <Mailer at MIT-OZ at MIT-MC>
To: JMC at SU-AI
Subject: Message of 22-Jan-83 01:50:32
Message undelivered after 1 day -- will try for another 2 days:
*PS:<hewitt>maiphilosophy-of-science.txt.1 at MIT-OZ: Append access required
------------
Date: 21 Jan 83 2249 PST
From: John McCarthy <JMC@SU-AI>
To: minsky@MIT-OZ, gavan@MIT-OZ
CC: phil-sci@MIT-OZ
-------
∂23-Jan-83 0114 JMC
To: phil-sci@MIT-OZ
Reading the article "The coherence theory of truth" in the
Encyclopedia of Philosophy and recalling Marvin's remarks about
sparsenes leads me to change somewhat my views expressed previously.
The article emphasizes the fact that we don't verify single statements,
but whole complexes of statements, i.e. theories and languages.
Moreover, within such a theory it is often, and perhaps necessarily,
unclear which statements are unadulterated observations, which are
definitions, and which are theoretical statements whose truth may
be verified.
All this doesn't interfere with the correspondence theory
of truth as applied to single statements in a given language such
as English. Nor does the fact that theories are verified as wholes,
rather than statement by statement, necessarily affect a viewpoint
to which truth has little to do with the method of verification.
However, it suggest we can do better.
Suppose someone supplies us with a 50,000 word textbook on Newtonian
mechanics written in Martian. Suppose that there are some errors
in the text and moreover we don't know the subject matter, there
are no diagrams, and we can't read Martian. It may be very difficult
and take a long time to figure out what the document is.
Someone may initially advance the theory that the document is a cookbook
or a novel. Nevertheless, it is extremely improbable, e.g. of the
order of the molecules rushing to the other side of the room, that
this textbook admits any coherent interpretation than as a textbook
of Newtonian mechanics with a few errors. Whether cryptography
and linguistics are presently up to guessing it, I don't know.
Remember that the still undeciphered inscriptions in unknown
languages are almost certainly not narratives or exposition but
mostly lists, whether of kings or the contents of warehouses.
What this suggests is that a long enough document may
have certain statements in it true or false in an absolute,
language independent sense. Of course, it may also assert
a myth in the same absolute sene of admitting that and no other
interpretation. Some of the background for these assertions
is in Shannon's 1948 Bell System Technical Journal article
on the probabilistics of cryptography. These ideas may be
combined with those of Solomonoff, Kolmogorov and Chaitin.
∂24-Jan-83 0111 JMC objective physical and mathematical worlds
To: phil-sci@MIT-OZ
Subject: objective physical and mathematical worlds
In reply to: mainly GAVAN
Whether the world should be regarded as a construct from
sense data and/or other experience or should be regarded as
existing independently of mind has been argued for centuries.
The same question arises about whether mathematical facts are
to be regarded as independent of the existence of human or other
minds. I believe that I have convinced most of the participants
in the debate that I am an actual adherent of "realism" in both
cases, and this took some doing. However, I haven't addressed
the issue itself, mainly because what little I can
add to the debate seems unlikely to change many minds. However,
GAVAN keeps emitting rhetorical questions like "What world?", so
perhaps I should say something.
Descartes tries to begin his consideration of philosophy
with a clean slate and argues "Cogito ergo sum". He does not even
accept the existence of other minds a priori, but considers
their existence to be a consequence of his reasoning. In order
to get such results, he adopts methods of reasoning so strong that
he can deduce the whole of the Catholic religion - which might raise
suspicions about his "rules of inference" among non Catholics.
Positivists often also propose to start from bare sense experience
and see what can be gotten from that.
There is, however, another principle from which one might
start, and I'd like to give it the fancy name of "Principle of
philosophical relativity". Consider taking as a starting principle:
"There is nothing special about me". Unless there is positive
reason to believe otherwise about some aspect of reality,
I will assume that I am not in a unique position. If I have
experiences and thoughts of a certain kind, very likely other
people have similar thoughts and experiences. This corresponds
to common sense prejudice, and indeed we seemed to be programmed
that way. A week old baby will open its mouth in response to its
mother's open mouth - presumably without having gone through the process of
deducing the existence of other minds and automatically making
a connection between the sight of its mother's mouth, and the
position of its own mouth, which it has never seen. We may regard
the baby as jumping to mistaken conclusions.
If we refrain from overcoming this apparently built-in principle
of philosophical relativity, we get other minds, other physical
objects and lots more rather early in our philosophical investigation.
Another argument that impresses me is the following: I
was taught in school about how the earth was formed from
the solar nebula, cooled off, developed life which evolved more
complicated forms, one form of which evolved intelligence, evolved
a culture, and eventually developed institutions of higher learning
in which some of us are even paid to think and argue about
philosophy. Now I am asked to believe that all this about
life and intelligence evolving isn't to be taken seriously as
something that actually occurred but is to be taken merely as
a convenient way of organizing my experience and predicting
future experience. I suppose I could manage this change of
viewpoint but am insufficiently motivated by any hope of benefit.
The question of objective mathematical reality is harder (for me)
to argue about. Would it be at all convincing to meet extra-terrestrials
and discover that while their mathematics had gone farther in some
directions than ours and less far in others, they talked
about the same basic systems of algebra, topology, analysis and
logic? Does anyone expect something drastically different?
I'm inclined to take what apparently is a relatively extreme position
among mathematicians, although it was Godel's position, and say (for example)
that the continuum hypothesis is either true or false although it
is much less certain that humans will ever know or will ever even
have a strong opinion.
There is also a question about what level of certainty
should be demanded before accepting the existence of other minds,
etc. Many people profess uncertainty about whether the physical
world exists, but don't seem to give the slightest weight to the probability
that they don't in their practical actions. This suggests that
a test be devised for the seriousness of sense data theorists.
It would involve offering a prize that could be won if there were
relations between sense data apart from those mediated by material
objects. Someone who put effort into trying to win the prize would
be showing some seriousness about the sense data view. Perhaps someone
can come up with a better way of formulating such a test.
Well that's all I can come up with at the moment, though
there's lots more in the literature.
∂24-Jan-83 1415 JMC "your version of reality"
To: gavan@MIT-OZ, phil-sci@MIT-OZ
Subject: "your version of reality"
In reply to: GAVAN
"Belief in OBJECTIVE reality is surely not pragmatically useful
(depending upon what pragmatics means for you). If you believe your
version of reality is objective, then be prepared to beat your head
against a wall for the rest of your life."
The phrase "your version of reality" leads to two kinds
of confusion:
1. Belief in the existence of objective reality, i.e. that
there are facts independent of human experience in general and one's
own in particular, does not require belief in a particular "version
of reality". Thus I am prepared to learn that there is a wall
where I previously thought there was an opening. Moreover, this
experience reinforces the doctrine that my beliefs are true only
if they correspond to reality.
2. A person's beliefs cannot be summarized as a "version
of reality" for two reasons. First a version of reality would involve
more detail than a human holds - the names of all the people in
the world to begin with. Our opinions cover only a tiny part of
reality. Second, even when an AI program's reality is restricted to a
tiny part of the world, e.g. a collection of blocks on a table, its
view cannot in general be regarded as a version of reality. It may
not have an opinion about the location of some block or it may have
a disjunctive opinion: e.g. it may believe that a certain box
contains a red block or a green block. This requires distinguishing
states of belief from belief in states of the world - or even in
partial states of the world. Bob Moore in his M.I.T. master's
thesis emphasized how AI programs whose belief structures were
whole worlds or partial worlds are limited in their capabilities.
The first approximation to a state of belief is a THEORY in the
sense of mathematical logic. A possible state of reality corresponding
to the state of belief would be a MODEL of the THEORY. I'm
adopting a convention of capitalizing technical terms. Unfortunately,
it may be that more sophisticated notions are required.
∂24-Jan-83 1426 JMC patience with GAVAN
To: kdf@MIT-MC
Indeed I sometimes find him both arrogant and thick. A hint is not
sufficient for him, but when he misunderstands he makes his misunderstanding
entirely clear. Most people are more polite, and when they misunderstand,
this is not obvious. Almost always, when GAVAN seems to me to have
misunderstood, I find that I have not been sufficiently explicit - a
fault of mine. If I later get around to summarizing my views on these
philosophical issues, I will go through the whole exchange and see if
I can obviate some of the misunderstandings by being more explicit -
and perhaps more long-winded in the first place. Please don't inform
me where he is on the scale between sophomore and professor of
philosophy.
∂24-Jan-83 1508 JMC correspondence theory
To: gavan@MIT-OZ, phil-sci@MIT-OZ
Subject: correspondence theory
In reply to: GAVAN
GAVAN: "Please remember that I am not denying the existence of
reality, only the objectivity of anyone's experience of reality
(and also the idea of a correspondence). Sceptics have always been
able to demonstrate that the existence of reality is unprovable but
they've never been able to disprove its existence either. Even the
sceptics engaged in practice, so they did really assume that the
world exists, as I do. Their point and my point is not that the
world doesn't exist. The point is that there's no necessary
correspondence between what's in the world and what's in your mind
(or what's in your sentences)."
Can it be that most of our arguments have been based on mere
misunderstanding? The correspondence theory does not require the
correctness of anyone's opinion of reality. Correspondence is instead the
criterion for the truth of a belief. In this interpretation I claim to
also speak for the authors referred to in the Encyclopedia article on the
correspondence theory.
There used to be a further issue about the "objectivity of
observation", i.e. whether trees (directly observed) are as real as
elementary particles, but I think arguments on this subject have died down
- both are real.
∂25-Jan-83 1131 JMC
To: RPG
Call Ron Ohlander 202 694-5051 about proposal and Common Lisp.
∂25-Jan-83 1509 JMC correspondence theory, misunderstanding thereof
To: gavan@MIT-OZ
CC: phil-sci@MIT-OZ
Subject: correspondence theory, misunderstanding thereof
In reply to: GAVAN
GAVAN: `As I understand the correspondence theory, it posits that there is
a one-to-one correspondence between things in the world and things
in the mind -- that we "copy" the world, starting with a "clean
slate."'
I think you misunderstand the correspondence theory. I don't believe that
any of its adherents would claim that the correspondence between the
things in the world and things in the mind is one-to-one. I refer again
to the Enclyclopedia article for others' views. For myself, the correspondence
involved in the truth conditions for a sentence can be quite complex.
The important point is the existence of reality independent of human
experience. You can call it an article of faith if you like; I'd
call it a bet.
∂25-Jan-83 1817 JMC Ad nauseum
To: batali@MIT-OZ
I think it's about time to change the subject. I have no additional
arguments that have any hope of convincing GAVAN, and I fear others
are merely being bored.
∂25-Jan-83 1858 JMC afghanistan
To: lederberg%SUMEX-AIM@MIT-MC
The January 24 Wall Street Journal has an article "Lifting the Curtain
on Afghanistan's Horror" by Rosanne Klass of Freedom House that reports
on the Paris hearings of the "Permanent Tribunal of the Peoples", the
successor to the Bertrand Russell tribunal. Its December 20 press
conference produced no news coverage that any of us saw. Anyway it took
much testimony on Soviet atrocities in Afghanistan, though not mainly
about chemical and biological warfare. It seems that the a major part
of the problem might be taken as publicizing their findings.
I'll send you a copy if it isn't still accessible to you.
∂26-Jan-83 0037 JMC letters
To: DFH
takasu.1 and lee.2 are ready for decoration, except that I don't
know whether, Tsing-hua University on Taiwan is formally in Taipei
or some other city. Perhaps you can find out.
∂26-Jan-83 1116 JMC isaacson%isi,phil-sci%oz/cc
In reply to: Isaaacson
As I understand intutionism, the syllogism mentioned by the puzzled
soul is a perfectly good intuitionist syllogism. It's just that
establishing "All birds fly" requires a constructive method.
All birds fly;
Fred is a bird;
Then Fred flies.
∂26-Jan-83 1121 JMC intuitionism
To: isaacson%USC-ISI@MIT-MC
CC: phil-sci%MIT-OZ@MIT-MC
In reply to: Isaaacson
As I understand intutionism, the syllogism mentioned by the puzzled
soul is a perfectly good intuitionist syllogism. It's just that
establishing "All birds fly" requires a constructive method.
All birds fly;
Fred is a bird;
Then Fred flies.
∂26-Jan-83 1156 JMC
To: DFH
schank.1
∂26-Jan-83 1248 JMC Common Lisp proposal
To: bscott@SU-SCORE, RPG@SU-AI
Ohlander would like the proposal officially submitted as soon as
possible. I believe both budget and content are ok.
Dick should phone him at earliest convenience.
∂26-Jan-83 1351 JMC
To: JK
The sentence about the use of EKL in CS206 needs to be revised to
say that EKL has been tested by its use in CS206. DARPA has to
be sensitive about possible charges that it is supporting education
per se. Thus someone might say that the Stanford professors are
using DoD money rather than Educatin department money to support
their classes, and this is unfair both to defense and to Upper
Podunk College, which doesn't have DoD money.
∂26-Jan-83 1824 JMC
To: TRG
Suggest you check and then send * a message about what authority plans.
∂27-Jan-83 1309 JMC
To: DFH
campbe.1
∂27-Jan-83 1528 JMC
To: DFH
daily.10 is a letter to the Stanford Daily.
∂27-Jan-83 1557 JMC
To: DFH
cory.1,paulso.1
∂27-Jan-83 1756 JMC world library
To: reddy@CMU-CS-A
The files librar[s79,jmc] and librar[e82,jmc] contain my ideas
about the world library. The latter is a partial proposal draft.
∂27-Jan-83 1929 JMC
To: HPM
I agree with your estimate of the interview - foutaise, connerie.
∂28-Jan-83 2333 JMC sentences
To: phil-sci@MIT-OZ
My grade school teacher told me that the virtue of sentences was that
a sentence expresses a complete thought.
A philosopher could undoubtedly have made her feel
foolish by showing that she didn't have a clear idea of what constitutes
a complete thought. Nevertheless, the grade school teacher was right.
Complete sentences are less context dependent
than incomplete sentences. Speaking or writing complete sentences
reduces the probability of presuming more context than the listener
or reader actually.
This is why careful speech and writing in all languages involves
complete sentences.
I believe that children could, in an unethical
experiment, be brought up in a language that did not involve complete
sentences. If they were then left alone, one of them might invent
the concept of complete sentences and persuade his fellows of their
advantages in avoiding misunderstanding.
It seems to me that archeology tells us that the first written
languages did not have complete sentences and were therefore useful
only for limited communication - the dates of kings and battles and
the contents of warehouses.
∂29-Jan-83 1405 JMC typology
To: uw-beaver!ubc-vision!Reiter%LBL-CSAM@MIT-MC
This message is mainly a check to see if the path works,
but I've some immediate comments on your typology.
1. I agree that in general circumcription doesn't cover probabilistic
reasoning, although perhaps probabilistic reasoning in its usual
form may well be treated non-monotonically. My intent was to cover
a limiting case. This disparity isn't worth too much attention by
either of us, because it isn't clear that it concerns a substantial
point of how to design systems.
2. Does one of your categories cover Occam's razor? Also, and I'm not
sure this isn't Occam's razor again, does one of them cover a conjecture
that there isn't a bridge when this conjecture is made on the basis of
seeking the simplest model of the facts rather than seeking for a typical
model?
In order to check out the path, please acknowledge receipt even if you
haven't any comments on the substantive matters.
∂29-Jan-83 1523 JMC
To: jcm@MIT-OZ
Did you mean "jives"? Wow, man.
∂29-Jan-83 1627 JMC innateness, sentences, etc.
To: phil-sci@MIT-OZ
Chomsky and Piaget and others consider many aspects of human
behavior to be innate, often disagreeing about just what is innate.
Even if each of them has defined, as exactly as he can, what he means
by innate, we may be too lazy to find the precise reference. It is
unsound, however, to attribute the silliest meaning we can think of,
even though it may temporarily massage the ego. It is a much better
approximation to attribute the most sensible meaning we can think of,
although if precision is sought, there is no substitute for reading
the literature.
Piaget, I think, thought that certain concepts were innate
in the sense that they arise at a certain stage of development of
all normal humans. I don't know what qualifications he made about
how normal the environment had to be.
With regard to sentences, Minsky's off hand reference to
what I said was correct - if I interpret him correctly. I doubt
that sentences are innate in the following sense. 1. If adults
brought up a child never uttering sentences, the child might not
come to use them. 2. A population of children initiallized without
sentences would develop them in time. I have no opinion about whether
this development would take place at the age of ten in the first
generation or would be an invention after several generations. I
suspect this differs from Chomsky's opinion, because his ideas about
innate universal grammar involve sentences.
To the extent that I understand Chomsky's argument I consider
it faulty. He argues that an innate universal grammar is required,
because a child acquires the grammars of their native language from
a small amount of experience, and this grammar permits judging
the grammaticality of an indefinitely large collection of sentences.
It seems possible to me that a major evolutionary step towards
human intelligence occurred when the output of a pattern recognition step
could be fed back into the input and combined with early data.
This is a step beyond the simple chain suggested by anatomy, where
the first visual or auditory cortex passes signals to the second,
which transforms them and passes them on but never back to its own
inputs. However, this capability is needed for thought processes
apart from language and might be a general intellectual mechanism
developed earlier than language. The Chomsky strategy of studying
grammar first and thought later wouldn't uncover it. Perhaps I
misrepresent Chomsky's point of view. In principle, the point is
testable by looking for either behavioral or anatomical evidence
for such feedback processes. For example, mental goal-seeking is
often a top down
process analogous to top down parsing. "In order to achieve C, I
need to perform an action that has preconditions B and B', which requires
actions that have preconditions ... ".
DAM interprets me correctly as believing that some early written
languages lacked sentences, although the oral languages of the
same people included sentences. In fact this is known. The oral
language of the Aztecs included sentences, although their inscriptions
didn't. Much that is known of their culture was written by Aztec
priests, etc., after the Spanish taught them to write their own
oral language in the Latin alphabet.
I am even willing to entertain the possibility that the first oral
languages didn't have sentences. I would also suspect that some present
day primitive oral languages are impoverished in some respect, especially
in the way they are ordinarily used. The anthropologists and bible
translators who study these languages may mistakenly impose European
linguistic categories on them.
"I bet GAVAN make big fella mistake Friday" in his speculation
about the Creole languages - for any of the possible meanings of "Creole".
Consult Webster's Collegiate for at least four. The Creole languages
weren't invented by masters for the use of slaves. Moreover, I'll bet
that all of them can distinguish past events from present events when this
is necessary for the communication without ever using tenses.
Apart from that it would be a mistake for GAVAN to suppose that we
advocates of the correspondence theories consider ourselves defeated. We
merely have trotted out all the arguments we care to and don't want to
repeat ourselves. Therefore, it is pointless for him to repeat flat
assertions he has made already about the meaningless of
correspondence statements.
∂29-Jan-83 1649 JMC proposal
To: JK
Come to think of it, we need a lot more. We have been asked to propose
for a three year continuation. Therefore, we have to divide up the
tasks according to what will be accomplished in what year. Sorry!
∂29-Jan-83 2111 JMC protection?
To: hewitt@MIT-OZ
∂29-Jan-83 2051 Carl Hewitt <Hewitt at MIT-OZ at MIT-MC> protection?
Received: from MIT-MC by SU-AI with NCP/FTP; 29 Jan 83 20:51:14 PST
Date: Saturday, 29 January 1983, 23:50-EST
From: Carl Hewitt <Hewitt at MIT-OZ at MIT-MC>
Subject: protection?
To: John McCarthy <JMC at SU-AI>
Cc: hewitt at MIT-OZ at MIT-MC, Hewitt at MIT-OZ at MIT-MC
In-reply-to: The message of 23 Jan 83 03:14-EST from John McCarthy <JMC at SU-AI>
John,
It looks OK to me. Are you still getting the messages?
Cheers,
Carl
I think not, but only some of my messages elicited them, and I haven't
sent much to phil-sci recently. Thanks for looking into the problem,
and I'll let you know if it recurs.
∂30-Jan-83 1233 JMC
To: dam@MIT-OZ
CC: phil-sci@MIT-OZ
It seems to me that DAM is doing those who have not studied
mathematical logic a disservice with his substantially correct but
informal explanation of first order indistinguishability of structures.
The ideas are precise and beautiful, but they require some study for
understanding. A person with a completely accepting attitude can get some
notion from such a non-technical exposition, but a person who finds them
in conflict with his own ideas will haggle vaguely rather than readjust
his intuition.
The best source I know for what can and cannot be formalized in
first order logic is the first chapter, by Jon Barwise, in the Handbook of
Mathematical Logic. Even that requires a slight acquaintance with group
theory in order to understand the examples. Unfortunately, the paperback
edition of the book, the rest of which is much more technical, costs $39.
Good beginning books are Robert Rogers's "Mathematical Logic and
Formalized Theories" and Patrick Suppes's "Introduction to Mathematical
Logic"; the former is specifically oriented toward philosophers. The text
most referred to by mathematical logicians is Joseph Shoenfield's
"Mathematical Logic".
Philosophers have a much greater tendency than mathematicians and
scientists to base their work on past formulations. It seems to me that
one reason, besides habit, is that the ideas are unclear. If you don't
read and quote Aristotle, someone will accuse you of getting something
important wrong. No-one will accuse you of getting calculus or Newtonian
mechanics wrong just because you haven't read Leibniz or Newton or
propositional calculus wrong because you haven't read Boole. This
phenomenon is a weakness of philosophy.
My own opinion is that ideas from mathematical logic, computing,
and even artificial intelligence research are essential to anyone who
wants to study epistemology. I also believe that its problems will be
solved as decisively as Newton solved the problems of mechanics, and the
proof will be computer programs that can make scientific discoveries.
Reading past philosophers, and probably even present day artificial
intelligence researchers, will not be necessary in order to understand how
the programs work, although it will be needed to understand the history of
the subject. There is an enormous amount of somewhat relevant past
philosophy, but it is probably a better strategy to concentrate on recent
work and above all, to think directly about the problems rather than about
winning debates on behalf of one's already held beliefs.
Incidentally, the results DAM cites on indistinguishable
structures depend heavily on using first order logic with the further
restriction that the only individuals in the domain of the logic are the
elements of the set being studied. An example is the statement, proved in
Barwise, that any sentence true in all torsion free groups is true in some
group with torsion, and therefore torsion freeness is not a first order
concept. Torsion freeness can be formalized in second order logic or in
set theory or even in a different first order theory in which subgroups
are permitted to be objects. Why the interest in first order logic then?
Because it is complete (Godel), and because those concepts that have first
order formalizations are worth distinguishing for mathematical reasons. I
am sure, however, that AI will not want to restrict itself to elementwise
first order formulations, and I doubt that they will have any special
importance for AI.
A final recommendation: Aaron Sloman's book "The computer
revolution in philosophy". Most of what he says, I agree with.
∂30-Jan-83 1240 JMC my error
To: dam@MIT-OZ
CC: phil-sci@MIT-OZ
I misquoted the result on indistinguishability. The correct statement is
"The set of first-order sentences true in all torsion abelian groups is
true in some abelian group H which is not torsion". These are first order
sentences where the individuals are the elements of the group and the constant
symbols of the language are just the group operation and the identity element.
∂30-Jan-83 1515 JMC
To: vardi@SU-HNV
Do keep me posted on the evolution of your typology and, of course,
anything you write. My new net address is:
"uw-beaver!ubc-vision!Reiter"@lbl-unix
The double quotes are necessary. If that doesn't work try
"uw-beaver!ubc-vision!Reiter"@lbl-csam
With best wishes,
Ray.
∂30-Jan-83 1657 JMC innateness
To: dam@MIT-OZ, minsky@MIT-OZ
CC: phil-sci@MIT-OZ
I don't see a need to make each science self-contained as laudable. I
just was chairman of an economics PhD oral in which the candidate and
the professors hypothesized about what information corporations
find it optimal to give customers, ignoring the fact that there is a
huge (business school) literature on marketing covering precisely this
point. Economists often mistakenly treat technology as a capital good
which a firm buys a certain quantity of - ignoring any specific characteristics
of specific inventions and processes.
In the present case, it isn't laudable for linguists to ignore
the relation between intelligence and problem solving on the one hand
and language on the other.
∂30-Jan-83 2216 JMC There you go again, Gavan.
To: gavan@MIT-OZ
CC: phil-sci@MIT-OZ
A string search indicates that your latest message is the third in which
you have referred to the correspondence theory as a religious
dogma. Perhaps two more tries and we'll all admit it. I'll
have more to say in favor of the correspondence theory when I
have something new to say about it.
∂31-Jan-83 1818 JMC CORRESPONDENCE, etc. and meta-epistemology again
To: phil-sci@MIT-OZ
This rather long exposition is not intended primarily as an argument
for the correspondence theory of truth, although it presents my
position on the relations among CORRESPONDENCE, COHERENCE and CONSENSUS.
It is addressed primarily to people who are already base their thinking
on some kind of correspondence theory and outlines some research ideas.
The primary idea is that abstract meta-epistemology is worth studying.
By abstract I mean that we consider a knowledge seeker in a world,
and we consider the effectiveness of strategies as functions of the
world. For this purpose, it is often appropriate to consider model
worlds that are not candidates as theories of the real world.
By studying strategies in abstract worlds, both theoretically and
experimentally, we may develop candidate strategies for application
to the real world. These candidate strategies may discussed as
philosophies of science and imbedded in programs interacting with
the incompletely known physical and mathematical worlds.
Even meta-level questions such as the appropriate theory of truth may
be studied in these abstract systems.
Here are my views on the relations between CORRESPONDENCE, COHERENCE,
and CONSENSUS baldly stated. Arguments are later.
1. The truth of a statement about the world is defined by its
CORRESPONDENCE to the facts of the world. The truth of a statement about
mathematical objects is determined by its correspondence to the facts
about these mathematical objects. Of course, both of these presuppose the
existence of the world and of mathematical objects. Tarski says: "Snow is
white" is true if snow is white. This has an unfortunate but inevitable
circularity, because we use language for talking about the world, and
we're talking about sentences in the same language. The circularity has
the consequence that the definition doesn't itself provide a means of
determining facts about the world. Unfortunate but inevitable.
2. Our means of trying to determine the truth involves the COHERENCE
of large collections of statements including reports of observation.
We do not take COHERENCE as the definition of truth, because we always
want to admit the possibility that a collection of statements may
be coherent but wrong. Naturally, we will only come to believe that
it is wrong if some other collection of statements is found to
be more COHERENT, but the new one may be wrong also.
3. CONSENSUS is a mere sociological phenomenon whereby groups of
people come to more or less agree about the truth of somm collection
statements. At any given time there may or may not be CONSENSUS in
various groups of people.
Meta-epistemology again:
A Toy mathematical example illustrating use of the above concepts:
Consider a mathematical system consisting of a computer
C, a language L, and a collection D of interacting automata
to which C is connected. We suppose that the language L includes
a predicate symbol holds, and we interpret holds(i,s,t) as
asserting that the i th subautomaton of D is (was) in state
s at time t. We further interpret a certain list B
of sentences in the memory of the computer as the list of what
the program BELIEVES. Sentences elsewhere in memory are considered
mere data. We suppose that the automaton system, including the
computer part is started in some initial configuration. At some
times during the operation of the system, certain sentences will
be in the list B. Suppose holds(17,5,200) is in that
list at some time t1. We regard it as true, and the program as
BELIEVING it correctly if subautomaton 17 is in state 5 at time 200.
In fact, whether the program BELIEVES it is irrelevant to its truth,
since its truth depends on the evolution of the automaton system,
in interaction with the program, and not on the contents of the list
B. However, our interest is in designing knowledge seeking programs,
and we are interested in what programs connected to what automaton
worlds will have lots of true beliefs.
One important class of programs, to be compared in effectiveness
with others, are programs that use data structures interpretable
as sentences about the world, mathematics, goals, etc. - in short
the kind of program now used in much AI research. The program
may be provided with an initial stock of sentences. Some of these
sentences may be regarded as presuppositions about the kind of
world to which the program is connected. Of course,
it wouldn't be interesting to include such assertions as holds(17,5,200)
in the presuppositions, and then admire the result if
the program moves the sentence to the list B. In evaluating
programs, it would be most interesting to consider connecting them
to a variety of automaton systems in a variety of initial states.
Remarks:
1. There will be programs that can be ascribed more true
beliefs if we use a different language and some other location
than in the list B. Indeed programs that evolve intelligence
are unlikely to use this specific langauge. However, since we
are talking about DESIGNING the program, it is difficult enough
to make it smart in the way we intend and quite unlikely that
it will turn out to be smart in some entirely different interpretation.
Therefore, we'll stick to the language L and the list B.
2. Finite automaton worlds are discussed as an example only.
If I were smarter and I thought your patience were greater, I would
have the program interacting with systems more like those discussed
by current theories of physics. Even within the automaton model,
there are more interesting kinds of assertions than holds(i,s,t)
which is rather like an assertion that a particular molecule
has a certain position and velocity at a given time. Assertions
about the structure of the system of automat, e.g. what is connected
to what more closely resemble present day scientific assertions.
Indeed the "obstacles and roofs" world that I mentioned earlier
is EPISTEMOLOGICALLY and HEURISTICALLY more like our own world.
The automaton model is only METAPHYSICALLY ADEQUATE for our present
purpose. (These terms are used in the sense of McCarthy and Hayes "Some
Philosophical Problems from the Standpoint of Artificial Intelligence").
3. What kinds of programs should we design? This depends
on what kinds of automaton systems we intend to connect to the
program. If we connect it to worlds that behave like those
that behaviorist psychologists were in the habit of connecting to
their rats and sophomores stimulus-response models of the world
will be fine. Indeed the sentences of the form holds(i,s,t) may
be quite superfluous for success in worlds designed by behaviorists,
and sentences like responds(s,r) interpreted as "If I give it
the signal s I will get back the response r" may be more
appropriate. In terms of its ability to predict, a correspondence
concept of truth will be irrelevant, because the behaviorist will
have designed his automaton system to give the intended responses
to the stimuli, and the actual mechanism whereby this is done
will be hidden from the program.
However, we might consider designing programs of a different
kind. These programs would hypothesize very large systems
of interacting simple automata connected in a regular way and
such that the inputs to the program were the result of averages
of large numbers of "microscopic" events. The states and transitions
of the individual microscopic automata would not affect the
inputs of the program, i.e. would not be observable.
Nevertheless, the laws connecting the individual "microscopic"
automata would be permanent features of the world and could
be used to explain and predict otherwise unpredictable events
that were more directly observable.
In worlds that I would design, such strategies would
be more effective than strategies hypothesizing stimulus-response
laws. I would design such worlds for my program, because I
believe such worlds are moe like the world to which I am connected
and of which I am a part.
4. If we give the program worlds composed of interacting
parts, sentences interpreted as asserting that the world is so
constructed would be true. Moreover, research programs aimed at
discovering such parts, their internal structures and their interactions
would be likely to generate true beliefs, and would be more successful
than other strategies in predicting the experiential consequences
of actions. This is almost tautologous, since if we connected a
program a world constructed of interacting parts, its beliefs will
be true if they assert this fact, and its predictions of the experiential
consequences of actions are more likely to be correct if the strategy
takes into accoun the facts. Therefore, this is not evidence for
the appropriateness of a correspondence theory of truth in dealing
with human experience. However, if humans have in fact evolved
in a world composed of interacting parts, then considering
epistemological models of the kind proposed here can help us
devise intelligent strategies for learning programs.
5. Besides its "official beliefs" which I would have the
program exhibit for my inspection on the list B, the program
would keep on various lists many other kinds of sentences "about" the laws
of interaction of the automata making up the world. We could inspect
these lists and try to interpret the sentences as assertions about
its world. Sometimes we would succeed and interpret the sentences
as true or false. Sometimes we would fail and say that a certain
sentence has no clear interpretation because the concepts are
confused. For example, some sentence might be analogus to one
about how much phlogiston a rat produces per day.
6. In certain kinds of world, the best strategy for
accumulating beliefs would be a COHERENCE strategy. The strategy
would have collections of assertions about large numbers of aspects
of the world, some of which would be alternatives to each other.
A strategy that put in the list B of official beliefs the
most COHERENT collections of assertions would probably be
most effective in generating beliefs that CORRESPOND to its
world. It would also be most effective in predicting the experiential
consequences of actions.
7. If the knowledge seeking program were composed of many
semi-independent subprograms, each connected to the automaton world
in a different way, strategies of co-operation might well develop.
Such strategies might involve inter-knower lists of of beliefs
obtained by CONSENSUS. This is especially likely if the individual
knowers were limited by short lives from independent access to the
phenomena and so were forced to develop collective institutions
of science.
8. So far our epistemological statements have all been at
the meta level. We have discussed the beliefs and truth seeking
strategies of the programs in the automaton world from outside
that world. If the world is complex, and complex worlds are the
primary interest, it will sometimes be effective for the program
itself to have theories of truth and belief and use these theories
in its knowledge seeking strategy. We might, for example, include
sentences expressing such meta-beliefs in the initial supply of
sentences we give the program. We might include the whole general
theory including a CORRESPONDENCE theory of truth, a COHERENCE
strategy of search and a CONSENSUS theory of co-operation
in the initial stock of sentences provided we could formalize
it suitably. We might try out rival theories, suitably formalized.
Alternatively, we might leave out any theories and see if they
develop.
9. As long as we provide a language L and examine what
sentences in it appear in the list B, we minimize our problems
of interpretation. However, if the system develops other languages,
or if we adopt some more "natural" approach than having a L and
B, we will have problems of whether certain data structures can
be interpreted as sentences making assertions about the world, i.e.
in inventing a translation rule into the language L or whatever
language we use for describing the world. However, I don't think
we will face a problem of having to alternate translations that
both "make sense". As I said in a previous message, cryptography
experience and the Shannon theory suggest that such problems are
extremely unlikely provided we take symmetries and isomorphisms
into account. My paper "Ascribing Mental Qualities to Machines"
discusses some of these points.
10. Symmetries and isomorphisms of the world or parts of
it raise interesting problems. The world to which we connect
the computer may have symmetries and it may be isomorphic to
structures other than those we design. The program may consider
rival theories and then discover that they are isomorphic.
If we consider final theories of the whole world, the preferred outcome
is clear. It should find the isomorphic theories and
recognize their isomorphism. Moreover, many isomorphisms
can be kept implicit by using a formalism that is canonical
with respect to the transformations involved.
However, we are not primarily interested in programs
that will create a final theory of the world write it in list
B and then stop. More facts may break an isomorphism, so the
machine must be more sophisticated. On the one hand, it can't
spend time trying to decide between theories isomorphic with
regard to the means it has for interacting with the phenomenon
involved. On the other hand, it needs should keep the equivalent
theories on hand just in case the equivalence breaks down later.
11. All this is methodology intended as a guide to research.
There are two directions in which research might proceed, theoretical
and experimental. On the one hand, we can develop theories of
what can be found out about what kinds of worlds. E. F. Moore's
"Gedanken Experiments with Sequential Machines" in Automata Studies
should be read by anyone who contemplates research in this area.
Its merit is that it makes important distinctions and proves some
theorems about investigating automaton worlds. Its fault is that
these worlds have too little structure for a sophisticated research
strategy to be effective. They aren't as bad as the stimulus-response
worlds, however, since at least they contain memory.
I fear that it is beyond our present knowledge to formulate
sophisticated conjectures about the effectiveness of different theories
of truth in guiding research in automaton worlds. I suppose the
theoretical state of meta-epistemology is that we need to work on
establishing interesting conjectures.
Experimental research in this area seems inappropriate at
present until there are some conjectures. For example, a program
for solving "obstacles and roofs" worlds might turn out to be just an
exercise in programming. I would also be uninterested in a proof
that "obstacles and roofs" is NP-complete.
12. It might be interesting for an adherent of the COHERENCE
theory of truth to attempt a meta-epistemological model. I wouldn't
know how to begin. He might start in the same way as I did - consider
systems consisting of a computer program connected to something with
which it interacts. My knowledge seeker attempts to find out the
structure of the something. However, just considering the knowledge
seeker connected to something involves a something, i.e. a world,
and makes the problem one of finding out about the world. Someone
who rejects "the world" and an associated correspondence theory
might well consider that these have already been presumed by
connecting the program to something. Well, that's their problem.
Perhaps even Gedanken experiments are inappropriate from their
point of view.
Here is another way of putting the question. Are Gedanken
experiments or real experiments with knowledge seeking programs
appropriate from the point of view of any non-correspondence
theory of truth? If so, what is the experimental environment of
the program, and what kinds of sentences does it attempt to ascribe
truth to? Would an obstacles-and-roofs world be appropriate,
or does it presume to much of a "real world"?
13. Finally, I hope for some reaction, which is why I wrote
this. The reaction I hope for, isn't primarily further debate on
the correctness of the CORRESPONDENCE theory or even applause for
stoutly maintaining it, although I am willing to take part in limited
further debate. Aside to GAVAN: I have not specifically attacked
coherence or consensus theories, because I have not formulated straw
men to be attacked. However, if you formulate something to attack,
I'll attack it if I disagree with it.
I mainly seek reaction to the idea of research in
abstract meta-epistemological models, i.e. the theory of knowledge
seeking programs connected with abstract worlds. Are there interesting
conjectures about what strategies and what presuppositions
will succeed in what worlds? Are experiments and appropriate,
and which?
Also, I would like to know if people find the ideas clear and/or
interesting or whether they require more detailed exposition to be even
comprehensible. This length is my limit for this forum, but it may be
appropriate to try to develop more specific research questions if there is
interest.
∂31-Jan-83 2030 JMC narrowness
To: dam@MIT-OZ, phil-sci@MIT-OZ
Economists should take the specifics of technology into account, and
linguists should take semantics into account in studying parsing.
The trouble isn't so much that some theories don't take into account
facts that aren't pure economics or linguistics, as the case may be,
but that the fields develop methodologies in terms of which it is seen
as wrong to go outside.
∂31-Jan-83 2350 JMC criticism of coherence and consensus
To: gavan@MIT-OZ
CC: phil-sci@MIT-OZ
As I remarked in my long message of 1818PST, which you may not have
got to yet or noticed the aside to you in it, I will give my opinions
of coherence and consensus if you give me a summary of your views,
or if you would rather references to previous messages or the literature.
I have read the article on the coherence theory in the Encyclopedia of
Philosophy, but the author of the article doesn't seem much more friendly
to the idea than I am, so I would prefer to criticize a presentation
by a partisan of it.
The index to the Encyclopedia mentions consensus only in
connection with the consensus gentium argument for the existence
of God, so I suppose the "consensus theory of truth" is due to Kuhn
or Feyerabend or someone like that. I don't promise to pursue
them very far, because, believe it or not, I am trying to cure myself
of being a controversialist, and will go only to limited lengths in
trying to win arguments. The little I have read of Kuhn has left me
with the impression that there is unlikely to be anything useful for
AI in what he says. I would, as it happens, find a reference to Putnam
more interesting, and I have found my copy of volume 2.
∂01-Feb-83 0040 JMC
To: gavan@MIT-OZ
Is "Reason, Truth and History" a book or paper? If a paper, where?
∂01-Feb-83 1040 JMC Paper
To: DFH
∂01-Feb-83 0648 John Batali <Batali at MIT-OZ at MIT-MC> Paper
Received: from MIT-MC by SU-AI with NCP/FTP; 1 Feb 83 06:48:41 PST
Date: Tuesday, 1 February 1983, 09:48-EST
From: John Batali <Batali at MIT-OZ at MIT-MC>
Subject: Paper
To: JMC at SU-AI
Could you send me a copy of your paper "Ascribing Mental Qualities to
Machines" that you mentioned in your recent message?
Or provide a fecund pointer?
John Batali
MIT AI Lab
545 Technology Square #753
Cambridge, MA 02139
Thanks thanks thanks thanks thanks thanks thanks
∂01-Feb-83 1254 JMC Ascribing ...
To: DFH
I seem to be out of "Ascribing Mental Qualities to Machines". Can you
get another 10 from publications?
∂01-Feb-83 1615 JMC
To: JRP
Philosophy Library catalog lacks Putnam's "Reason, Truth and History".
∂01-Feb-83 1845 JMC
To: LGC
No, just the principal investigators of whom there seem to be enough to
fill the room. What I do want is a three year proposal, with the
usual goals and milestones. I have been lazy and should have been
after you before, but I was expecting to encounter you. A two page
draft for my own use some time tonight as a message would be helpful.
∂01-Feb-83 2213 JMC checking circuit
To: HAP
If it is equally convenient for you, it would be better if you
would come to my house between 2pm and 3pm, because Carolyn will
be home anyway to receive a piano tuner. If not, 3:45 as planned,
meeting in my office, will be ok.
∂01-Feb-83 2252 JMC library
To: DFH
Please telephone Green Library and cancel my request for
"Reason, Truth and History" by Hillary Putnam; I bought the book.
∂01-Feb-83 2341 JMC loneliness
To: philosophy-of-science-request%MIT-MC@USC-ECL
I haven't received any mail from phil-sci for a day, and I suppose it
is related to MC dropping NCP. Perhaps the list needs a new routing
to JMC-LISTS@SU-AI. Or perhaps GAVAN going on jury duty has sequestered
all the philosophers.
∂02-Feb-83 0135 JMC
To: gavan@MIT-OZ
Where in "Reason, Truth and History"?
∂02-Feb-83 1440 JMC messages
To: gavan@MIT-OZ, gavan%mit-oz@SU-SCORE,
gavan%mit-oz@USC-ECLC
I received your messages via both udel and eclc. SAIL is waiting
for the TOPS-10 version of TCP which will require further adaptation.
Mid March is the current guess. I have received no phil-sci messages
for two days, so I suppose that the routing to SAIL hasn't been
revived since MC dropped NCP. I am called JMC-LISTS@SU-AI on that
list and would be grateful if you could arrange a new routing.
Please let me know which of three routings reach you.
I have acquired "Reason, Truth and History" but haven't read it yet.
∂02-Feb-83 1453 JMC
To: ME
There are 25 DD's in use and I got "people waiting" message.
∂02-Feb-83 1455 JMC
To: ZM
Zohar: It isn't a question of agreeing to something. It's a question
of making sure the scope items are comprehensive enough. If you
don't take part, it may turn out that the only approach to temporal
reasoning is Elephant.
∂02-Feb-83 1505 JMC
To: HAP
Many thanks.
∂02-Feb-83 1607 JMC
To: ZM, DCL
ARPA[W83,JMC] contains a sample of scope statement for 2 areas.
∂02-Feb-83 1732 JMC Thanks.
To: GAVAN%MIT-OZ@USC-ECLC
You are back on the phil-sci list, using the eclc routing kludge. We
can alternatively use the udel-relay route if we experience any
difficulties. Let me know if you experience any on your end.
JMC-LISTS received your message via usc-eclc. The last previous
message received from phil-sci was from JCM and was sent at 0433EST 1 Feb.
It had a nice simile about how arguing about Feyerabend's position by
people, none of whom advocated it, was like "watching people watch TV".
I'd like the intervening messages if its not too much trouble.
Incidentally, JMC-LISTS is for PHIL-SCI itself. Single messages
addressed only to me are still to JMC@SU-AI.
∂02-Feb-83 1828 JMC
To: csd.bscott@SU-SCORE
Please go ahead full speed on Gabriel proposal.
∂03-Feb-83 0051 JMC Heat and temperature
To: kdf%oz@USC-ECLC
Please thank Susan Carey for her paper. I hope to use its ideas in
studying how to formalize imprecise concepts and to figure out how
an intelligence can find out that it is confused and try to correct
its confusion.
I don't understand the sentence on page 294, "... as long as only one
substance was involved it did not matter whether one considered heat to
be exchanged in the mixture, or temperature itself, or some entity with
properties of both". It seems to me that unless one always mixes the
higher and lower temperature substances (say water) in the same ratio
(say equal ratios) one will discover the difference between heat and
temperature using a single substance. A small amount of water at high
temperature will raise the temperature of a standard amount less
than a larger amount. Moreover, seems to be a fact of common experience,
as well as the outcome of any experiment that might be made, so one
has to be quite confused in order to miss it. Therefore, it would
seem necessary to look for sources of confusion for the later experimenters
analogous to those found for the earlier.
∂03-Feb-83 0125 JMC
To: gavan%mit-mc@USC-ECLC
Please transmit my thanks for Habermas book.
∂03-Feb-83 1147 JMC phil-sci
To: RWW
philsc[jnk,jmc] has the phil-sci file up to today purged of Stanford
announcements. New stuff continues to come into msg.msg[jnk,jmc].
∂03-Feb-83 1450 JMC
To: DFH
Please prepare grade change forms accordingly.
∂03-Feb-83 1257 JJW 206 grades
I would recommend the following final grades.
Agosta B-
Kopelman B
LeVine A-
Moller B-
I'll get the forms or whatever is necessary to Diana for you to sign.
There are still others who have incompletes; I'll try to contact them
and see how they are progressing on projects.
∂03-Feb-83 1908 JMC Does Robby need the predicate true?
To: phil-sci%mit-oz@USC-ECLC
I have been silent (or rather its equivalent for written communication)
for a while, partly because of net problems, and partly because of having
undertaken to make a considered comment on coherence and consensus theories
on the basis of some references GAVAN gave me. Habermas isn't here yet,
and I haven't got through Putnam's "Reason, Truth and History" in which
he seems at least partly to go back on his earlier realist positions.
However, the colloquy on Robby and Tarskian semantics is too
much for me to stay out of.
1. What use is the predicate true(<sentence>)? It is of little use
applied only to constant sentences, because in almost every
place where "true("Snow is white")" appears, you might as well write
"Snow is white". Its real usefulness is in quantified assertions,
such as
1. Whatever JMC says is true, formalized as:
(all (x) (implies (says jmc x) (true x)).
2. If Smith says something he knows is not true, Smith is a liar.
3. Any true assertion about the blocks on Robby's table follows
from those listed in the variable FACTLIST in Robby's program.
4. Any assertion about LISP programs "proved" by the Boyer-Moore
theorem prover is true.
A common use of the predicate true(<sentence>) will involve
substituting a particular sentence in a general statement involving
true having the form of an implication, then proving the antecedent
(which may involve only a syntactic computation or lookup
in a database on the sentence), concluding by modus ponens that
the sentence is true, and then using the Tarski equivalence to
assert the sentence itself.
Informalized example: Whatever Tarski says is true. I just heard
Tarski say "Snow is white". I believe my ears are working, so
Tarski said "Snow is white". Hence, "snow is white" is true. Hence,
Snow is white.
Therefore, to answer DAM's question, if Robby has to reason
from the form of sentences or has to reason from the source of sentences,
he will have use for the predicate true(<sentence>).
2. In our discussion of Robby's function we may conclude that Robby's
beliefs about what is on the table is COHERENT but not true or that
all the robots have come to a CONSENSUS that snow is black, but they're
all wrong.
3. Question for clarification. Does the CONSENSUS theory of truth
apply to assertions like "The E key on this damn terminal
was sticking for a day last week, and I really should have complained
about it while it was happening, because they can't investigate
a problem that has gone away temporarily"?
∂03-Feb-83 2301 JMC
To: minsky%MIT-MC@USC-ECL
Thanks for your long message with phil-sci discussion. Just got it.
∂03-Feb-83 2307 JMC
To: jcma%mit-oz@USC-ECLC
Thanks for transmission of phil-sci segment.
∂04-Feb-83 1441 JMC Job available
To: JMC
∂03-Feb-83 (from Score BBOARD)
Mail-From: CSD.DORIO created at 3-Feb-83 10:58:56
Date: Thu 3 Feb 83 10:58:56-PST
From: Nancy Dorio <CSD.DORIO@SU-SCORE.ARPA>
Subject: Job available
To: su-bboards@SU-SCORE.ARPA
Stanford-Phone: (415) 497-2273
Programmer wanted to produce a common stock Database Query Program on
an Apple or Atari Computer. Interested parties should contact Jay
Yaudegis at 408 249-3287. Hours flexible, salary negotiable.
-------
∂04-Feb-83 2105 JMC heat
To: kdf%mc@USC-ECLC
and temperature
I don't think your preference for naive (common sense) physics
over automaton worlds is a bug. In the main I share it, but
I use automaton worlds for theoretical and argumentative purposes.
Since we can completely specify the world, the desired outcome of theory-forming
processes is determined, and we can discuss or even test methods
for expressing theories, testing theories or discovering theories.
So far I have used automaton models only theoretically, analogously
to the way physicists use the kinetic theory of gases, i.e. to
derive (speculated on) laws. However, it seems as inappropriate to
represent actual phenomena by automaton models as it would be to
represent an actual gas by giving the positions and velocities of
molecules. Well, not quite as inappropriate.
Common sense physics is more useful since it applies to the real world
and is open ended, but because it is open ended, it's harder.
Notice that heat per unit mass is the quantity that doesn't have
to be distinguished from temperature if there is only one substance.
They never mention this quantity, but it seemed to me that sometimes
it fit what they were saying better than heat itself.
∂05-Feb-83 1719 JMC unique factorization
To: phil-sci%oz@USC-ECLC
Unique factorization is an interesting case. It doesn't take much
experience with factoring numbers to develop an intuition for the
uniqueness of factorization. The reasoning from Peano's axioms
or other first principles is non-trivial. Also the statement
of unique factorization as a formula is non-trivial, and I would
suppose that natural language was used until fairly late. It
is a rather direct consequence of "If p divides ab, then p
divides a or p divides b". This in turn is proved by applying
the Euclidean algorithm to p and whichever of a and b which
p is assumed not to divide (say a), getting 1 = xa+yp, which we
then multiply by b getting b = xab+ybp. Since ab is assumed
divisible by p, this shows that b is divisible by p. The
Euclidean algorithm is gotten from the division algorithm, and
the latter depends on the fact that any m can be divided by
any n leaving a remainder less than n. This depends on the
ordering properties of the natural numbers. A simpler proof is
not to be expected, because the theorem isn't true in many rings
of algebraic integers that don't have division algorithms:
e.g. 6 = 3x2 = (1+sqrt(-5))(1-sqrt(-5)), and both are primes
in R(sqrt(-5)). Have I got that right? The point, if any, is that
mathematical experience can lead to intuitions that may be non-trivial to
verify from what one is inclined to take as the basic facts.
∂05-Feb-83 1725 JMC "knowledge" is useful
To: phil-sci%mit-oz@USC-ECLC
I agree with Marvin that the idea of knowing as justified true belief
has bugs, but I also think that it is a useful idea. The point is
to avoid claiming that a single formalism captures all useful ideas
of knowing. I have a paper in which knowing is axiomatized in such
a way that it is closed under inference. It will have a footnote
suggesting that the reader replace each occurrence of KNOW in the
paper by KNOW7, formalizing KNOW1 thru KNOW6 to suit himself.
When we have a flexible system of handling contexts, then KNOW
itself can be used in its different senses without confusion.
It seems to me that outlawing approximate concepts such as knowledge used
in ordinary language is not likely to be fruitful for either AI or
philosophy. Moreover, simplified versions of these concepts are
likely to be useful. Consider the afore-mentioned Robby:
We may find it advantageous to include in his database that Pat knows
Mike's telephone number and that travel agents know airline schedules.
Notice the following facts about English. We can say, "Pat knows
that Mike's telephone number is 333". We can also say, "Pat believes
that Mike's telephone number is 333". We can say, "Pat knows Mike's
telephone number", and this is a different use of the verb "knows"
than in the previous sentence. English doesn't have a parallel
use of "believes", i.e. we can't say, "Pat believes Mike's telephone
number", but have to say "Pat believes that he knows Mike's telephone
number". In my formalization of concepts as objects (reported in Machine
Intelligence 9), I was tempted to introduce a parallel usage of
"believes", i.e. believes(pat, Telephone Mike). There is no obvious
reason why such a locution shouldn't be present in some natural
languages.
Marvin complains that we can't give an operational definition of "knows".
It would be nice if we could, but the world isn't as we would like
it to be, and I think that confining robots to operationally definable
concepts won't work. In fact, I hope it will be a theorem of abstract
meta-epistemology that systems only using operationally definable
concepts in their thinking will be ineffective in finding out about
their worlds, i.e. perhaps Life World physicists would be unable to
define "glider" and "puffer train" operationally.
Marvin seems to propose dropping "knows" from scientific
language and use "believes" combined with asserting p itself.
However, there are large areas of usage, such as most use of the
above sentences, in which this would be inconvenient, and we would
promptly introduce a term for justified true belief, perhaps not
even defining it precisely. Even so, I might go along with Marvin
if making this concession to his scruples would make it likely
that epistemology for robots would thereby avoid future crises
of the same sort - attacks on approximate concepts. However,
"believe" is also subject to the same kind of attack.
Some philosophers have responded to the demand to base epistemology
on the facts as discovered by quantum mechanics by pointing out
that even the physicists don't think that present quantum mechanics
provides the last word. They advance the slogan, "It shouldn't be
necessary to solve all the problems of science before doing
epistemology". I agree, and advance the further slogan, "It shouldn't
be necessary to solve all the problems of philosophy in order
to do artificial intelligence. My goal is to formalize common
sense epistemology, not to engage in disputes with philosophers.
However, avoiding disputes isn't so easy, when one also wants to
interact with philosophers in order to get some help and when
certain recently discovered philosophical concepts, e.g. "natural
kinds" are of genuine use for AI.
I have been continuously disappointed in the direction this discussion
has taken. I had hoped that a discussion of philosophy of science
undertaken in an AI/computer science environment would concentrate
on identifying usable formalisms, and that the discussants would
advance their own positive ideas of what formalisms might be
useful. In fact, there has been much too little of this, and
much of the discussion is based solely on pre-computer philosophical
ideas. Again let me recommend Aaron Sloman's "The computer revolution
in philosophy" or even Daniel Dennett's "Brainstorms".
∂05-Feb-83 1933 JMC
To: JJW
I'd certainly prefer PROLOG.LOG in my aliased directory.
∂05-Feb-83 2039 JMC
To: cstacy%MIT-MC@USC-ECL
The person to ask about TCP at SAIL is Martin Frost, known as ME@SAIL.
However, his last response to the question was that we are still waiting
for the TOPS-10 version, which is being done at Wright-Patterson Air Forse
Base, and which will have to be further modified for the WAITS operating
system. I believe his estimate was the middle of March.
Your message took seven hours to get here, since it's now 8:30 pm here.
We are using USC-ECLC as a relay now. I'm not getting the copies of
messages addressed directly to me from OZ, although they are apparently
being correctly transmitted to JMC-LISTS@SU-AI, which is my preferred
address for "junk mail", i.e. all mail addressed to lists. Thanks for
your concern, and death to BBN.
∂05-Feb-83 2039 JMC this is a test
To: ME
∂05-Feb-83 2029 Christopher C Stacy <CSTACY@Mit-Mc> this is a test
Received: from UDEL-TCP by SU-AI with NCP/FTP; 5 Feb 83 20:29:05 PST
Date: 2 February 1983 16:15 EST
From: Christopher C Stacy <CSTACY@Mit-Mc>
Return-Path: <CSTACY@MIT-MC>
Subject: this is a test
Received: from MIT-MC.ARPA by udel-relay.ARPA (3.284 [1/5/83])
id AA00723; 5-Feb-83 23:31:09-EST (Sat)
To: JMC.SAIL at UDel-TCP
Cc: JCMA at Mit-Mc, GAVAN at Mit-Mc, CSTACY at Mit-Mc
This should get to you via the kludge relay at UDEL, and you should be
able to reply to CSTACY.MC@UDEL-RELAY if you got it. When do you
expect to have TCP running at SAIL? The Government claims you already
have it up, but obviously they are mistaken. Your machine is, as far
as all but a few remaining sites (none of which are at MIT), gone.
Chris
The person to ask about TCP at SAIL is Martin Frost, known as ME@SAIL.
However, his last response to the question was that we are still waiting
for the TOPS-10 version, which is being done at Wright-Patterson Air Forse
Base, and which will have to be further modified for the WAITS operating
system. I believe his estimate was the middle of March.
Your message took seven hours to get here, since it's now 8:30 pm here.
We are using USC-ECLC as a relay now. I'm not getting the copies of
messages addressed directly to me from OZ, although they are apparently
being correctly transmitted to JMC-LISTS@SU-AI, which is my preferred
address for "junk mail", i.e. all mail addressed to lists. Thanks for
your concern, and death to BBN.
∂05-Feb-83 2109 JMC
To: ME
∂05-Feb-83 2107 MAILER failed mail returned
Because the destination ARPA host(s) are not running NCP,
the following message was undeliverable to recipient(s)
dam@MIT-OZ:
∂05-Feb-83 2107 JMC Tarski si! Tarsky no!
To: dam@MIT-OZ
CC: phil-sci@MIT-OZ
DAM:
∂06-Feb-83 0208 JMC Tarski, si! Tarsky, no!
To: dam%oz%MIT-MC@USC-ECL
CC: phil-sci%oz%MIT-MC@USC-ECL
DAM:
In particular why can't we think of "P is true" as being
equivalent to (unquote P) where "unquote" is a particular syntactic
operator which takes a quotation and removes the quotes. This does
not seem to me to depend on Tarskian semantics. Of course the
function "unquote" will have a recursive definition with rules like
(unquote "P and Q") iff (unquote P) and (unquote Q). But I do not
consider such recursive rules to have much to do with Tarskian
semantics.
My formalizations of knowledge and, incidentally of truth,
were formulated axiomatically at first, and didn't explicitly use
the Tarskian semantics, although any first order theory has Tarskian
semantics that anyone if free to discuss. However, when I began
to try to formalize puzzles involving non-knowledge, such as a
strong version of the puzzle of the three wise men or the puzzle
of Mr. S and Mr. P, I had to use the Kripke semantics of modal
logic in an explicit way, and no modal logician has been able to
show how to solve either of these problems purely within modal
logic. The strong version of "Three wise men" requires that the
king tell his wise men that he will put white or black spots on
each of their foreheads at least one of which will be white. In
fact all three spots are white. He then asks "Do you know the
color of your spot?" three times. The formalization must express
these conditions, and the strong problem is to be able to prove
from the formalization that they answer know the first two times
the king asks, and yes the third time. Unfortunately, these results
haven't been yet put in publishable form, but anyway the semantic
formalization involving possible worlds as objects does the strong problem
nicely.
Your proposed formalization involving unquote isn't described
well enough for me to tell whether it will be more like a modal
formalism or like a first order formalization of the predicate true.
In any case, your proposal has the advantage that arguments about
the meaning of "unquote" may not get as emotional as those about truth,
and such sayings as "Unquote is Beauty, and Beauty is unquote" will
more clearly seem to demand explanation on the part of those who
utter them.
∂06-Feb-83 0217 JMC
To: DFH
kantro.1
∂06-Feb-83 0256 JMC fixing garbles
To: minsky%oz%MIT-MC@USC-ECL
I believe these two paragraphs span the garbles in two of my messages.
I agree with Marvin that the idea of knowing as justified true belief
has bugs, but I also think that it is a useful idea. The point is
to avoid claiming that a single formalism captures all useful ideas
of knowing. I have a paper in which knowing is axiomatized in such
a way that it is closed under inference. It will have a footnote
suggesting that the reader replace each occurrence of KNOW in the
paper by KNOW7, formalizing KNOW1 thru KNOW6 to suit himself.
When we have a flexible system of handling contexts, then KNOW
itself can be used in its different senses without confusion.
Unique factorization is an interesting case. It doesn't take much
experience with factoring numbers to develop an intuition for the
uniqueness of factorization. The reasoning from Peano's axioms
or other first principles is non-trivial. Also the statement
∂06-Feb-83 2346 JMC fixing garbles
To: phil-sci%oz%MIT-MC@USC-ECL
I believe these two paragraphs span the garbles in two of my messages.
I agree with Marvin that the idea of knowing as justified true belief
has bugs, but I also think that it is a useful idea. The point is
to avoid claiming that a single formalism captures all useful ideas
of knowing. I have a paper in which knowing is axiomatized in such
a way that it is closed under inference. It will have a footnote
suggesting that the reader replace each occurrence of KNOW in the
paper by KNOW7, formalizing KNOW1 thru KNOW6 to suit himself.
When we have a flexible system of handling contexts, then KNOW
itself can be used in its different senses without confusion.
Unique factorization is an interesting case. It doesn't take much
experience with factoring numbers to develop an intuition for the
uniqueness of factorization. The reasoning from Peano's axioms
or other first principles is non-trivial. Also the statement
∂07-Feb-83 0013 JMC Philosophy's debt to science
To: phil-sci%oz%MIT-MC@USC-ECL
Philosophy owes science some debts. In principle, this needn't
be the case if philosophy were to confine itself to what could be
determined by thinking, but it seems that we humans aren't smart
enough and have too much tendency to run in intellectual herds. Here
are some examples:
1. Before approximately the time of Galileo, there wasn't much separation,
but most philosophy, and much science, attempted to explain the world
in terms of purpose, i.e. the purpose of the ant is to teach us not
to be lazy, and the purpose of the rainbow is to remind us that God
told Noah that the next time he would destroy the world by fire not
by water. It was only by providing explanations that, as Laplace
put it, had no need of that hypothesis, that science rescued philosophy
from its concern with explanation by purpose.
2. Kant's notions of a priori facts, e.g. Euclidean geometry, were
determined by introspection and later proved wrong. It would be
interesting to know whether it was noted that non-Euclidean geometry
proved him wrong or whether it had to wait for the theory of
relativity.
3. Before Einstein, just about anyone, scientist or philosopher, wouldn't
have noticed that distant simultaneity needn't be axiomatic and that
expressing space-time as the Cartesian product of space and
and time, was possible in many ways, none of which was
canonical. If you don't know what "canonical" means
here, read some mathematics. Hint: the dual of a finite dimensional
linear space is isomorphic with it but not canonically, while the
dual of the dual is canonically isomorphic with the original space.
4. Quantum mechanics tells us that common sense notions of causality
are only an approximation to reality and leaves us with
basic problems about causality that have remained unsolved since
the 1920s.
5. Daniel Dennett's "Can a computer feel pain" in his Brainstorms
points out that the question, "Can a person feel pain without
knowing it?" is a scientific question rather than a philosophical
question.
6. Mathematics has also taught philosophy some lessons. Before
Godel and Tarski, no-one distinguished clearly between truth and
provability. Cauchy destroyed philosophical nonsense about
calculus by Berkeley, and pre-destroyed similar nonsense by Marx.
Cantor destroyed the philosophy of the infinite.
7. Scientists have also helped mislead philosophy. Behaviorism
and logical positivism are just as much the creation of scientists
as of philosophers.
Now that dialectics has been mentioned, I cannot resist
the following quotation, which I believe is from Schopenhauer.
"But the height of audacity in serving up pure nonsense, in stringing
together senseless and extravagant mazes of words, such as had previously
been known only in madhouses, was finally reached in Hegel, and became the
instrument of the most bare-faced general mystification that has ever
taken place, with a result that will appear fabulous to posterity, and
will remain as a monument to German stupidity".
∂07-Feb-83 0131 JMC
To: DFH
Chen.2.
∂07-Feb-83 1801 JMC I also once enumerated the 2 by 2 Turing machines, though not as
To: phil-sci%oz%MIT-MC@USC-ECL
completely as Marvin did, in the hopes that the simplest machines with
universal or other interesting behavior would lead to some new ideas about
what are the simple computing mechanisms. Maybe carrying the enterprise
further would turn up something interesting. However, it reminds me of
the last sentence in a Government pamphlet that gave all the known
techniques for panning for gold in streams.
"A strong man, who isn't afraid of hard work, can make as much as
fifty cents a day doing this".
My informant claimed that while Chinese laid off from building railroads
in the beginning of this century had panned out all the streams in
California, enough years had elapsed since even they gave up for enough
gold to wash down from the mountains so that one could make twenty dollars
a day.
Maybe there are enough new ideas about what might be interesting
to justify another look at small Turing machines and other tiny computers.
Is it interesting that Conway's Life, invented for an entirely different
purpose, can support self-reproducing universal computers? See reference
1. I don't see any justification for dogmatism either way about the
possible reward from this enterprise.
About numbers:
I'm not sure there is much to be gained by studying the concept of
number itself, whether philosophically, mathematically or psychologically.
The probability of empty profundity is rather high. Let me point out
(allege) that a child's first use of numbers is not to state the
cardinality of small sets. Rather it is to count, i.e. recite the number
sequence - a singularly useless achievement in itself. Presumably,
however, the rules for forming the numeral sequence, at first arbitrary
and then having increasing regularity, give a child the notion of
algorithm or at least the idea of a sequence formed by rule. Pleasure in
this exercising this algorithm seems to motivate feats of counting that
require a span of attention longer than a child of that age usually
devotes to such activities.
References:
1. Unsubstantiated M.I.T. rumor.
∂07-Feb-83 2251 JMC possible worlds and elaboration tolerance
To: phil-sci%oz%MIT-MC@USC-ECL
What Robby needs to know about depends on how smart we are trying
to make him, and I think that our first efforts have to be rather
modest, because we don't understand epistemology well enough.
I have been trying to develop an idea I call "elaboration tolerance".
We want Robby to use simple formalisms and then elaborate them
when he has to. One way to do this is to use predicates that
can be "elaborated" into more comprehensive predicates. Thus we
begin with at1(Robby, desk) and suitable axioms for at1. Later
we need at2(Robby,desk), which is an object rather than a truth
value. We say things like knows(Robby, at2(Robby,desk)) and have
an axiom
(x y).at1(x,y) iff true1(at2(x,y)).
Successive elaborations may involve such expressions as
true2(at2(x,y),s) where s is a situation, true4(true2(at(x,y),s),w)
where w is a possible world, etc. The possible worlds can in turn
be be mere objects whose values depend on more comprehensive
possible worlds. If you think this idea is half-baked in its present
form, I agree with you.
However, the AI and philosophical point is that we can use
a possible worlds formalism without having to commit ourselves to
a philosophical doctrine about what (if anything) possible worlds
"really" are. As a mathematical object, a possible world is a
member of a space of possible worlds. Such a space may be a set
the set of equivalence classes of a subset of outer level possible
worlds. Thus in the wise men puzzle, the possible worlds are
characterized by the colors of the spots on the foreheads of
the three wise men. However, we aren't really committing ourselves
to this as the only distinction worth making in the world as a
whole, and it is only in certain outer possible worlds that there
is a wise men puzzle at all.
I think the best strategy is to develop the possible
worlds idea for use in AI, and then to figure out what we have
got. If we try to pin down the idea now, it is likely to turn
out that we will want a somewhat different concept than the
one we have made precise at considerable trouble.
I will look at my numerous files on the wise men and
Mr. S and Mr. P puzzles and decide what to transmit. Since they
are lengthy and many won't be interested, I'll transfer copies
once to OZ and then give a reference in phil-sci. The S and P
puzzle is as follows:
Two numbers m and n are chosen such that 1 < m =< n < 99.
Mr. S is told their sum and Mr. P is told their product. The following
dialogue ensues:
Mr. P: I don't know the numbers.
Mr. S: I knew you didn't know. I don't know either.
Mr. P: Now I know the numbers.
Mr. S: Now I know them too.
In view of the above dialogue, what are the numbers?
The point is that to formalize Robby's reasoning in solving the
puzzle, we need to express the original ignorance in the form
of the existence of sufficient possible worlds.
Incidentally, DAM, if you meant that a possible world
in the wise men problem was fully characterized by the colors
of the spots, you were mistaken. In the Kripke formalism, a
possible world is characterized not only by the colors of the spots,
but also by what other possible worlds are "accessible" (possible)
from it. My own formalization uses a four term accessibility
relation A(world1,world2,person,time), and we can express
"person pp knows proposition p at time t in world w" by
(w1)(A(w,w1,pp,t) implies true(p,w1).
Here's that old devil "true" again.
My axioms for S and P were modified by Ma Xiwen, a visitor
from Peking, and he gave a proof in Weyhrauch's FOL proof checker of the
reduction of the knowledge problem to a mathematical problem.
Unfortunately, I found a bug in his axioms; they don't describe
the problem correctly. While I have fixed that bug I haven't
redone the proof, because I am dissatisfied with the expression
of the axioms for learning. Anyway, I'll shortly transmit
something.
∂08-Feb-83 1230 JMC
To: DFH
carey.1
∂08-Feb-83 1231 JMC
To: DFH
address on letter in file box
∂08-Feb-83 2300 JMC
a244 1603 08 Feb 83
AM-Top-Level Tests,460
Fewer Top-Level SAT Scores
By JOHN RICE
Associated Press Writer
SAN FRANCISCO (AP) - The cream of America's high school crop is
getting thinner, according to college entrance exams, and university
officials said Tuesday there's reason for concern.
''These figures even shocked me when I looked at them,'' said
Stanford University admissions dean Fred Hargadon, a former chairman
of the College Board, which sponsors the Scholastic Aptitude Tests
nationwide.
The decline in average SAT scores has been well publicized and
intensively analyzed. ''I'm not sure anybody's paid as much attention
to the drop in top scores,'' Hargadon said in an interview.
The number of students scoring 650 or higher on the SAT verbal
skills section dropped 45 percent between 1972 and 1982 - from 53,794
to 29,236 - while there was a 23 percent decline in students scoring
650 or higher on the math section - down to 71,916 from 93,868.
Over the same period, the median score slipped to 426 from 453 on
the verbal section and from 484 to 467 on math. The highest possible
score is 800.
The number of students taking the test dropped by only 3 percent and
Hargadon said studies indicate the tests, if anything, may be
somewhat easier today.
''I can't believe students today are inherently less capable than
they were in the past,'' Duke University Professor Robert Sawyer, who
directs a summer program for gifted high school students, said
Tuesday.
''I think probably we tend to overestimate the number of secondary
schools that are either equipped or motivated to teach at a very high
level,'' Hargadon said.
He agreed that a growing number of leisure activities create
potential distractions for students. ''I also think it's pretty clear
that students are required to do considerably less homework than was
the case 10 or 20 years ago,'' he added. ''Homework in many of the
schools is almost unknown now.''
Schools also are under tremendous pressure to save money and to be
egalitarian, Sawyer noted, which means that programs for the few
students who would score over 650 on the SAT are often cut to make way
for programs that benefit a larger group.
And schools are having trouble finding qualified teachers,
particularly in math and sciences.
''There are so many other professions that are more attractive than
teaching these days and young women that typically have gone into
education are not bound to go into education these days,'' Sawyer
said.
California state officials said last week that teachers who took a
basic skills test did much worse than college students training to
become teachers. Only 57 percent of the full-time teachers who took
the test passed it, compared with 71 percent of the students.
Hargadon stressed the need to challenge bright students.
''I happen to believe that if you set the standards high, students
are going to meet them,'' said Hargadon.
ap-ny-02-08 1902EST
***************
jmc - The unmentionable hypothesis is that the long term (maybe
a century old) low fertility of intelligent well educated people
is finally catching up with us. Perhaps the effects were disguised
for many years by increasing equality of opportunity, but eventually
the intelligent from all levels of society were culled. Perhaps
the hypothesis can be checked. If there were a long term
decline in intelligence, then the number of high level IQ scores
when the 1982 SAT group entered the first grade in 1971 would have been
reduced from the number of high IQ scores in 1961. Do such statistics
exist? There is nothing in the Statistical Abstract of the United
States about intelligence at all. If there were just as many smart
first graders, then it really is the fault of the schools.
∂08-Feb-83 2312 JMC Wessels
To: csd.golub@SU-SCORE
I think Quate's current proposal meets my substantive needs. Mike is old
enough so that a four year appointment will bring him into the tenure
decision age range, and anyway if we aren't decided by then, we can
re-open the issue. As I said I don't feel strongly substantively about
Lantz. However, in so far as there is a question of departmental
prerogatives, I will stick with the group.
∂08-Feb-83 2316 JMC
To: bboard
stupid.ns
∂08-Feb-83 2317 JMC
To: bboard@SU-SCORE
stupid.ns
∂08-Feb-83 2320 JMC
To: bboard
a244 1603 08 Feb 83
AM-Top-Level Tests,460
Fewer Top-Level SAT Scores
By JOHN RICE
Associated Press Writer
SAN FRANCISCO (AP) - The cream of America's high school crop is
getting thinner, according to college entrance exams, and university
officials said Tuesday there's reason for concern.
''These figures even shocked me when I looked at them,'' said
Stanford University admissions dean Fred Hargadon, a former chairman
of the College Board, which sponsors the Scholastic Aptitude Tests
nationwide.
The decline in average SAT scores has been well publicized and
intensively analyzed. ''I'm not sure anybody's paid as much attention
to the drop in top scores,'' Hargadon said in an interview.
The number of students scoring 650 or higher on the SAT verbal
skills section dropped 45 percent between 1972 and 1982 - from 53,794
to 29,236 - while there was a 23 percent decline in students scoring
650 or higher on the math section - down to 71,916 from 93,868.
Over the same period, the median score slipped to 426 from 453 on
the verbal section and from 484 to 467 on math. The highest possible
score is 800.
The number of students taking the test dropped by only 3 percent and
Hargadon said studies indicate the tests, if anything, may be
somewhat easier today.
''I can't believe students today are inherently less capable than
they were in the past,'' Duke University Professor Robert Sawyer, who
directs a summer program for gifted high school students, said
Tuesday.
''I think probably we tend to overestimate the number of secondary
schools that are either equipped or motivated to teach at a very high
level,'' Hargadon said.
He agreed that a growing number of leisure activities create
potential distractions for students. ''I also think it's pretty clear
that students are required to do considerably less homework than was
the case 10 or 20 years ago,'' he added. ''Homework in many of the
schools is almost unknown now.''
Schools also are under tremendous pressure to save money and to be
egalitarian, Sawyer noted, which means that programs for the few
students who would score over 650 on the SAT are often cut to make way
for programs that benefit a larger group.
And schools are having trouble finding qualified teachers,
particularly in math and sciences.
''There are so many other professions that are more attractive than
teaching these days and young women that typically have gone into
education are not bound to go into education these days,'' Sawyer
said.
California state officials said last week that teachers who took a
basic skills test did much worse than college students training to
become teachers. Only 57 percent of the full-time teachers who took
the test passed it, compared with 71 percent of the students.
Hargadon stressed the need to challenge bright students.
''I happen to believe that if you set the standards high, students
are going to meet them,'' said Hargadon.
ap-ny-02-08 1902EST
***************
jmc - The unmentionable hypothesis is that the long term (maybe
a century old) low fertility of intelligent well educated people
is finally catching up with us. Perhaps the effects were disguised
for many years by increasing equality of opportunity, but eventually
the intelligent from all levels of society were culled. Perhaps
the hypothesis can be checked. If there were a long term
decline in intelligence, then the number of high level IQ scores
when the 1982 SAT group entered the first grade in 1971 would have been
reduced from the number of high IQ scores in 1961. Do such statistics
exist? There is nothing in the Statistical Abstract of the United
States about intelligence at all. If there were just as many smart
first graders, then it really is the fault of the schools.
∂09-Feb-83 1511 JMC JMC si! JCM no!
To: dam%oz%MIT-MC@USC-ECL
Missspelling Tarski is one thing. Misspelling JMC causes your message
to go astray. Actually I haven't been getting the copies addressed
directly to me of mail addressed to phil-sci even when my initials
are correct, although I do seem to get much individual mail. Moreover,
I look at everything in phil-sci (assuming that everything is properly
forwarded by phil-sci), so I don't really need individual copies of
the messages in my mail file.
∂09-Feb-83 2330 JMC "God's eye view"
To: gavan%oz%MIT-MC@USC-ECL
CC: phil-sci%oz%MIT-MC@USC-ECL
Do you know if this phrase was invented by Putnam? I've decided that
it indeed describes the view that AI researchers should take of
truth for robots, and I'd like to give proper credit.
The book report on Habermas and Putnam will be along shortly.
∂10-Feb-83 0030 JMC
To: ME
The DM is working fine at 1200/1200, but it would be nicer to have
1800, and I'd like to try. Can we try it now or would it be easier
when you are at the Lab?
∂10-Feb-83 0105 JMC Book Report
To: phil-sci%oz%MIT-MC@USC-ECL
At GAVAN's instigation, I have read the translated chapter of
Jurgen Habermas's "Theories of truth", which he kindly supplied, and
Hilary Putnam's chapter "Two Philsophical Perspectives" reprinted in his
book "Reason Truth and History".
The first concerns the consensus theory of truth and the second
concerns an "internal realist" or coherence theory of truth. He hoped
that they would make me reconsider my adherence to a correspondence theory
of truth. I read them hoping for information relevant to making computer
program reason about their own knowledge or to helping computer scientists
reason about computer programs. As you may remember from previous
messages, I believe that Robby will need the predicate true as applied to
sentences (or more abstract entities called propositions).
Often I have found theories developed by epistemologists
useful in AI. Examples include the de re - de dicto distinction,
modal logic, and the concept of natural kinds, the latter a subject
to which Putnam has made contributions. Unfortunately, I could
not find anything usable in either of the present papers. I think
AI will have to start almost from scratch in building a theory of
the relation between minds and the world.
I will begin with Habermas, proceed to Putnam, and then
recapitulate why AI needs a theory of little minds in little
worlds and requires what I would describe as a correspondence
theory.
Habermas:
Habermas offers a consensus theory of truth, the best summary
of which is the following paragraph.
"On this view I may ascribe a predicate to an object if and only if
every other person who could enter into a dialog with me would also
ascribe the same predicate to the same object. In order to
distinguish true from false statements I make reference to the
judgments of others -- in fact to the judgment of all others with
whom I could ever hold a dialog (among whom, contrary to fact, I
include all the dialog partners I could find if my life-history were
coextensive with the history of mankind). The condition of the truth
of statements is the potential agreement of all others. Every other
would have to able to be convinced that I am justified in ascribing
to object x the predicate p and would then have to be able to agree
with me. Truth means the promise of attaining a rational consensus".
- from page 7 of the anonymous translation.
This is certainly different from the common sense notion
of truth, from Tarski, and from my own opinion. According to us,
everyone who ever lived could be mistaken about some question.
Moreover, an assertion or proposition or sentence could be true
if no-one ever thought of it, since truth is a relation between
assertions and the world and has nothing to do with whether anyone
ever thinks of it.
Habermas's notion of truth is imbedded in a theory of
speech acts - a mess for which John Searle of Berkeley bears
a substantial part of the responsibility. Exhibiting this mess
is the following citation from page 6.
"We can summarize the results of our preliminary deliberations in
three theses, which require further explication.
1. We call truth the validity-claim which we associate with
constative speech acts. A statement is true if the validity-claim of the
speech acts by means of which, in employing sentences, we maintain that
statement is justified.
2. Questions of truth arise only if the validity-claims naively
imputed in contexts of action become problematic. In discourses where
hypothetical validity-claims are tested, utterances about the truth of
statements are therefore not redundant.
3. In contexts of action assertions give information about objects
of experience; in discourses stements about facts are discussed.
Questions about truth are consequently raised not so much in reference to
the inner worldly correlates of action - with discourses that are
experience -- and action -- free. Whether states of affairs are or are
not the case is not decided by experiential evidence, but by the course of
argument. The idea of truth can be unpacked only in reference to the
discursive settlement of validity claims.
From these theses I would like to draw a few provisional
conclusions which suggest a consensus theory of truth."
The above citation contains the following terms all of which are
treated as technical terms of his theory: truth, validity-claim,
statement, justified, naively imputed, contexts of action, utterance,
assertion, discourse, facts, information, inner worldly correlate, action
related cognition, experience, experiential evidence and course of
argument. This is just the beginning; the number of technical terms in
this paper must exceed a hundred. It would be unfortunate but ok if
they were really needed to understand what truth is. However, my opinion
is that Habermas, no doubt with help from other obscurantist philosophers
like Searle, is in a mode whereby he keeps adding terminology in order to
try to get out of complications induced by the preceding terminology.
I fear that GAVAN will feel cheated by the fact that I decline
to make an analysis of this paper and prefer to make a few superficial
remarks about the views expressed and then start over. My reason is
that there isn't time. Each term would require a page of speculation
about what it might mean and each sentence would require another page
of analysis. A whole 500 page philosophy PhD thesis could be written
about this one paper, and many have about similarly obscure writings
of more famous philosophers.
Putnam:
Putnam is a lot more comprehensible than Habermas. and his "Two
philosophical perspectives" seems to be where GAVAN got his "God's eye
view" epithet. Putnam's argument has the following character. If the
correspondence theory were true, we would have to admit that we might be
brains in a vat connected electronically to the outside world. But then,
and I elide his argument a great deal, how could we really be referring to
the external world. In other words, metaphysical realism faces a
difficulty in referring.
The argument is intricate, but fortunately it is possible
to say, "So what?". There may indeed be difficulties in referring.
The Life World physicists may be harassed by a life world Putnam
(not to speak of a Life World GAVAN) claiming that their hard won
conjecture that their fundamental physics is a certain two dimensional
cellular automaton is a just a way of referring to a coherent set
of sentences. However, we, with our God's eye view of the Life
World, congratulate the Life World physicists on their success.
To repeat my previous argument. There is nothing about
evolution that guarantees that creatures who evolve intelligence
will have an easy time with epistemology. In the words of a
Moroccan proverb, "The world has not promised anything to anybody".
If I understand Putnam correctly, this makes me an even
more extreme externalist than his straw man. Putnam's straw man
externalist relates terms to things they refer to by some kind of
causal chain. In my view, the world has not promised that there
will be a causal chain between me and all objects that exist,
because there is no law that prevents intelligence from evolving
in a way that leaves it causally cut off from some aspect of
the world. Tough for us. We can't be sure that what we can
possibly discover about science, geography, history
or even mathematics is all there is.
The God's eye view - beginning with little minds in little worlds
Putnam accuses "external realists" of advocating a God's
eye view, and Gavan has accused me of arrogance in this connection.
After reading Putnam's paper, I have concluded that this is an
excellent short description of what I advocate.
Putnam's paper summarizes the history of the philosophy of
reference - e.g. the relation between red or long objects and what
represents them in the mind or brain. The history includes Aristotle,
Plato, Locke, Berkeley, Kant, Wittgenstein and Nelson Goodman.
Locke begins by arguing that whatever may correspond to a red object
in the brain, it isn't red - thus criticizing his predecessors, and
his successors jazz up this idea. Indeed.
Starting with a very little mind, we can also say that whatever
corresponds to a temperature of 72 degrees in a thermostat, it isn't
a temperature. Whatever corresponds to a red apple in the memory
of a computer connected to a TV camera isn't red, isn't round and
isn't tasty. What makes the relation between the data structure
in a person's brain and the objects he sees more problematical?
Nothing except that a human is discussing the relation. We will
face complications when we design a computerized thermostat that
reasons about whether its opinion of the temperature is justified.
Actually there are difficulties in designing an AI system
that can reason about the towers of blocks it sees on the table
with its TV camera. For example, I am having difficulty devising
a formalization that will include abstract designs (a red block on top of
two green blocks), concrete but unrealized designs (block A on
top of block E and block F where these blocks are a red and two
greens), realized designs existing in a the present or some
hypothetical situation. There is also the problem of a tower
retaining its identity while it is being improved, i.e. having
blocks added to it. Unfortunately, nothing Habermas or Putnam
has anything to say relevant to how we must go beyond the
concepts required to understand the simple thermostat in order
to represent facts about the blocks world.
The problem of the relation between information in the
human brain and phenomena in the world is much more complex than
the above-mentioned blocks world that already gives us difficulties.
However, I don't believe that Habermas or Putnam even intend to
help with it. If you have doubts, go back to the above Habermas
extracts and ask how they help with the blocks world or even
with the thermostat. If you believe in the relevance of all those
Habermas concepts, consider writing a LISP program to use them
for some practical purpose.
Unfortunately, this kind of philosophical discussion tends to
distract attention from the real, practical and interesting problems of
epistemology. Indeed the amount of thought already put into the PHIL-SCI
discussion might make a dent in the blocks world problem if it weren't
being dissipated.
Now that I have done a book report on two opuses suggested
by GAVAN, let me suggest to him that he do one on Aaron Sloman's
"Computer Revolution in Philosophy". I fear he won't like it any
better than I liked Habermas, but it will be more productive than
a paragraph by paragraph refutation of the foregoing. I don't agree
with even most of what Sloman says, but it has quite a number of
interesting and useful ideas.
∂10-Feb-83 0212 JMC Addendum to book report
To: gavan%oz%MIT-MC@USC-ECL
CC: phil-sci%oz%MIT-MC@USC-ECL
Indeed there are sometimes too many correspondences - equivalent
as to observable consequences. Putnam's example is the equivalence
between wave theories of electricity and magnetism and
action-at-a-distance theories. The truth may be one or the other
or neither. It may also be that there may be such an isomorphism
that a more abstract view is desirable. The problem is mitigated
by the fact that assertions like, "This electron exerts .01 dynes
repulsion on this other electron" are the same in both theories.
Again we must distinguish between what may or may not be true
and what we may or may not have the opportunity to determine.
Such grand problems won't arise soon with Robby.
∂10-Feb-83 1852 JMC
To: CLT
I went home. Please phone.
∂11-Feb-83 1114 JMC Course for IBM
To: csd.armer@SU-SCORE
I'm interested if someone else will organizes it. I could teach either
an introduction to program verification or the formalization of common
sense knowledge - preferably the latter.
∂12-Feb-83 1214 JMC
To: DFH
I swiped your little yellow memo sticker pads.
∂13-Feb-83 1939 JMC
To: buchanan@SUMEX-AIM
I can meet with Fujimoto in the late afternoon, more or less at his convenience.
∂14-Feb-83 1103 JMC
To: DFH
Fujimoto time ok.
∂14-Feb-83 1452 JMC
To: csd.armer@SU-SCORE
CC: CLT@SU-AI, LGC@SU-AI
Please add the following to the scope document for NAVELEX and ARPA.
1.5 Formalization (both procedural and declarative) of the heuristic
knowledge used in commonsense reasoning and problem-solving.
∂14-Feb-83 1607 JMC
To: csd.armer@SU-SCORE
CC: CLT@SU-AI
Carolyn Talcott is co-ordinating proposal material for us.
∂16-Feb-83 1654 JMC
To: SM%SU-AI@USC-ECL
Sorry I missed your message. Les's company, Imagen, is in phone book.
∂16-Feb-83 1911 JMC Computers & Standards--new journal
To: admin.library@SU-SCORE
We do need "Computers and Standards".
∂16-Feb-83 1206 ADMIN.LIBRARY@SU-SCORE Computers & Standards--new journal
Received: from SU-SCORE by SU-AI with PUP; 16-Feb-83 12:06 PST
Date: Wed 16 Feb 83 12:05:46-PST
From: C.S./Math Library <ADMIN.LIBRARY@SU-SCORE.ARPA>
Subject: Computers & Standards--new journal
To: su-bboards@SU-SCORE.ARPA
North Holland is publishing a new journal titled Computers and Standards which
began in 1982. It appears to be a quarterly. The following is from the blurb:
Topics covered in C&S include office automation, CAD/CAM, robotics,
programming languages, database and information systems, computer graphics, I/O
interfaces, nedia, data dictionary/directories, documentation, data elements,
data encryption, operating systems, storage and interchange media, and other
Issues relating to the impact of computer standards on technology, economics,
trade etc. are included. The major aims of C$S are to inform the international
EDP community about computer standards work at the international and national
levels and to provide an international forum for discussion of computer
standards. The editor in chief is John L. Berg. Editorial board includes Bill
Hollowary, M. Kato, Phil Locke, J.M. van Oorschot, Norman J. Ream, John Riganati,
Fawzi Hassan Hakeem, K. B. Fingerle, Reiner Durchholz, Grayce M. Booth,
Robert Bemer, Michael Blasgen, and Jeanne C. Adams.
Rcent articles have included: A guide to DBMS Standardization Activities;
The activities of the BCS/CODASTL DDLC DBAWG; ECMS the European Computer
Manufacturers Association; Information Processing Standards; the IBM PC Stan
dards as Marketing Strategy; Implications of the Supreme Court Decision
in the Hydrolevel Case; Report of the first meeting of ISO/TC97/SC18
text preparation and interchange; An examination of standards and practices
for software production; the World of EDP standards; Communication Protocol
and Service Standards.
This journal would cost $88 a year. Are you interested in the Math/CS Library
purchasing this title? Let me know if you feel this title is needed for
your research needs here at Stanford.
Harry Llull
-------
∂16-Feb-83 1913 JMC
To: admin.library@SU-SCORE
Sorry! We do need Computers and Standards.
∂17-Feb-83 1656 JMC interested in your new paper.
To: DFH
Diana: Please send him the paper with a note saying that there is more
to come eventually, and I'll put him on the list.
∂17-Feb-83 1653 YOU@UTAH-20 interested in your new paper.
Received: from SU-SCORE by SU-AI with PUP; 17-Feb-83 16:53 PST
Received: from UTAH-20 by SU-SCORE.ARPA with TCP; Thu 17 Feb 83 15:09:44-PST
Date: 17 Feb 1983 1602-MST
From: Jia-huai You <you@UTAH-20>
Subject: interested in your new paper.
To: McCarthy@SU-AI
cc: you@UTAH-20
Just aware that you wrote a paper entitled
"Coloring Maps and Kowalski Doctrine".
I'd appreciate it if you could send me a copy of it.
Please send to the following address:
Jia-huai You
Dept. of Computer Science
3160 Merril Engineering Building
University of Utah
Salt Lake City, Utah 84112
Thank you very much,
-You
-------
∂17-Feb-83 1819 JMC
To: DFH
daily.11
∂18-Feb-83 1443 JMC tentative declination of invitation
To: Pool%MIT-MULTICS@USC-ECL, Solomon%MIT-MULTICS@USC-ECL
I just received your letter, and I cannot accept your invitation to take
part given the present format. My objection is that the press release
lists three general areas of computer history, while the videotaping is
organized around specific M.I.T. projects. In my opinion this will
inevitably lead to a distortion of history. In particular, I do not
especially regard CTSS as the culmination of my own work on time sharing.
Please retransmit this to the other members of your executive advisory
board. I am willing to consider a revised proposal, but the revision
cannot be just oral or just in a letter to me.
∂18-Feb-83 1515 JMC
To: DFH
Please send copies of daily.11 to the following:
.<< bcc: Barton Bernstein, Diana Dutton, Sidney Hook, Edward Teller,
. Lowell Wood, Elliott Bloom, Steve Kline, Edwin Good, Robert McGinn>>
∂18-Feb-83 1700 JMC mailing list
To: MSK@SU-AI
Please put JMC-LISTS@SAIL on your mailing list for the defense research
seminar.
∂18-Feb-83 1706 JMC
To: chappell@SRI-AI
Could you send a copy of
TOWARD A HOMUNCULAR THEORY OF BELIEVING
to John McCarthy, Computer Science Dept., Stanford?
∂19-Feb-83 1557 JMC
To: solomon%MIT-MULTICS@USC-ECL
Your transmission worked; reply shortly.
∂26-Feb-83 1740 JMC
To: csd.restivo@SU-SCORE
There is a lot of documentation. Ask Joe Weening or Carolyn Talcott.
∂27-Feb-83 1502 JMC
To: YOM
Did I point you to minimi[e82,jmc]? If not, look at it.
∂27-Feb-83 2311 JMC Lowell return
To: pjb@S1-A
I infer from Lowell not having logged in recently that he is away.
When is he expected?
∂27-Feb-83 2350 JMC
To: pjb@S1-A
No need. Rod told me he's just come back.
∂28-Feb-83 1448 JMC
To: boyer%UTEXAS-20@USC-ECL
It is rumored that your system will be available in Maclisp.
∂28-Feb-83 2123 JMC
To: JJW
∂28-Feb-83 2015 CL.BOYER at UTEXAS-20
Received: from USC-ECL by SU-AI with NCP/FTP; 28 Feb 83 20:15:18 PST
Received: from UTEXAS-20 by USC-ECL; Mon 28 Feb 83 18:33:39-PST
Date: Monday, 28 February 1983 20:29-CST
From: CL.BOYER at UTEXAS-20
To: John McCarthy <JMC%SU-AI at USC-ECL>
Moore and I are planning to convert our theorem-prover to
run on a Symbolics 3600 this summer. We have two 3600s on
order, and we will probably order more if they work as we
expect.
Although an Interlisp "compatibility package" is available
for the 3600, I am fairly sure we will make the program work
in Zetalisp, which is very close to Maclisp.
∂01-Mar-83 0126 JMC texts
To: AAM
The Grimm's came somehow from IBM and the Wuthering from III. I can't
imagine any objection in either case. I don't think either is copyrighted.
∂02-Mar-83 1638 JMC Arden House conf.
To: DFH
When must I leave New York to arrive at SFO by 6:30?
∂02-Mar-83 1759 JMC
To: DFH
I have decided to take part in Brian Randell's Newcastle meeting.
∂04-Mar-83 0059 JMC
To: CLT
Wed. at 2pm looks good, but see the following:
∂03-Mar-83 2050 CSL.VER.RJT@SU-SIERRA my thesis proposal
Received: from SU-SIERRA by SU-AI with PUP; 03-Mar-83 20:33 PST
Date: Thu 3 Mar 83 17:54:27-PST
From: Richard Treitel <CSL.VER.RJT@SU-SIERRA.ARPA>
Subject: my thesis proposal
To: jmc@SU-AI.ARPA
cc: CSL.VER.RJT@SU-SIERRA.ARPA
Since I gave you that document, the more I have thought about it the less I
have liked it. Compared with LCF, there seems to be very little that is new.
Unless you (or Jussi) can see something of value in it, I feel like dropping
the whole thing and switching to a different project.
- RJT
-------
∂04-Mar-83 1534 JMC
To: vardi@SU-HNV
I don't remember any Bob Doyle, but Jon Doyle is in the CS Department
at CMU.
∂04-Mar-83 1535 JMC
To: vardi@SU-HNV
Bob Moore is at SRI. Combining the two is an operation for whose
consequences I would not take responsibility.
∂07-Mar-83 1350 JMC time-sharing history
To: rich%oz%MIT-MC@USC-ECL
The documentation of time-sharing history is in bad shape. However, once
you acknowledge receipt of this message (so I can be sure the path
to OZ is working), I'll send you two files. The first contains a copy
of a 1959 January 1 memo from me to Philip Morse proposing time-sharing.
It also contains a letter from Christopher Strachey to Donald Knuth
saying that the concept of time-sharing of his 1959 paper was not
the one I proposed. Time-sharing in Strachey's sense, where programs
are debugged to run together, goes back to the SAGE system. The
concept where each many users of a computer can behave as though
he has a computer of his own was invented by me.
The second file is a reminiscence written in the last week stimulated
by Ithiel Pool and Richard Solomon. It is solely based on my present
memory and may misstate what others did. It also is critical of others,
and some of the criticisms may be mistaken. Please do not quote it
or take what it says as anything but providing hypotheses to be checked
and leads to be followed.
As to the connection of this with AI:
1. I was doing both.
2. AI research always seemed likely to be an interactive activity,
and my proposed Advice Taker would have been an interactive program.
3. However, you will note from the memo to Morse that time-sharing
was advocated for its general virtues, and AI wasn't mentioned.
∂07-Mar-83 1429 JMC phone call
To: DFH
Please phone Douglas Hofstadter (see PHON) and tell him (or whoever
takes the message) that my answer is C. My reason is that this is
my normal behavior (and I believe almost everyone's normal behavior)
in the normal circumstances closest to his problem. If I'm in if
you get him, I'll talk to him. If you call his home, call collect.
∂07-Mar-83 1435 JMC
To: DFH
I forgot to say that Hofstadter wants the reason.
∂07-Mar-83 1448 JMC
To: RPG
My part is formal[w83,jmc]. It needs fixing personnel and references.
∂07-Mar-83 1455 JMC New York abstract
To: DFH
What is common sense and what programs need it
Abstract: Common sense thinking combines certain general knowledge about
the world with the facts of a particular situation in order to decide what
to do. Even after thirty years of research in artificial intelligence, we
still don't have a full picture of what knowledge and what reasoning
ability is involved. While many impressive and useful artificial programs
don't have or need much common sense ability, general intelligence
requires it, and so do many important AI programs. Moreover, until
programs have some general common sense, a human will have to use his
common sense to determine whether a program is usable in a given
situation.
We will try to identify what knowledge and what reasoning methods are
required for common sense. We will also try to identify what practical AI
tasks require it. This constitutes part of an agenda for AI research.
∂07-Mar-83 2200 JMC
To: CLT
cats 239-6222
∂08-Mar-83 0044 JMC
To: csl.bkr@SU-SCORE
16th is fine with me.
∂08-Mar-83 1107 JMC 3330 disk drives available
To: REG
If there is any interest in 3330 disk drives cheap, perhaps free,
call John Altstatt 824-5300.
Perhaps Suppes can use them.
∂08-Mar-83 1820 JMC
To: su-bboards@SU-AI
Spring 1983 Industrial Lectureship
ANNOUNCEMENT
INDUSTRIAL LECTURESHIP IN COMPUTER SCIENCE
The Computer Science Department of Stanford University
is pleased to announce the Industrial Lectureship in Computer
Science and Engineering starting in Spring Quarter 1983.
The purpose of the lectureship is to increase interaction between
Computer Science Department faculty and students and computer scientists in local
industry.
Each quarter the Computer Science Department will invite
one outstanding computer scientist from the local industry to give
a course in his specialty. Office space, computer use and salary
appropriate to the teaching of one course will be provided. It is
expected that the balance of the lecturer's salary will be paid by
his permanent employer.
The Spring 1983 course is as follows. Watch for an announcement
of the first meeting. The 1983-84 courses have been determined, and they
will be in the regular Stanford Courses and Degrees.
COMPUTATIONAL APPROACHES TO VISION
Alex Pentland and Stephen Barnard
Artificial Intelligence Center
SRI International
Vision may be studied as a problem in physics, psychology, physiology
or as a computational problem. Recently, research in computational
vision has attempted to take greater advantage of these other
paradigms, and so has gone in directions which are somewhat separate
from ``mainstream'' artificial intelligence research. In particular,
more emphasis has been placed on data concerning biological vision and
on mathematical models of image formation. This seminar will examine
representative examples of these approachs and will explore how, and
to what extent, research in computer vision can take advantage of
these other paradigms. The initial portion of the seminar will
attempt to provide the student with a sophisticated, albiet
necessarily superficial, grasp of human visual psychophysics and
visual neurophysiology.
Qualifications:
Alex Pentland:
* Computer Scientist, in vision research, SRI AI Center.
* Phd Psychology, MIT (1982), in conjunction with MIT AI Lab (Marr's
vision group).
* Assisted in teaching computer vision seminar at MIT during 3 terms.
co-taught course entitled ``Psychophysics And Neurophysiology'' in
MIT psychology dept.
* 10 publications and papers in area of human perception. I have
fairly extensive knowledge of current neurophysiology through
association with the Schiller lab at MIT.
* 21 publications and papers in various types of computer vision
(primarily AI and remote sensing) over the last 10 years,
while at MIT, Arthur D. Little and Environmental Research
Institute of Michigan (ERIM).
Steven Barnard:
* Senior Computer Scientist, vision research, SRI AI Center
* PhD Computer Science, University of Minnesota (1979)
* Many quarters of teaching basic computer science courses
* Several publications in computer vision
∂08-Mar-83 1822 JMC
To: pool%MIT-MULTICS@USC-ECL
timesh[w83,jmc] Reminiscences on the history of time sharing
I remember thinking about time-sharing about the time of
my first contact with computers and being surprised that this wasn't
the goal of IBM and all the other manufacturers and users of computers.
This might have been around 1955.
By time-sharing, I meant an operating system that permits each
user of a computer to behave as though he were in sole control of a
computer, not necessarily identical with the machine on which the
operating system is running. Christopher Strachey may well have been
correct in saying in his letter to Donald Knuth that the term was already
in use for time-sharing among programs written to run together. This idea
had already been used in the SAGE system. I don't know how this kind of
time-sharing was implemented in SAGE. Did each program have to be sure to
return to an input polling program or were there interrupts? Who invented
interrupts anyway? I thought of them, but I don't believe I mentioned the
idea to anyone before I heard of them from other sources.
My first attempts to do something about time-sharing was in the
Fall of 1957 when I came to the M.I.T. Computation Center on a Sloan
Foundation fellowship from Dartmouth College. It was immediately clear to
me that the time-sharing the IBM 704 would require some kind of interrupt
system. I was very shy of proposing hardware modifications, especially as
I didn't understand electronics well enough to read the logic diagrams.
Therefore, I proposed the minimal hardware modification I could think of.
This involved installing a relay so that the 704 could be put into
trapping mode by an external signal. It was also proposed to connect the
sense switches on the ccnsole in parallel with relays that could be
operated by a Flexowriter (a kind of teletype based on an IBM typewriter).
When the machine went into trapping mode,
an interrupt to a fixed location would occur the next time the
machine attempted to execute a jump instruction (then called a
transfer). The interrupt would occur when the Flexowriter had
set up a character in a relay buffer. The interrupt program would
then read the character from the sense switches into a buffer,
test whether the buffer was full, and if not return to the
interrupted program. If the buffer was full, the program would
store the current program on the drum and read in a program to
deal with the buffer.
It was agreed (I think I talked to Dean Arden only.) to install the
equipment, and I believe that permission was obtained from IBM
to modify the computer. The connector to be installed in the
computer was obtained.
However, at this time we heard about
the "real time package" for the IBM 704. This RPQ (request
for price quotation was IBM jargon for a modification to
the computer whose price wasn't guaranteed),
which rented for $2,500 per month had been
developed at the request of Boeing for the purpose of allowing
the 704 to accept information from a wind tunnel. Some element
of ordinary time-sharing would have been involved, but we did not seek
contact with Boeing. Anyway it was agreed that the real time
package, which involved the possibility of interrupting after
any instruction, would be much better than merely putting
the machine in trapping mode. Therefore we undertook to beg
IBM for the real time package. IBM's initial reaction was
favorable, but nevertheless it took a long time to get the
real time package - perhaps a year, perhaps two.
It was then agreed that someone, perhaps Arnold Siegel,
would design the hardware to connect one Flexowriter to the
computer, and later an installation with three would be designed.
Siegel designed and build the equipment, the operating system
was suitably modified (I don't remember by whom), and demonstration
of on-line LISP was held for a meeting of the M.I.T. Industrial
Affiliates. This demonstration, which I planned and carried out,
had the audience in a fourth floor lecture room and me in the
computer room and a rented closed circuit TV system. Steve
Russell, who worked for me, organized the practical details including
a rehearsal. This demonstration was called time-stealing,
and was regarded as a mere prelude to proper time-sharing.
It involved a fixed program in the bottom of memory that collected
characters from the Flexowriter in a buffer while an ordinary batch
job was running. It was only after each job was run that a job
that would deal with the characters typed in would be read in
from the drum. This job would do what it could until more
input was wanted and would then let the operating system go
back to the batch stream. This worked for the demonstration,
because at certain hours, the M.I.T. Computation Center operated
at certain hours a batch stream with a time limit of one minute on any job.
Around the time of this demonstration, Herbert
Teager came to M.I.T. as an assistant professor of Electrical Engineering
and expressed interest in the time-sharing
project. Some of the ideas of time-sharing overlapped some ideas
he had had while on his previous job, but I don't remember what
they were. Philip Morse, the Director of the Computation Center,
asked me if I was agreeable to turning over the time-sharing project
to Teager, since artificial intelligence was my main interest.
I agreed to this, and Teager undertook to design the three Flexowriter
system. I'm not sure it was ever completed.
There was a proposal for support for time-sharing submitted to
NSF and money was obtained. I don't remember whether this preceded
Teager, and I don't remember what part I had in preparing it or
whether he did it after he came.
This should be an important document, because it will contain that
year's conception of and rationale for time-sharing.
Besides that, IBM was persuaded to make substantial modifications
to the IBM 7090 to be installed at the M.I.T. Computation Center.
These included memory protection and relocation and an additional
32,768 words of memory for the time-sharing system. Teager was the
main specifier of these modifications. I remember my surprise when
IBM agreed to his proposals. I had supposed that relocation
and memory protection would greatly slow the addressing of the computer,
but this turned out not to be the case.
Teager's plans for time-sharing were ambitious and, it
seemed to many of us, vague. Therefore, Corbato undertook an
"interim" solution using some of the support that had been obtained
from NSF for time-sharing work. This system was demonstrated some
time in 1962, but it wasn't put into regular operation. That wasn't
really possible until ARPA support for Project MAC permitted
buying a separate IBM 7090.
Around 1960 I began to consult at BBN on artificial intelligence
and explained my ideas about time-sharing to Ed Fredkin and
J. C. R. Licklider. Fredkin, to my surprise, proposed that time-sharing
was feasible on the PDP-1 computer. This was D.E.C.'s first computer,
and BBN had the prototype. Fredkin designed the architecture of
an interrupt system and designed a control system for the drum to
permit it to be used in a very efficient swapping mode. He convinced
Ben Gurley, the chief engineer for D.E.C. to build this equipment.
It was planned to ask NIH for support, because of potential medical
applications of time-sharing computers, but before the proposal could
even be written, Fredkin left BBN. I took technical charge of the
project as a one-day-a-week consultant, and Sheldon Boilen was hired
to do the programming. I redesigned the memory extension system
proposed by D.E.C. and persuaded them to build the modified system
instead of the two systems they were offering, but fortunately
hadn't built. I also supervised Boilen.
Shortly after this project was undertaken, D.E.C. decided
to give a PDP-1 to the M.I.T. Electrical Engineering Department.
Under the leadership of Jack Dennis, this computer was installed
in the same room as the TX-0 experimental transistorized computer
that had been retired from Lincoln Laboratory when TX-2 was built.
Dennis and his students undertook to make a time-sharing system
for it. The equipment was similar, but they were given less memory
than the BBN project had. There wasn't much collaboration.
My recollection is that the BBN project was finished first
in the summer of 1962, but perhaps Corbato remembers earlier
demonstrations of CTSS. I left for Stanford in the Fall of 1962,
and I hadn't seen CTSS, and I believe I hadn't seen Dennis's system
operate either. BBN didn't operate the first system and didn't
even fix the bugs. They had few computer users and were content
to continue the system whereby users signed up for the whole
computer. They did undertake a much larger follow-on project
involving a time-shared PDP-1 that was installed in Massachusetts
General Hospital, where it was not a success. The computer was
inadequate, there were hardware and software bugs, and
there was a lack of application programs, but mainly the project
was premature.
At the same time that CTSS, the BBN system, and the EE Department
systems were being developed, M.I.T. had started to plan for
a next generation computer system. The management of M.I.T. evidently
started this as an ordinary university planning exercise and
appointed a high level committee consisting of Philip Morse, Albert
Hill and Robert Fano to supervise the effort. However, the actual
computer scientists were persuaded that a revolution in the
way computers were used - to time-sharing - was called for.
The lower level committee was chaired by Teager, but after his
ideas clashed with those of everyone else, the committee was
reconstituted with me as chairman. The disagreement centered around
how ambitious to be and whether to go for an interim solution.
Teager wanted to be very ambitious, but the rest of us thought
his ideas were vague, and he wanted M.I.T. to acquire an IBM 7030
(Stretch) computer as an interim solution. As it turned out, acquiring a
Stretch would have been a good idea.
Our second report to M.I.T. proposed that M.I.T. send out
a request for proposals to computer manufacturers. On the basis
of the responses, we would then ask the Government for the money.
The RFP was written, but M.I.T. stalled perhaps for two reasons.
The first reason was that our initial cost estimates were very
large for reasons of conservatism. Secondly, IBM asked M.I.T. to
wait saying that they would make a proposal to meet M.I.T.'s needs
at little or no cost. Unfortunately, the 360 design took longer
than IBM management expected, and along about that time, relations
between M.I.T. and IBM became very strained because of the patent
lawsuit about the invention of magnetic core memory.
As part of the stall, President Stratton proposed a new
study with a more thorough market survey to establish the demand
for time-sharing among M.I.T. computer users. I regarded this
as analogous to trying to establish the need for steam shovels
by market surveys among ditch diggers and didn't want to do it.
About this time George Forsythe invited me to come back to
Stanford with the intention of building a Computer Science
Department, and I was happy to return to California.
In all this, there wasn't much publication. I wrote a
memo to Morse dated January 1, 1959 proposing that we time-share
our expected "transistorized IBM 709". It has been suggested
that the date was in error and should have been 1960. I don't
remember now, but I believe that if the memo had been written
at the end of 1959, it would have referred to the 7090, because
that name was by then current. In that memo I said the idea of
time-sharing wasn't especially new. I don't know why I said that,
except that I didn't want to bother to distinguish it from what
was done in the SAGE system with which I wasn't very familiar.
Most of my argumentation for time-sharing was oral, and when
I complained about Fano and Corbato crediting Strachey with time-sharing
in their 1966 Scientific American article, Corbato was surprised to
find my 1959 memo in the files. Their correction in Scientific
American was incorrect, because they supposed that Strachey and I
had developed the idea independently, whereas giving each user
continuous access to the machine wasn't Strachey's idea at all.
In fact, he didn't even like the idea when he heard about it.
Teager and I prepared a joint abstract for an ACM meeting shortly
after he arrived, and I gave a lecture in an M.I.T series called
Management and the Computer of the Future. In this lecture I referred to
Strachey's paper "Time-sharing of large fast computers" given at the 1959
IFIP Congress in Paris. I had read the paper carelessly, and supposed he
meant the same thing as I did. As he subsequently pointed out, he meant
something quite different that did not involve a large number of users,
each behaving as though he had a machine to himself. As I recall, he
mainly referred to fixed programs, some of which were compute bound and
some input-output bound. He did mention debugging as one of the
time-shared activities, but I believe his concept involved one person
debugging while the other jobs were of the conventional sort.
My 1959 memo advertised that users generally would get the
advantage of on-line debugging. However, it said nothing about how
many terminals would be required and where they would be located.
I believe I imagined them to be numerous and in the users' offices, but
I cannot be sure. Referring to an "exchange" suggests that I had
in mind many terminals. I cannot now imagine what the effect was on
the reader of my failure to be explicit about this point. I'm afraid
I was trying to minimize the difficulty of the project.
The major technical error of my 1959 ideas was an underestimation
of the computer capacity required for time-sharing. I still don't
understand where all the computer time goes in time-sharing installations,
and neither does anyone else.
Besides M.I.T.'s NSF proposal, there ought to be some letters
to IBM and perhaps some IBM internal documents about the proposal,
since they put more than a million dollars worth of equipment into
it. Gordon Bell discusses D.E.C.'s taking up time-sharing in Bell
and Newell book, but I don't recall that they discuss Ben Gurley's
role. Fredkin and perhaps Alan Kotok would know about that.
After I came to Stanford, I organized another PDP-1 time-sharing
project. This was the first time-sharing system based on display
terminals. It was used until 1969 or 1970 for Suppes's work on
computer aided instruction.
∂08-Mar-83 1824 JMC
To: pool%MIT-MULTICS@USC-ECL
January l, l959
To: Professor P.M. Morse
From: John McCarthy
Subject: A Time Sharing Operator Program for our Projected IBM 709
l. INTRODUCTION
This memorandum is based on the assumption that MIT will be
given a transistorized IBM 709 about July l960. I want to propose
an operating system for it that will substantially reduce the time
required to get a problem solved on the machine. Any guess as to
how much of a reduction would be achieved is just a guess, but a
factor of five seems conservative. A smaller factor of improvement
in the amount of machine time used would also be achieved.
The proposal requires a complete revision in the way the machine
is used, will require a long period of preparation, the development
of some new equipment, and a great deal of cooperation and even
collaboration from IBM. Therefore, if the proposal is to be con-
sidered seriously, it should be considered immediately. I think
the proposal points to the way all computers will be operated in
the future, and we have a chance to pioneer a big step forward in
the way computers are used. The ideas expressed in the following
sections are not especially new, but they have formerly been con-
sidered impractical with the computers previously available. They
are not easy for computer designers to develop independently since
they involve programming system design much more than machine design.
2. A QUICK SERVICE COMPUTER
Computers were originally developed with the idea that
programs would be written to solve general classes of problems and
that after an initial period most of the computer time would be
spent in running these standard programs with new sets of data.
This view completely underestimated the variety of uses to which
computers would be put. The actual situation is much closer to the
opposite extreme, wherein each user of the machine has to write
his own program and that once this program is debugged, one run
solves the problem. This means that the time required to solve
the problem consists mainly of time required to debug the program.
!This time is substantially reduced by the use of better programming
languages such as Fortran, LISP (the language the Artificial
Intelligence Group is developing for symbolic manipulations) and
COMIT (Yngve's language). However, a further large reduction can
be achieved by reducing the response time of the computation center.
The response time of the MIT Computation Center to a performance
request presently varies from 3 hours to 36 hours depending on the
state of the machine, the efficiency of the operator, and the
backlog of work. We propose by time sharing, to reduce this
response time to the order of 1 second for certain purposes. Let
us first consider how the proposed system looks to the user before
we consider how it is to be achieved.
Suppose the average program to be debugged consists of 500
instructions plus standard subroutines and that the time required
under the present system for an average debugging run is 3 minutes.
This is time enough to execute 7,000,000 704 instructions or to
execute each instruction in the program l4,000 times.
Most of the errors in programs could be found by single-
stepping or multiple-stepping the program as used to be done.
If the program is debugged in this way, the program will usually
execute each instruction not more than 10 times, 1/1400 as many
executions as at present. Of course, because of slow human re-
actions the old system was even more wasteful of computer time
than the present one. Where, however, does all the computer time
go?
At present most of the computer time is spent in conversion
(SAP-binary, decimal-binary, binary-decimal, binary-octal) and in
writing tape and reading tape and cards.
Why is so much time spent in conversion and input output.
1. Every trial run requires a fresh set of conversions.
2. Because of the slow response time of the system it is
necessary to take large dumps for fear of not being able to find
the error. The large dumps are mainly unread, but nevertheless,
they are necessary. To see why this is so, consider the behavior
!of a programmer reading his dump. He looks at where the program stopped.
Then he looks at the registers containing the partial results so far
computed. This suggests looking at a certain point in the program. The
programmer may find his mistake after looking at not more than 20
registers out of say 1000 dumped, but to have predicted which 20 would
have been impossible in advance and to have reduced the 1000
substantially would have required cleverness as subject to error as his
program. The programmer could have taken a run to get the first register
looked at, then another run for the second, etc., but this would have
required 60 hours at least of elapsed time to find the bug according to
our assumptions and a large amount of computer time for repeated loading
and re-runnings. The response time of the sheet paper containing the dump
for any register is only a few seconds which is OK except that one dump
does not usually contain information enough to get the entire program
correct.
Suppose that the programmer has a keyboard at the computer
and is equipped with a substantial improvement on the TXO interro-
gation and intervention program (UT3). (The improvements are in
the direction of expressing input and output in a good programming
language.) Then he can try his program, interrogate individual pieces
of data or program to find an error, make a change in the source
language and try again.
If he can write program in source language directly into the
computer and have it checked as he writes it, he can save additional
time. The ability to check out a program immediately after writing
it saves still more time by using the fresh memory of the programmer.
I think a factor of 5 can be gained in the speed of getting pro-
grams written and working over present practice if the above-
mentioned facilities are provided. There is another way of using
these facilities which was discussed by S. Ulam a couple of years
ago. This is to use the computer for trial and error procedures
where the error correction is performed by a human adjusting
parameter.
! The only way quick response can be provided at a bearable
cost is by time-sharing. That is, the computer must attend to
other customers while one customer is reacting to some output.
3. THE PROBLEM OF A TIME-SHARING OPERATOR SYSTEM
I have not seen any comprehensive written treatment of the
time-sharing problem and have not discussed the problem with
anyone who had a complete idea of the problem. This treatment is
certainly incomplete and is somewhat off the cuff. The
equipment required for time-sharing is the following:
a. Interrogation and display devices (flexowriters are possible
but there may be better and cheaper).
b. An interrupt feature on the computer -- we'll have it.
c. An exchange to mediate between the computer and the
external devices. This is the most substantial engineering problem,
but IBM may have solved it.
In general the equipment required for time-sharing is well
understood, is being developed for various advanced computers, e.g.,
Stretch TX2, Metrovich 1010, Edsac 3. I would not be surprised if
almost all of it is available with the transistorized 709. However,
the time-sharing has been worked out mainly in connection with
real-time devices. The programs sharing the computer during any
run are assumed to occupy prescribed areas of storage, to be
debugged already, and to have been written together as a system.
We shall have to deal with a continuously changing population of
programs, most of which are erroneous.
The major problems connected with time-sharing during pro-
gram development seem to be as follows:
l. Allocating memory automatically between the programs.
This requires that programs be assembled in a relocatable form
and have a preface that enables the operator program to organize the
program, its data, and its use of common subroutines.
2. Recovery from stops and loops. The best solutions to these
problems require
a. Changing the stop instructions to trap instructions.
This is a minor modification to the machine. (At least it will
be minor for the 704.)
b. Providing a real time alarm clock as an external device.
! 3. Preventing a bad program from destroying other programs.
This could be solved fairly readily with a memory range trap which
might not be a feasible modification. Without it, there are pro-
gramming solutions which are less satisfactory but should be good
enough. These include:
l. Translations can be written so that the programs
they produce cannot get outside their assigned storage areas. A
very minor modification would do this to Fortran.
2. Checksums can be used for machine language programs.
3. Programming techniques can be encouraged which make destruction
of other programs unlikely.
4. There is an excessive tendency to worry about this
point. The risk can be brought down to the present risk of having
a program ruined by operator or machine error.
4. SUMMARY
l. We may be able to make a major advance in the art of
using a computer by adopting a time-sharing operator program for
our hoped-for 709.
2. Such a system will require a lot of advance preparation
starting right away.
3. Experiments with using the flexo connection to the
real-time package on the 704 will help but we cannot wait for the
results if we want a time-sharing operator program in July l960.
4. The cooperation of IBM is very important but it should
be to their advantage to develop this new way of using a computer.
5. I think other people at MIT than the Computation Center
staff can be interested in the systems and other engineering
problems involved.
!
OXFORD UNIVERSITY COMPUTING LABORATORY 45 Banbury Road
PROGRAMMING RESEARCH GROUP Oxford OX2 6PE
1st May 1974
Professor D. E. Knuth
Stanford University
Computer Science Department
Stanford, California 94305
U.S.A.
Dear Don:
The paper I wrote called `Time Sharing in Large Fast
Computers' was read at the first (pre IFIP) conference at
Paris in l960. It was mainly about multi--programming (to
avoid waiting for peripherals) although it did envisage this
going on at the same time as a programmer was debugging his program
at a console. I did not envisage the sort of console system which
is now so confusingly called time sharing. I still think my use
of the term is the more natural.
I am afraid I am so rushed at the moment, being virtually
alone in the PRG and having just moved house, that I have no
time to look up any old notes I may have. I hope to be able to
do so while settling in and if I find anything of interest I
will let you know.
Don't place too much reliance on Halsbury's accuracy. He
tends to rely on memory and get the details wrong. But he was
certainly right to say that in l960 `time sharing' as a phrase
was much in the air. It was, however, generally used in my sense
rather than in John McCarthy's sense of a CTSS-like object.
Best wishes,
Yours sincerely,
C. Strachey
Professor of Computation
University of Oxford
∂09-Mar-83 1027 JMC
To: lantz@SU-HNV
You are undoubtedly right, and I forgot, for which I apologize.
However, unless you had an additional candidate, the issue is moot,
because there were exactly four candidates for four slots. The
other three were John Greenstadt from IBM who will lecture on the
finite element method, Bob Moore from SRI and Stan Rosenschein from
SRI. Had I remembered, I wouldn't have had to do all the work.
∂10-Mar-83 0222 JMC lunch on Tuesday
To: admin.mrc@SU-SCORE, ME@SU-AI
Why?
∂09-Mar-83 2118 CSD.GOLUB@SU-SCORE lunch on Tuesday
Received: from SU-SCORE by SU-AI with PUP; 09-Mar-83 21:18 PST
Date: Mon 7 Mar 83 22:10:19-PST
From: Gene Golub <CSD.GOLUB@SU-SCORE.ARPA>
Subject: lunch on Tuesday
To: faculty@SU-SCORE.ARPA
I've asked Bob White who is head of the EE department to come to
lunch on Tuesday to talk to us about being an engineering department.
Pete Veinott of OR will probably come to. If you have any thoughts
about our possible move to Engineering be sure to come.
GENE
-------
∂11-Mar-83 1021 JMC
To: DFH
Please make fac. club reservation for two for 1:15 today.
∂11-Mar-83 1136 JMC
To: DFH
No need to call Faculty Club; I finally got them.
∂11-Mar-83 1150 JMC
To: RPG
I suppose short biographies will suffice.
∂11-Mar-83 1529 JMC
To: reid@SU-SHASTA
CC: csd.rita@SU-SCORE
Wednesday at 9:30 then.
∂11-Mar-83 2127 JMC
To: csd.tucci@SU-SCORE
Yes to both questions.
∂13-Mar-83 1733 JMC LOTS
To: su-bboards@SU-AI
1. Most likely, the University won't spend a lot of money soon on relieving
the LOTS situation in the near future. In fact, this situation is
perceived by them (I believe correctly) to be much better than it was.
2. While it is often convenient for a course to make the last large
assignment due at the end of the quarter, present LOTS capacity makes
it the lesser evil to co-ordinate among courses. This has happened at
times in the past, and perhaps the situation this quarter would make it
happen again. The best way to do it is for LOTS management to tell
the four or five largest courses what their latest deadlines should be
and rotate this from quarter to quarter. They could cut off students
in the courses with early deadlines at appropriate times.
3. It is said that the 2080 will have more than 10 times the power
of the 2060. If this applies to time-shared operation, then Stanford
is well advised to wait for it unless there is a more cost-effective
other solution.
4. The cost-effectiveness of other solutions needs to be demonstrated
in personnel as well as hardware costs. CSD was not very successful
in keeping large numbers of Altos running and supplied with software.
The personnel costs of LOTS are considerably lower than its hardware
costs. The personnel costs of C.I.T. are very much higher.
5. It is probably not cost-effective to go with a brand new architecture
no matter what the promised software is. Part of LOTS's success comes
from the wide variety of software available, and this includes much
software unused and almost unknown in the Computer Science Department.
Engineering faculty in particular have their favorite packages.
6. Most likely a micro based system will require a network and a file
server to be cost-effective. Keeping disk supplied and working and
copying software onto them is likely to lose in manpower requirements.
7. Proponents of micros should advocate and design an experiment with
(say) 25 terminals.
8. There is much that can be done to improve the efficiency of the
2060 in LOTS type operation. However, this work is difficult for
computer scientists, especially computer science students, to undertake.
First, the culture is impressed by features and not by efficiency.
Second, improving the efficiency of a system requires understanding
its operation, and it's much more fun to start afresh.
One readily believes that the problems one knows about are all the
problems there are, and this leads one to believe that the new system will
surely be more cost-effective. At Xerox PARC they said the Altos would
cost $500 in five years five years ago.
9. There are a number of areas in which the efficiency of LOTS and
SCORE, etc. could be improved. These include
a. Improving EMAX. Its improvement should compete with the
project to replace it. The latter project is likely to succumb to
creeping featurism anyway. The designers will brag about its wonderful
features and forget to measure its efficiency.
b. Display service. In this respect TOPS-20 is far inferior
to WAITS, and WAITS is none too good.
c. System inefficiencies detectable by reading the code. For
example, it is said that the time to open a file is proportional to
the number of files already open. With such anomalies, there is no
guarantee that the 2080 will be a lot more powerful than the 2060
even though benchmark problems will show the advertised improvement.
10. There are theses projects to be undertaken in the area of operating
system efficiency. There are also more modest projects.
∂15-Mar-83 1105 JMC
To: pourne%MIT-MC@USC-ECL
"for you, but I do need (1) a hard copy of that essay you wrote
on the subject, and (2) a complete vita and bibliography, at"
What essay on what subject? I'll send the vita. I'll be in the
New York Thursday through Sunday noon, but if there are phone
numbers where you can be reached Sunday afternoon, perhaps we can
get together, since I will be in SF for the pocket opera Sunday
evening.
∂15-Mar-83 1328 JMC
To: JJW
Sorry, I'll send it anon.
∂15-Mar-83 2102 JMC problem with line
To: ME
I am getting garbage back from the computer along with the
correct information. Perhaps the modem at one end or the other
is having trouble. Transmission to the computer is correct and
fragments of the output come back. The terminal works fine when
used over the telephone.
∂15-Mar-83 2107 JMC calendar
To: CLT
Next week is free except for Ohlander on Wednesday and our dinner
in Berkeley on Friday. The Datamedia is working normally over the
telephone; only the private line is ill.
∂15-Mar-83 2320 JMC
To: ME
It seems to have fixed itself. Maybe it was a loose connection.
∂16-Mar-83 0101 JMC tty20
To: ME
It has gone bad again. I tried the loop-back feature here,
and everything worked fine. Also using the Datamedia over
the phone works fine. Therefore, its either the Stanford
modem or more likely the phone line. I'd like to be in
a position to complain to the phone company as soon as possible.
Or rather to have you complain. I apologize for
possible garbling of this message, but I can't really read what
I have written.
∂16-Mar-83 0140 JMC Chicago paper
To: DFH
It is chicag.xgp[w83,jmc]. Please
xs chicag.xgp[w83,jmc]
several times until you get a full set of pages without too many
blotches. The send it to Chicago special delivery.
∂16-Mar-83 0203 JMC L-5 article
To: HPM
It's an excellent article. However, does it work with the existing
shuttle or does it depend on a new rocket with the shuttle technology?
Since the shuttle is volume limited for many missions, allowing a lower
terminal velocity might not help much. Have you thought about what
revisions of the shuttle would help.
∂16-Mar-83 0203 JMC
To: DFH
Please send a copy of my long biography to Jerry Pournelle.
∂16-Mar-83 0954 JMC
To: csd.golub@SU-SCORE, DEK@SU-AI, csd.ullman@SU-SCORE
It would be nice to do something in honor of Art Samuel. I don't
have a definite idea, but I'll think. It's "Samuel" by th way.
∂16-Mar-83 1557 JMC
To: DFH
Diana: Please make out the grade change.
∂16-Mar-83 1512 JJW 206 grade
Mai Hoang's project (an algebraic simplifier) somehow got into the pile of
graded assignments without being graded. I looked at it, and it deserves
about a B. Since her exam results were among the poorest in the class (42 on
the midterm and 10 on the final), I think her final grade should be a B- which
would agree with others doing approximately the same level of work. If this
is OK with you, can you have Diana fill out the grade card?
∂16-Mar-83 1602 JMC
To: DFH
The private telephone line between my home terminal and SAIL
has become noisy in the last day. Please complain to the phone
company by calling 611. The circuit number is 77KD1618, and one
end is at my home and the other is in room 020A or 020B in Margaret
Jacks Hall. Martin Frost can answer questions.
∂16-Mar-83 2148 JMC foot-in-mouth and fashionable scapegoating
To: reid@SU-SHASTA
CC: admissions@SU-SHASTA
Unless you are prepared to accuse Ross or someone else of preferring
less qualified applicants and back it up with examples of people
who should be but weren't offered admission, you should withdraw
your accusation.
∂17-Mar-83 1015 JMC →15553 (21-Mar-83)
To: "#___JMC.PLN[2,2]"
In New York till Sunday, March 20.
∂21-Mar-83 0001 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:
In New York till Sunday, March 20.
∂21-Mar-83 0046 JMC
To: EJS
timesh[w83,jmc]. Not for passing around copies.
∂21-Mar-83 1354 JMC Art Samuel day
To: csd.golub@SU-SCORE
Getting John Cocke on the phone by calling his number is a rare event,
but I won the first time. He thought that many people at IBM would
be interested in attending an Art Samuel appreciation day and will
get back to me with a list. My current idea is that there would be
two or so appreciatory speeches, a gift or so, and a short scientific
program in areas of his interest, including game playing, speech recognition,
and something characteristic of his IBM activity. When I mentioned
the idea, John asked whether we wanted help from IBM, but told him that
mainly we were interested in personal participation. If IBM wants to
help organize it, that will be great. My opinion is that IBM participation
will constitute bread cast upon the water and will eventually be
returned in better relations.
∂21-Mar-83 1539 JMC meeting with Knuth
To: nilsson%SRI-AI@USC-ECL
He is enthusiastic, and you should call him to arrange a meeting. He
is at present on a schedule of waking up at noon.
∂21-Mar-83 1633 JMC
To: SU-BBOARDS@SU-AI
Anyone who might take the AI Qual this Spring should inform me.
∂21-Mar-83 1650 JMC
To: nilsson%SRI-AI@USC-ECL
Also Zohar and Ed.
∂21-Mar-83 2054 JMC
To: nilsson%SRI-AI@USC-ECL
Brown, Irving Director, Department of International Affairs
AFL-CIO
815 Sixteenth St., N.W.
Washington, D.C. 20006 (202) 637-5050
20, rue de la Paix 261-7471
75002 Paris, France 261-6874
∂22-Mar-83 0950 JMC
To: DFH
Please pub and print malik.re1[let,jmc].
∂22-Mar-83 1420 JMC Slagle paper
To: bobrow%PARC-MAXC@USC-ECL
I have serious doubts about whether it is a good idea to publish a paper
about military applications in an international journal. I have nothing
against applying AI to the defense of the U.S., but I think they should
publish in a U.S. journal, assuming the article should be published as
a paper.
∂22-Mar-83 1530 JMC
To: HHB
Sure. Come up.
∂22-Mar-83 1531 JMC
To: bobrow%PARC-MAXC@USC-ECL
I am willing to review the Levesque paper.
∂22-Mar-83 1640 JMC
To: HHB
I spoke to Betty Scott who confirms that nothing will happen immediately.
∂22-Mar-83 1850 JMC
To: bobrow%PARC-MAXC@USC-ECL
Also I'll be at Karlsruhe (and Washington) and approve Loveland.
∂23-Mar-83 1554 JMC
To: HHB
It is done.
∂24-Mar-83 1547 JMC
To: csd.tajnai@SU-SCORE
No, I don't have time nor can I re-establish the state of mind. Also
I must say that I regard the lecture itself as a bombastic disaster.
∂24-Mar-83 1757 JMC
To: RSF
I don't know whether my defense of you ever got through the mailer.
∂26-Mar-83 1556 JMC
To: TOB
The Industrial Professorships for the year 1983-84 are already determined.
This was necessary so that we could get them into the catalog. Please
tell Mundy to apply for the academic year 84-85 in a letter to Golub
with a copy to me by January 1984. The letter should include a vita
and a catalog description of the proposed course suitable for inclusion
in the catalog without further modification.
∂26-Mar-83 1558 JMC
To: TOB
Addendum to previous: There is nothing in principle to prevent our
having more than one Industrial Professor at a time, but we don't
know how large the demand is, and we do pay them. Therefore, I think
we should leave it as is for now.
∂26-Mar-83 1717 JMC
To: JJW
congratulations
∂27-Mar-83 1458 JMC Seminar on short notice
To: "@DIS.DIS[1,CLT]%SU-AI"@USC-ECL
Edward Zalta will talk on "Possible worlds and stories" in
MJH 352 at 1pm, Monday, March 28. This is of possible interest
for AI and computational linguistics.
∂27-Mar-83 1713 JMC
To: DFH
The letter to Kirkland is a draft going to Chudnovsky for suggestions.
∂28-Mar-83 1820 JMC
To: csd.scott@SU-SCORE
Diana Hall's work is entirely satisfactory, and she has mastered
all necessary computer skills and some optional ones.
∂28-Mar-83 2355 JMC
To: RPG
Have you probed Lowell?
∂28-Mar-83 2357 JMC
To: llw@S1-A
I'll talk to Joe about it, and if this is what he wants to do, I'll help
him. I think a good thesis in this area would be very valuable.
However, there are those who like theory and also like system
work but don't like to or can't combine them. It's worth trying.
∂29-Mar-83 1411 JMC Congratulations!
To: JJW
∂28-Mar-83 2217 Lowell Wood <LLW at S1-A> Congratulations!
Received: from S1-A by SU-AI with NCP/FTP; 28 Mar 83 22:17:11 PST; for: jmc
Date: 28 Mar 1983 2213-PST
From: Lowell Wood <LLW at S1-A>
Subject: Congratulations!
To: jmc at SU-AI
CC: LLW at S1-A
Dear John:
Congratulations! You are now the proud owner of a brand-new Hertz Fellow,
Joe Weening. Although he's not a genuinely low-mileage graduate student, I
think that you'll find that, with regular maintenance and careful use,
you'll get many years of excellent service from him.
More seriously, I couldn't help but note in the course of my interview
with him ten days ago that he spoke with much more animation of his
systems-type activities than he did of his machine reasoning work. (As is
rather self-evident, I also came to share your high opinion of him as an
individual, as expressed in your much-appreciated, better-late-than-
too-late letter-of-reference).
He spoke with particular interest of delving into the question of why, if
it takes an average of, say, 10,000 instructions executed per second to
provide truly high quality terminal service to each user of a computer
system and you have a 10 MIPS system, you can't provide such service to
1000 users (instead of the 100 or so that actually seems to be the case).
Both you and I have expressed non-trivial interest in this question in the
past. I'm becoming quite immediately intrigued by it, and wonder if you
might still be sufficiently interested in it to devote an uncommonly
capable student to at least exploratory work on the question.
I'm prepared to throw in summer support, lots of semi-continuous peer
review by similarly interested S-1 staff, and all the access to a real 10
MIPS machine which could possibly be of interest, if you would care to
commit Joe's time (and perhaps a bit of your own, in supervision) for the
coming summer quarter, with the understanding that Joe would document his
findings in either a "Here's why not" report or an ad hoc dissertation
proposal in early Fall.
Please let me know how you see this possibility.
Lowell
I'll talk to Joe about it, and if this is what he wants to do, I'll help
him. I think a good thesis in this area would be very valuable.
However, there are those who like theory and also like system
work but don't like to or can't combine them. It's worth trying.
∂30-Mar-83 0015 JMC
To: TOB
Is your part of the proposal in?
∂30-Mar-83 1543 JMC
To: JEF@SU-AI, GHG@SU-AI
I have been math sciences adviser for 3 years and want to be replaced in Sept.
∂31-Mar-83 0225 JMC
jmc - CIT's main need is to have its budget drastically reduced.
Central planning of computing at Stanford would be harmful,
because anyone competent to exercise the authority could
spend his time more usefully in other ways. Someone
incompetent, like the incumbent, would involve
the whole campus, including CSD, in battles to prevent
the authority from being abused.
Central planning should not be done for computing at Stanford.
The potential benefits are small and are far outweighed by the
risk that it will be done badly or (more likely) that large amounts
of faculty time will have to be spent in politicking to prevent
it from doing harm.
Unless the situation has changed, CIT (it used to be Jon Sandelin)
had to approve the acquisition of the DEC-20 for Sierra. However,
the EE Department was entirely correct in ignoring CIT even if they
knew the nominal rule, since EE will prove competent to manage a
DEC-20, and CIT is not.
Ask David Cheriton points out, CIT is a bloated unnecessary empire
and should have as little authority as possible.
Ed Shaw moved to his position from a purely bureaucratic
one in the Provost's office. He had no technical knowledge
in computing or any other branch of information technology.
He seems to have the usual tendency to expand his organization
and its power to the maximum extent.
Siegmann's example that a central authority should require that
all computers be connectable to a specific network is precisely
the kind of thing that is more trouble than it is worth. If a
network has been set up, departments will connect to it if they
find it in their interest. The probability that a central authority
is needed to avoid rival networks is low enough to take the chance.
As I understand it, anyone with serious number crunching does it
at SLAC or outside Stanford.
The key cost of any computer installation is personnel. Any empire
This can be kept down in one of three ways:
1. The Administration has sufficient technical knowledge
to refuse unnecessary personnel increases and to force personnel
reductions when possible. This has never happened at Stanford
or at any other university or corporation or government agency
that I know about.
2. Services are bought from profit making outside organizations
that compete in price. Many will be unable to prevent personnel growth
even when the consequence is bankruptcy, but a few may survive.
3. The computing unit is small enough so that its personnel
are hired by the boss of the activity that benefits from the computing.
This person should face the choice between computing personnel and
(say) research personnel. At present this gives the lowest costs
most of the time.
The ideal isn't that new users should tell CIT what they
want. They will simply have to learn by experience whether CIT
services, outside services or the micros sold commercially is
best for them.
∂31-Mar-83 0227 JMC
To: su-bboards@SU-AI
Central planning of computing at Stanford would be harmful,
because anyone competent to exercise the authority could
spend his time more usefully in other ways. Someone
incompetent, like the incumbent, would involve
the whole campus, including CSD, in battles to prevent
the authority from being abused.
Central planning should not be done for computing at Stanford.
The potential benefits are small and are far outweighed by the
risk that it will be done badly or (more likely) that large amounts
of faculty time will have to be spent in politicking to prevent
it from doing harm.
Unless the situation has changed, CIT (it used to be Jon Sandelin)
had to approve the acquisition of the DEC-20 for Sierra. However,
the EE Department was entirely correct in ignoring CIT even if they
knew the nominal rule, since EE will prove competent to manage a
DEC-20, and CIT is not.
Ask David Cheriton points out, CIT is a bloated unnecessary empire
and should have as little authority as possible.
Ed Shaw moved to his position from a purely bureaucratic
one in the Provost's office. He had no technical knowledge
in computing or any other branch of information technology.
He seems to have the usual tendency to expand his organization
and its power to the maximum extent.
Siegmann's example that a central authority should require that
all computers be connectable to a specific network is precisely
the kind of thing that is more trouble than it is worth. If a
network has been set up, departments will connect to it if they
find it in their interest. The probability that a central authority
is needed to avoid rival networks is low enough to take the chance.
As I understand it, anyone with serious number crunching does it
at SLAC or outside Stanford.
The key cost of any computer installation is personnel. Any empire
This can be kept down in one of three ways:
1. The Administration has sufficient technical knowledge
to refuse unnecessary personnel increases and to force personnel
reductions when possible. This has never happened at Stanford
or at any other university or corporation or government agency
that I know about.
2. Services are bought from profit making outside organizations
that compete in price. Many will be unable to prevent personnel growth
even when the consequence is bankruptcy, but a few may survive.
3. The computing unit is small enough so that its personnel
are hired by the boss of the activity that benefits from the computing.
This person should face the choice between computing personnel and
(say) research personnel. At present this gives the lowest costs
most of the time.
The ideal isn't that new users should tell CIT what they
want. They will simply have to learn by experience whether CIT
services, outside services or the micros sold commercially is
best for them.
∂31-Mar-83 0228 JMC
To: su-bboards@SU-AI
jmc - CIT's main need is to have its budget drastically reduced.
[The above should be first line of message beginning "Central planning ..."]
∂31-Mar-83 1027 JMC another task
To: RPG
It occurs to me that I have been looking for someone to do a study of
factors affecting the efficiency of time-sharing systems. Joe might
want to work in that area. Therefore, I want to include a possible
task just in the off chance that it becomes big enough to merit special
funding. If there is already such a task in the umbrella put in by
someone, nothing more need be done. Otherwise, we need a task like:
To study the effects of the structure of an operating system on
the number of Lisp and other users that can be supported in various
parts of the program development cycle and in the interactions typical
of AI research.
(This should be in the Common Lisp section, and it needs to be broad
rather than clear, since we'll have to write a task document if we
ever decide to pursue it on a scale requiring additional support).
∂31-Mar-83 1208 JMC
To: shortliffe@SUMEX-AIM
Thanks. My MYCIN session suffices for now.
∂31-Mar-83 1726 JMC
To: elyse@SU-SCORE
Please co-ordinate meeting times for me with Diana Hall, DFH@SAIL. She
has my schedule. Also I'm JMC@SAIL for ordinary messages and
JMC-LISTS@SAIL for inclusion on mailing lists, e.g. on the list for
memos to all faculty or senior faculty.
∂31-Mar-83 1730 JMC Central computer planning
To: su-bboards@SU-AI
CIT's main need is a reduction of budget and personnel.
Central planning of computing at Stanford would be harmful,
because anyone competent to exercise the authority could
spend his time more usefully in other ways. The insufficiently competent
incompetent, like the incumbent, would involve
the whole campus, including CSD, in battles to prevent
the authority from being abused.
Central planning should not be done for computing at Stanford.
The potential benefits are small and are far outweighed by the
risk that it will be done badly or (more likely) that large amounts
of faculty time will have to be spent in politicking to prevent
it from doing harm.
Unless the situation has changed, CIT (it used to be Jon Sandelin)
had to approve the acquisition of the DEC-20 for Sierra. However,
the EE Department was entirely correct in ignoring CIT even if they
knew the nominal rule, since EE will prove competent to manage a
DEC-20, and CIT is not.
Ask David Cheriton points out, CIT is a bloated unnecessary empire
and should have as little authority as possible.
Ed Shaw moved to his position from a purely bureaucratic
one in the Provost's office. He had no technical knowledge
in computing or any other branch of information technology.
He seems to have the usual tendency to expand his organization
and its power to the maximum extent.
Siegmann's example that a central authority should require that
all computers be connectable to a specific network is precisely
the kind of thing that is more trouble than it is worth. If a
network has been set up, departments will connect to it if they
find it in their interest. The probability that a central authority
is needed to avoid rival networks is low enough to take the chance.
As I understand it, anyone with serious number crunching does it
at SLAC or outside Stanford.
The key cost of any computer installation is personnel. Any empire
This can be kept down in one of three ways:
1. The Administration has sufficient technical knowledge
to refuse unnecessary personnel increases and to force personnel
reductions when possible. This has never happened at Stanford
or at any other university or corporation or government agency
that I know about.
2. Services are bought from profit making outside organizations
that compete in price. Many will be unable to prevent personnel growth
even when the consequence is bankruptcy, but a few may survive.
3. The computing unit is small enough so that its personnel
are hired by the boss of the activity that benefits from the computing.
This person should face the choice between computing personnel and
(say) research personnel. At present this gives the lowest costs
most of the time.
The ideal isn't that new users should tell CIT what they
want. They will simply have to learn by experience whether CIT
services, outside services or the micros sold commercially is
best for them.
∂31-Mar-83 2135 JMC
To: atp.bledsoe%UTEXAS-20@USC-ECL
I can't do automatic theorem proving at all. I could do
program verification but not a survey. It would have to be on one
of my current research topics. I could also talk abut non-monotonic
reasoning as an AI topic. I simply haven't been following the literature
well enough to do a proper survey.
∂01-Apr-83 1456 JMC
To: PB
1. I have been kibitzing the library on such matters for many years, but
since I haven't volunteered to actually do anything - even politick for
money - they have been only polite.
2. To me a program for handling checkout, recalls, etc. is not enormously
interesting, and I'm willing to wait for the central library to get
around to it.
3. I did ask when we could DIAL to Arlen from SAIL and use its
facilities to look up books in the general catalog and find out
what specific library they are in. The answer was in a few months
two years ago, but the 3081 with its software has proved inadequate, and public
access has been indefinitely postponed.
4. Another feasible project has occurred to me. The library already has
its catalog in the 3081 - at least everything acquired since 1971. We
could copy the computer science part of the file onto tape and put it
on disk on SCORE or SAIL and arrange for its updating. Then we could
examine the catalog from our terminals.
5. You overestimate our ability to direct master's degree students -
at least mine.
6. Politicking to get the Stanford Library system to allocate
50 cents to an ad hoc solution of the problems of the Computer
Science Library and CSD separately from the general library system
is probably hopeless. Anyway, I'm not inclined to try.
For example,I doubt the library system pays for the library's account
on SCORE.
Therefore, what CSD does must be done with CSD resources. I believe
the Department would allocate the disk space on SCORE or SAIL or on
the future file computer to store the catalog. Whether someone can
be got to move the file is more problematical. Do you volunteer?
∂01-Apr-83 1514 JMC
To: JEF@SU-AI
Dick Gabriel can represent my interests.
∂01-Apr-83 1709 JMC
To: PB
Let's begin by talking to Harry and Richard.