staff_bob@gsbacd.uchicago.edu (01/25/89)
>In article <1470@tank.uchicago.edu>, staff_bob@gsbacd.uchicago.edu writes: >> It might be convenient for >> Minsky to ignore this sound, scientific research, just as it's convenient >> for him to ignore the fact that he hasn't made a significant contribution >> to the field of AI since it was in its infancy, and most of the trails >> he blazed have turned out to be dead ends, but that makes it no less >> valid. > >And statements that a major figure in AI research hasn't >made a significant contribution since the field was in its infancy are all >the better for evidence to back them up. > I regretted making this remark almost as soon as I had posted it. By way of excuse, I was responding to the following comment In article <7301@venera.isi.edu>, smoliar@vaxa.isi.edu (Stephen Smoliar) writes... >Granting that you HAVE >the sort of right hemisphere you want to "clean up" (and I tend to agree >with Minsky that one should be suspicious of such brain division), and was somewhat miffed at the idea that someone should cite Minsky as a refutation of brain laterality. As it turns out, there seems to have been something of a misunderstanding here, so we shall let that rest. By way of defense, I took a quick survey of some of the AI reference books that I was able to quickly get my hands on. Here are the results: In "The Artificial Intelligence and Expert Systems Sourcebook", Hunt(1986), the bibliography contains approximately 200 references. None cite Minsky. In "Inside Computer Understanding", by Schank and Riesbeck, out of approximately 100 references in the bibliography, one cites Minsky. It is dated 1963 and entitled "Steps Towards Artificial Intelligence." In "Artificial Intelligence", by Ranan B. Banerji, the bibliography contains 81 references. The single citation of Minsky is : Minsky and Papert, 1969. "Perceptrons: An Introduction to Computational Geometry." MIT Press In "An Introduction to Artificial Intelligence", out of 75 some references, 1 cites Minsky. That citation is dated 1975, and entitled "A framework for representing Knowledge", in "The Psychology of Computer Vision" ed. P.H.Winston. In "The Handbook of Artificial Intelligence" we find 5 references to Minsky out of some 400. Of these only two are less than twenty years old. These are the reference cited above from 1975 and a publication with the same title "A framework for representing knowledge" in J. Haugeland(ed.) "Mind Design: Philosophy, Psychology and Artificial Intelligence". Finally, in 'The Knowledge Frontier "Essays in the Representation of Knowledge"', Minsky is cited 3 times in a bibliography of some 350 citations. These citations are the 1975 paper mentioned above, a 1980 MIT Tech Report entitled "Learning Meaning" and a 1980 article in "Artificial Intelligence" entitled "Why People Think Computers Can't". This is not a very impressive resume for such a giant in the field, especially if we keep in mind that someone of Minsky's stature is capable of publishing almost at will. I'm sure that he has published a good deal, but in my cursory review, he doesn't seem to get quite as much mention as some othe big names, such as Berliner, Charniak, McCarthy, Newell, Simon, Winograd and Winston. I wish I could have gotten my hands on Winston's latest AI book. As both Winston and Minsky are from MIT, I would expect more mention of Minsky in Winston's bibliography. In this review, though, it seems that only Minsky's "Framework for representing knowledge" can be considered any sort of contribution that is generally cited and has been made in the last twenty years. As for my 'dead end' statement, I would like to retract it. It was based primarily on my apparently mistaken association of Minsky with perceptrons. I am now advised that Minsky's primary association with perceptrons was that he showed that they are computationally equivalent to the modern computer, effectively killing them. I am not attacking Minsky here, only defending myself. It should be easy enough to see how I could form such an opinion. Dr.Minsky is no doubt a giant in the field. Given that fact, and given all of the resources available to him, doesn't it strike you as odd that he is so little cited in broad scoped AI reference books? that was the basis for my remark, which I do consider somewhat ill advised.
demers@beowulf.ucsd.edu (David E Demers) (01/25/89)
It might be interesting to see who Minsky's grad students were, and what they have accomplished. He no doubt has had more influence than simply his own publications. Dave
sean@cadre.dsl.PITTSBURGH.EDU (Sean McLinden) (01/26/89)
To reduce the significance of Minsky's contribution to our understandind of automated intelligence and cognition to the number of citations which have appeared in the recent literature sounds like something that a tenure committee might do, but not a historian of artificial intelligence. How many people in this group who have never read anything of Minsky's have, nonetheless, been influenced by his ideas? I daresay nearly all. To criticize "The Society of Mind" and his opinion on the basis of a lack of supportive evidence might be entirely appropriate but to impeach him on the basis of a publication record seems to be a bit narrow minded. Minsky's contributions as a catalyst are, probably, immeasurable. Sean McLinden Decision Systems Laboratory University of Pittsburgh
bwk@mbunix.mitre.org (Barry W. Kort) (01/29/89)
I was distressed to see the recent postings on this newsgroup disparaging Minsky for insufficient publications. First, I would like to quote a brief passage from _The Tomorrow Makers_ by Grant Fjermedal, from his interview with Gerald Jay Sussman: "I think Minsky has got a good chance of having influenced more people than almost anybody I know in the computer world. Just take the number of heads of departments of computer science that are Minsky's students." [p. 78] Second, I would like to point out that early in my career, I wrote extensively, noting that my superiors were grading me on my publication record. I later discovered that almost no one was reading anything I wrote. Since then, I have scaled back my writing in favor of more one-on-one interactions which appear to be a more effective form of communication. Perhaps Minsky, too, has discovered more effective ways of disseminating knowledge than by burying his ideas in pounds of little-read papers. Creating brilliant minds would seem to be a greater achievement than creating brilliant documentation. (Ask any software engineer.) --Barry Kort
converse@tartarus.uchicago.edu (Tim Converse) (01/31/89)
The comments I've seen in this thread have mainly addressed the
question of to what extent research contribution is necessarily reflected in
publication record. Those on the pro-Minsky side have argued that there are
many ways that a person can influence a field other than through publication.
I agree, and Minsky is a great example of this sort of impact.
The running assumption, though, has been that Minsky's publication
record *is* sparse. Is this even true? I was surprised to find four
references to Minsky in Hopcroft and Ullman's text on automata and formal
language theory, including attribution of the proof that a two-counter
machine can simulate an arbitrary Turing machine. I suppose I shouldn't
be surprised, in that Theory and AI were probably closer and friendlier in
those days. :-)
====================================================================
-- Tim Converse-- 1606 E. Hyde Park #4E, Chicago, IL 60615
converse@gargoyle.uchicago.edu ..!ihpn4!gargoyle!converse
"If it wasn't for disappointments / I wouldn't have any appointments"