Author: Enrique Irazoqui
Date: 08:10:02 06/28/00
Go up one level in this thread
On June 27, 2000 at 18:13:26, Fernando Villegas wrote: >Tricky issue is to use the "understanding" word for this kind of problems. Some >tends at once to think of it as something with a human-like kind of >understanding and from then on they say not just one program understand a shit. >But as far my knowledge and experience of CSTAl tell me, this program really >"understand" in the sense it has code lines to tackle these kinds of situations. >That is, in hardware or software realm, understanding. >Fernando Fernando, take a look at this: ______________________________ The forgotten, drowsy shack under the black pond this cloud forgets itself ______________________________ Let us go then, you and I, When the evening is spread out against the sky Like a patient etherized upon a table. . . . ______________________________ One is a poem written by T.S. Eliot. The other piece, by a 2K. True Basic program of Kemeny, consisting of an elementary set of rules and tiny databases of articles, nouns, pronouns, adjectives, verbs and adverbs. One is poetry, the other is garbage. Now let's imagine that the program has a truly sophisticated set of rules and all the vocabulary, but still no semantics, no information about the meanings of words and how they influence each other. This program could write a perfect sonnet, or in due time come with something starting with "Life is a tale told by an idiot..." The similarity with Shakespeare's "Life is a tale told by an idiot..." would be only apparent and meaningless. "Pierre Menard, autor del Quijote" is a quite fascinating short story of Borges that talks about this. Nevertheless, if it is a matter of constructing a formally perfect sonnet, the machine could regurgitate one after another, and yet none of them would be poetry. I mean that being a versifier and being a poet are very different things, similarities notwithstanding. If we measure (god forbid!) poetry by the formal perfection of sonnets, this program could be a strong candidate to the Olympus, or at least to the Nobel prize. If we don't, it would be considered as what it is: an idiotic machine that doesn't understand a damn thing of what it does. That's an extreme case, but to what extent do you think it applies to computer chess? Let's say that versifying is search, and then what Tal and Spassky and Shirov try to do is missing in CC, because it is not only difficult but it will slow the search down and the perfect sonnet won't come in time? At least I think we could agree about the fact that programs versify and little else, so they play a competent and ugly game of chess, and that's why their tactics, but seldom if ever their "ideas", are valuable to us. Enrique
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.