Author: Roy Eassa
Date: 11:41:48 04/12/02
Go up one level in this thread
On April 12, 2002 at 14:30:23, Dana Turnmire wrote: >>>Wouldn't a very fast stupid program be at a disadvantage to a much newer and >>>smarter one? From what I have read Deep Blue wasn't that sophisticated when it >>>came to forward pruning and general chess knowledge. >> >>Deeper Blue was _anything_ but a stupid programme, I see you really haven't any >>hard data on the whole Deep Blue project, which goes all the way back to >>Chiptest, which evoloved into Deep Thought, which became the "Manhatten Project" >>of chess playing machines. Deeper Blue was "smarter" than Fritz. >> >>Terry > >Maybe I was confusing Deep Blue with Deeper Blue. In an article by IM Larry >Kaufman in the Computer Chess Reports/Volume 5/page 37 he said concerning Deep >Blue which Kasparov beat: "Aside from the opening book, Blue's problems seem to >be a tendency towards planless play in closed positions, and in general a crude >evaluation function, which allowed Kasporov to outplay Blue positionally." > >Would it be said of the current top programs that they have a "crude evaluation >function?" Not at all. Their eval factors have to be highly tuned because they are in direct conflict for CPU time with each other and with the search. Deeper Blue had the luxury of having its eval factors implemented in parallel so more could be added without decreasing the use of the others or of slowing the search.
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.