Author: David Eppstein
Date: 21:22:38 11/06/99
Go up one level in this thread
On November 06, 1999 at 20:04:27, Rajen Gupta wrote: >I do not know why such a big fuss is made about ''how ''deep blue makes its >moves or whatever-after all its just a supercomputer powered by hundreds of >regular powerpc chips with the added bonus of having an additional few hundreds >of chips specially optimised to do chess calculations together with an analysed >database of every single gm game ever played on this planet. You are the one making a big fuss, I was just reporting on a recent computer chess article that happened to involve a program you don't like. I was interested in this article because it gave a reasonably clear explanation of how to do something constructive with the huge piles of GM games one can download easily these days. In the only other program for which I've seen a description of this phase of the game (crafty), this information is used differently, by condensing it down to the traditional move-tree type book, so it was interesting to compare with the approach here of using the database for bonuses in a regular search. One thing I wonder is how they give bonuses to moves rather than to positions. Doesn't that mess up the search? There are other situations where you'd like to do that sort of thing, e.g. to encourage developing moves when out of book in the opening. It didn't sound like anything in Campbell's article required special hardware. Anyone else could be using similar techniques.
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.