Author: James Swafford
Date: 08:10:47 01/09/00
Go up one level in this thread
On January 08, 2000 at 22:23:28, Arshad Syed wrote: >Quantum technology seems to be the next big leap in the field of computing. From >my understanding, it supposedly will be relatively much faster than todays >semi-conductor technology. Probably even a PC based on this method would be much >faster and powerful than todays supercomputers (DEEP BLUE??). Would it be >possible that such a computer would finally play 'perfect' chess - losing no >games, probably drawing a few in worst case scenario? > My understanding of a quantum computer is that it changes the fundamental composition (or quantum state) of the atoms that compose it. That description may be a little off, but I think it's close... So even if this entire (small) planet that we're on were somehow transformed into a quantum computer, there would not be enough atoms to represent the entire game tree. So I say no, quantum computers will not solve chess. (Which is fine by me, that would take all the fun out of it. :-) ) >Here are just my opinions: > >1.) A chess program/computer is only as good as its evaluation function gets. >While monsters like DB are good at calculations - tactical play, their >positional knowledge - light squares/dark squares, mobility etc. - is below that My feeling is that "knowledge" isn't absolute. It is some hybrid of the evaluation function *and* the search. I believe that as we search deeper and deeper, we can trim our evaluation functions (allowing us to search deeper still!)! This makes sense, if you think about it this way: We know *how* to solve chess - tablebases. Problem is that it's **practically** impossible to do so. But if we did have the means to construct that 32 man tablebase - we wouldn't need *any* evaluation terms. The score for each position is actually +1 (WIN), 0 (DRAW), or -1 (LOSS), isn't it? >of a GM. DB for instance in an earlier version (DEEP THOUGHT?) lost convincingly >to Kasparov. It was only after a year of programming by Joel Benjamin that it >managed to defeat Kasparov. I feel that computers at quantum speed, would find >more time to do extensive positional analysis in the eval function at the nodes, >which would help plug the holes in the program, which currently seem to me to be >more due to deficiencies in positional analysis. > >2.) Even the most powerful computers now are unable in the middle game to search >every single position. As a result some moves are ignored due to horizon effect. >With the new technology, maybe this will no longer be an issue. Add the improved I'm afraid it will be an issue for quite a while. >positional analysis, and the PC programs would be playing at the level of a >World Champion. > >3.) With ply-depth no longer a programming issue, programmers would be able to >focus on other issues such as AI - I mean genuine machine learning, which would >allow programs to even simulate human styles based on methods other than >weights. > I don't understand why programmers can't focus on AI with the "ply depth issue." If ply depth wasn't an issue, there would be no need for learning algorithms. >Just my 2 cents.... > >Regards, >Arshad Thanks... interesting post. :-) -- James
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.