Author: Ralf Elvsén
Date: 06:09:10 04/18/01
Go up one level in this thread
On April 18, 2001 at 03:38:46, Uri Blass wrote: >I suggest the sponsors the following idea: > >Kramnik will get no information about the program that he is going to play >against(he is not going even to know if he plays against DeepFritz,Deep >Shredder,Deep Junior,Ferret,Gambittiger, Dark thought,Crafty or maybe Deeper >blue). >He is only going to know the fact that he is playing against a chess program. > >If kramnik cannot win the match then he is going to get a revenge match against >the same program at the conditions that he wants(he get a copy of the program >that he is going to play). > >I guess that kramnik is going to agree to this idea if enough money is involved. >He should get more money for winning the first match when the program is unknown >but also a lot of money if he wins only the second match and the same money that >is suggested to him today for participating if he is losing the 2 matches. > >Uri I think your idea is interesting if one has the perspective of "data collector" for estimation of machine vs human - strength. I doubt this is the perspective BGN has :) But one half of your idea, namely that Kramnik musn't have a copy to practise against should be a demand from the programmers in my opinion. Note: I am not fully aware of the meaning of this "hand over a copy beforehand" but if it means that Kramnik can observe how the program plays I think it is unacceptable. Without knowing the history of computer chess in full, my view is that the question "can we make a combination of software + hardware that plays better chess than the best human" is the most fundamental. The rules of this game is that the human plays as against an ordinary opponent. The computer is whatever the programmer wants to throw into a box before the game (HW, SW, EGTBs, opening book etc). When the game begins the "box" must do on its own (apart from an hopefully unimportant operator). If the rules are changed so that the human can experiment with the "box" before the game, it is unfair IMO. This should only be done after a win by the computer, along the lines Uri suggest. If Kramnik wants e.g. copys of testgames played by the program before the match that would be acceptable, since the programmer has the corresponding information. A copy to practise against would be very strange, and if this is the case I am surprised that the authors of Fritz and Junior accept it. Or will the copy just be locked into a valve for three months? Doesn't make sense... Ralf
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.