Author: Howard Exner
Date: 12:52:49 09/06/99
Go up one level in this thread
On September 06, 1999 at 11:47:16, Lanny DiBartolomeo wrote: >On September 06, 1999 at 01:14:49, Jouni Uski wrote: > >>After 5(?) games no wins so far. Bad, bad. Probably an aggressive program like >>Fritz, Chessmaster, Mchess have already won couple of games... >> >>Jouni > >I dont really know about this, > The Gm players have a month to prepare for only one game and use all there >energy in only one game,what if a GM says to himself ill play Rebel, and waits >and prepares 3months then says to Ed ok Ill play ? Also what if it was 4 or 5 >games in a row, this possibly would make a difference. This is an important observation that you make. I can think of no other person having to attain that recognition of GM strength by playing a series of games in this manner that Rebel is now doing. It's not the natural way of proving that one is of GM playing strength. GM's typically acheive their status through tournaments by playing a variety of players, both GM and non GM. Not in this long arduous way of playing well prepared opponents once a month. The cards are somewhat stacked against the computer opponent as computers do not possess that ability to prepare against an opponent as effectively as a GM. It just is not a strong component of computer play to prepare and adjust depending on who the opposition is. I would caution against using this format as a comparison to how humans gain that ultimate recognition of, "that person truly plays like a GM". Now in the end if Rebel manages to attain a GM performance of say 2550, that could tell us that there is a possibility that given ordinary playing patterns (ie: tournaments) that the 2550 could translate into something higher. Somehow the pendulum has swung. In the past we often pointed out unfavourable conditions for humans as in playing in noisy conditions, or that they received no monetary incentive to play their best. Now we are in a sense asking a player to play someone who has a month or so notice while it in a sense has no notice at all( for a computer, "knowing" ones opponent is marginally helpful and that only when the programmer attempts to include information on the opponent, but I can't imagine that as being much of a factor). This is not to make excuses for Rebel's two losses (If one counts Anand's win as well as Rhode's - toss the Hoffman game out unless your into counting ones and zeros), but simply to draw a comparison to the actual way humans have to acheive GM strength and recognition.
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.