Author: Uri Blass
Date: 04:50:12 02/18/02
Go up one level in this thread
On February 18, 2002 at 07:36:26, Enrique Irazoqui wrote: >On February 18, 2002 at 03:23:40, Ed Schröder wrote: > >>When reorganizing my mailboxes I came across and old posting of 2000, here is: >> >>========================================================================== >>Posted by Enrique Irazoqui on July 17, 2000 at 07:33:41: >> >>In Reply to: Re: About head or tail (was Upon scientific truth - the nature of >>informati posted by Ed Schröder on July 17, 2000 at 07:15:45: >> >>On July 17, 2000 at 07:15:45, Ed Schröder wrote: >> >>>>I know. Prove me wrong. :-) >>> >>>How about a 10 game match....? >> >>What for? What a waste... Comp-comp won't prove a thing no matter how many games >>you play. Let's take a quick look: >> >>1 - Programs are helpless against anti-computer strategy, like Fritz in >>Frankfurt and Junior in Dortmund. Their performance is inversely proportional to >>human awareness of this shortcoming, and search alone won't solve the problem, >>or at least it won't solve it before we all become very bald. Oh yes: in >>comp-comp search is everything. >> >>2 - Programs are essentially polite social beings: they behave like GMs amongst >>GMs, like 2300s amongst 2300s. For instance, look at Junior's performance in >>Dortmund and in the Israeli league. >> >>3 - If program A has extra code to avoid closed positions and program B does >>not, comp-comp won't show the difference as an advantage for A. If B is a faster >>searcher, the extra code will harm A when playing B. >> >>4 - Comp-comp games show a partial and rather uninteresting picture, their >>results don't necessarily correlate to human-comp and watching them can even >>become a threat to one's mental health. >> >>Now go figure the statistic certainty of 10, 100 or 1000 comp-comp games. >> >>Enrique >> >>========================================================================== >> >>I think this was one of the best postings ever made in CCC. Point (2) states >>more or less (and I deliberately put it strongly) that in Man vs Machine the elo >>of the GM is irrelevant. >> >>If true it makes "little difference" to play a 2500, 2600, 2700 or even a 2800 >>elo rated player if he (she) doesn't know how to play a computer. >> >>If the above is true and if GM van Wely doesn't know how to play a computer >>Rebel should win, same as Fritz should beat Kramnik. > >I don't know about van Wely's preparationagainst computers, but if Kramnik wants >he would beat Fritz 8-0. Yes, this is my usual understating self. :) The question is if you assume that kramnik gets a determnistic program that he play against(I assume for this discussion that he gets the program that he plays against it). A possible option for the programmers is to make Fritz not deterministic even if it means that it is 50 elo weaker so it can play different moves in different days and even if kramnik gets the program he cannot prepare the win before the match. They may for example decide that the pawn gets a random number between 0.95 and 1.05 and not 1. I believe that changing the value of the pieces by not more than 5% is going to reduce the strength of the program by probably less than 50 elo in comp-comp games because chess is mainly tactics but it may increase the strength of Fritz against kramnik. I expect kramnik to win but I guess that the result is not going to be 8-0 for kramnik even if he gets the program if the Fritz guys use some randomness in the evaluation. Uri
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.