Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Let's say Over 2300

Author: Terry McCracken

Date: 20:33:38 03/06/02

Go up one level in this thread


On March 06, 2002 at 15:52:50, ALI MIRAFZALI wrote:

>On March 06, 2002 at 02:49:48, Joshua Lee wrote:
>
>>Ok really Any program as strong as Rebel Century is over 2300 but let me put
>>this in perspective : Rebel lost two games for which it performed 400elo less
>>than it's opponent. Rebel won two games but how many people can Beat Rebel.
>>To beat Rebel once in 1000 games you would have to be within 399.2elo
>>We don't yet know exatly what Rebel's elo would be so for safety's sake 2697
>>because that's what it's performance was. To beat Van Wely 1 out of 100 games
>>Rebel would have to be atleast 2305 , 1 game out of 50  : 2313elo  , 1 game in
>>10 : 2377elo , we don't know how the performance would turn out if 20+ games
>>were played but what if rebel lost the rest +2 -22  which equates to 2363elo
>>So in my opinion more games are needed to get a more accurate picture.
>>To win 2 out of 1000 Rebel had to be atleast 2298elo  (within 398.4elo) , 2 out
>>of 100  2313elo (within 384) 2 out of 50 2329 (within 368), 2 out of 25 2361elo
>>(within 336) I can say for sure that Rebel is over 2300 but by how much??
>>To win 2 out of 6 Rebel would have to be 2563 (within 134 elo)
>>So to recap Rebel is within 800elo 1897  too low , within 400 ..2297elo  this
>>still sounds too low ....
>>To win 2 out of 8 Rebel Had to be atleast 2497 (within 200elo) this sounds right
>>as well as 2377.  2 out of 10 means that rebel is 2457 or within 240elo)
>>this sound reasonable as Rebel would have to be somewhere within 300elo to score
>>a win in the first place and secondly Rebel had played many other GM's on weaker
>>Hardware and it's average after many games was over 2400
>>
>>To beat rebel one would have to be within 400elo of whatever rating it is,
>>There are 3009 people on the FIDE list with a rating of 2377 so I would like to
>>see more games or i would like to see if it can get an actual FIDE rating.
>>
>>I didn't figure out draws so it could be very acurate even with the four games
>>but it is possible that there are lots of draws , and a few more wins maybe one
>>more
>>
>>
>>It would've been nice if just one more game had been played as even if Rebel
>>Lost this would put it at 2617 performance and in my opinion more than 4 games
>>is always necessary to tell if one player is stronger than the other. I think
>>Statistics Agree.
>>
>>I would like to hear what others think the outcome would be if more games are
>>played.
>If more games were played the GM would have been crushed completely.AS I have
>said before you underestimate Computers immensely.

Computers "underestimate" me! :o)



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.