Author: Christophe Theron
Date: 15:58:23 11/18/00
Go up one level in this thread
On November 18, 2000 at 16:14:53, Uri Blass wrote:
>On November 18, 2000 at 15:52:11, Stephen A. Boak wrote:
>
>>On November 18, 2000 at 14:47:09, Uri Blass wrote:
>>
>>
>>>2-3.46 times faster mean 70-130 elo better based on the assumption of 70 elo
>>>from doubling the speed and I do not consider deminishing return from speed.
>>>
>>>It means not more than one point in 6 rounds.
>>
>>Uri, my friend, you forget your statistics (mathematics). You did not say 'on
>>the average', but I assume that is what you meant.
>
>It is clearly what I meant.
>
> tsk tsk! You know you can't
>>talk in such absolute figures!
>>
>>There is a good reason for the concept "bell-shaped curves". It reminds us that
>>measurements and performances vary from time to time. That is why the concept
>>"average" has a useful meaning.
>>
>>Many results are above or below average. Is the current result of GambitTiger
>>above or below average? How can you tell, from a single tournament result?
>>
>>How can you extrapolate (predict) the result of using a faster computer (on the
>>'average') if you don't think about this issue?
>>
>>>
>>>The difference between Junior and Gambittiger is bigger than one point so the
>>>expected results from better hardware are not enough for gambit to win the
>>>tournament.
>>>
>>
>>It is possible that the current difference in score between Junior and
>>GambitTiger would not be greatly altered (change of tournament leader) based on
>>GT using a faster CPU.
>>
>>On the other hand, GambitTiger may be performing below its average level in this
>>particular tournament. [Note: its opponents may be playing a bit above their
>>average levels.]
>>
>>Its results may be a bit below 'average', relative to the opposition. If so, it
>>might easily perform at or above average, in another tournament. Combined with
>>a faster CPU, it might even win another such tournament.
>
>I agree and I did not say that Junior is better than Gambittiger but only that
>the result of Junior is better than the result of Gambittiger in this tournament
>if you translate it to equal hardware result.
>
>The final result of Junior is 1.5 points more than Gambittiger and hardware that
>is 2 times faster can explain in average difference of less than 1 point out of
>7.
>
>I explained in a previous post that the result of chesstiger in chessfun's games
>suggest the opposite and my point was not to say that Junior is better but to
>say that the difference is not big and every program out of the 2 can win a
>tournament.
>
>I posted it in response to another post of someone who decided to uninstall
>Junior,Fritz and Nimzo after buying Gambit.
>
>Uri
Then maybe, to be fair, you could also say how many elo points above the
competition the results of Gambit Tiger in the Dutch championship means?
You'll find the results of this championship here:
http://ourworld.compuserve.com/homepages/thstorm/docc00.htm
Christophe
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.