Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: CEGT: testing and presentation of results

Author: Graham Banks

Date: 20:30:53 10/20/05

Go up one level in this thread


On October 20, 2005 at 17:38:07, Uri Blass wrote:

>On October 20, 2005 at 10:00:11, Heinz van Kempen wrote:
>
>>On October 20, 2005 at 06:26:37, Ómar Skúlason wrote:
>>
>>>On October 20, 2005 at 00:04:43, Heinz van Kempen wrote:
>>>
>>>- snip -
>>>>
>>>>I regret that the discussion will surely have been annoying to the readers and
>>>>will try to avoid unkind answers of any kind from my side, seeing that there are
>>>>great efforts to make this forum friendlier overall.
>>>>
>>>>Best Regards
>>>>Heinz
>>>
>>>Hi Heinz!
>>>
>>>I think you are much to soft! Those discussions have not been annoying at all.
>>>Please keep posting as much as you can.
>>>Uri is one of the sharpest brains in this forum, and he is never rude.
>>>
>>>Ómar.
>>
>>Hi Omar,
>>
>>usually I am open to sharp discussion as long as not all turns to unfounded
>>speculations without real proof.
>>
>>Uri is nevr rude, you are right, but maybe a bit tenacious and pedantic and not
>>the kindest person on earth. But I already know him from Winboard Forum, old and
>>new one, and so I got accustomed. I was also one of his beta testers in former
>>times when Movei versions were very weak, without expecting a word of thanks for
>>many thousand test games.
>>
>>Well, he know proposed that CEGT should not test further Movei versions anymore
>>and this after more than 1100 games 40/40 2Ghz for his version 00_8_310, an
>>engine most other testers do not even care about in Blitz.
>
>
>I suggested it because it seemed to me that the CEGT testers are against me
>regardless of what I do so if they are against me they should never test movei
>regardless of playing strength of it in the future.
>
>I can also add that I almost did nothing about Movei lately and I do not expect
>a new version that is significantly stronger than movei00_8_295 or movei00_8_310
>in the near future.
>
>Uri


Hi Uri,

I'm not against you as such.
I do think your choice of words was unnecessarily harsh and certainly seemed to
question the integrity of CEGT testing even if this wasn't what you intended to
do. This is what upset people.
Remember that the CEGT team is made up of experienced testers who would
certainly notice anything unusual occurring in their testing.

The elostat system is notorious for being too generous with initial ratings over
a small number of games as Heinz has mentioned. You only need to look at the
decline of CM10th Pestilence in the rating list to see this is a common
occurrence.
It may well be that both Fruit 2.2 Uri, Shredder 9 Eccentric and Shredder 9
Columbus's Egg 9g are ALL better than the default settings at either shorter or
longer time controls than CEGT uses. I certainly know with 100% certainty and
have provided the proof to back up my claims that many CM10th settings perform
much more strongly than the default settings at longer time controls (much
different to the relative performances at CEGT time controls).
What you need is for many people who're prepared to check this out.

Regards, Graham.




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.