Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Is SSDF detrimental to advances in Computer Chess?

Author: Amir Ban

Date: 12:49:01 11/19/03

Go up one level in this thread


On November 19, 2003 at 06:14:12, James T. Walker wrote:

>On November 18, 2003 at 19:38:56, Amir Ban wrote:
>
>>On November 18, 2003 at 09:17:11, James T. Walker wrote:
>>
>>>What do you think?  Does the SSDF list which is so important to commercial
>>>programmers cause them to ignore what is important in beating the top humans?
>>>Are almost all amateurs falling into the same trap in trying to compete with the
>>>professionals?
>>>Jim
>>
>>If you take SSDF ratings at face value the top programs are actually
>>overperforming against humans.
>>
>>Amir
>
>Geez, I'm not sure what you mean by that statement.  It seems obvious to me that
>the SSDF ratings are not comparable to FIDE ratings.  If they were then I guess
>what you are saying is right.  Do you actually believe that Shredder on a 1200
>Mhz cpu would carry a 2800+ Elo in human events over several hundred games?
>Just curious.
>Jim

It's my turn not to understand your meaning.

You say if top programs did not focus on SSDF, they would have a better chance
of beating the top humans, i.e. to be rated ELO 2850-2900.

The top programs can only draw the top players, i.e. they perform about 50
points better than their SSDF rating.

Your question about Shredder means you believe the SSDF ratings are inflated.
Maybe, but how does a draw with Kasparov show that ?

If the top programs are overrated, then it doesn't show. Starting with Deep
Junior in Dortmund 2000 they have consistently performed over 2700 against
humans.

If it were SSDF that's holding them back, then presumably there should be a
not-so-top program that is able to beat the top humans. I don't think so.

Amir




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.