Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Updating engines during tournaments? (Odyssee Tournament)

Author: Robert Hyatt

Date: 21:18:59 03/07/01

Go up one level in this thread


On March 07, 2001 at 14:05:20, Andreas Schwartmann wrote:

>On March 06, 2001 at 18:37:15, Robert Hyatt wrote:
>
>>Run an event any way you want.  That is what this seems to be about.  But I
>>don't see anything wrong with Thorsten's approach of letting people fix bugs
>>as they are detected.  Will it _really_ corrupt the final results?  Did
>>introducing changes after each round of the WMCCC event prevent shredder
>>from winning?  Did shredder get changed?  Does it matter?
>
>
>Sure it does: Who really won? Shredder 5? Shredder 5.01? Maybe even Shredder
>5.2? The only thing you actually CAN say is SMK has won the tourney with his
>engine(s). And I guess you can say that it's not that the strongest engine won,
>but the best-modified-between-rounds-engine did win. Maybe Shredder only won
>because SMK put in some hours in modifying the engine between rounds in order to

Quick time-warp back to 1973.  Fischer is saying "it isn't clear who is the
best player with today's system, it is more a statement of who has the best
group of "seconds" that are up all night analyzing adjourned games, preparing
opening TNs for the next game, etc..."

Don't humans do _exactly_ the same thing?

The concept of a "constant" chess player is simply not reality.

In the 1986 WCCC, the CB that played rounds 1 and 2 was different by 4 lines
of code from the CB that played the last 3 rounds.  I have no doubt but that
had we used the same program for all 5 rounds we would have been lucky to
break even at 2.5-2.5.  By fixing a bug I introduced, we finished 4-1 and
won the tournament.  Were we the best there?  Based on my chess skill and
knowledge, yes we were.  Would we have won without the bug fix?  Not a real
chance in hell.

Did that taint the result?

Not IMHO...





>make adaptions regarding the next opponents. Maybe Gandalf did not win, because
>Steen was not present at the tourney and could not make modifications. Maybe
>Gandalf was Gandalf 4.32h all the time, while Shredder came in many
>incarnations, each one fine tuned to a particular opponent. So these Shredders
>won. Does this say anything about Shredder's real strength? Nope. Which Shredder
>version anyway?
>
>Andreas


If you don't go, you _definitely_ give up a bit.  No doubt about it.  I am
usually unable to attend WMCCC events... they are too long for me to miss a
week of classes.  The trips are too expensive when family vacation time is
considered a precious commodity.  And I never have any real hope of winning
such an event not being there.  If I watch, I learn something _every_ round,
and can often spot problems that don't lose in the current round, but which
might lose in future rounds if the right circumstance arises.  I try to fix
them a.s.a.p. of course.  If I am not there, that 'feedback loop' doesn't
happpen.



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.