Computer Chess Club Archives


Search

Terms

Messages

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

Author: Mogens Larsen

Date: 01:52:42 03/07/01

Go up one level in this thread


On March 06, 2001 at 22:45:11, Chessfun wrote:

>The fact that Thorsten's are closer to the WMCCC neither makes it better nor the
>results more trustworthy. I would believe the SSDF or Enrique's results
>more than either the WMCCC or Thorsten's tourney run as it is. The number of
>games played and the fact that those programs are available to all users being
>the main reasons.

What does your preferences about tournament format has to do with the
possibility of modifying the setup of the participants from round to round? The
reasons you give for preferring the SSDF method is independent of that fact.
Some of the programs wouldn't be available to the user anyhow, nor would the
number of games increase.

The problem is that you're putting all the different formats into one pot and
decides their validity by the same standards. That is a narrow approach IMO and
the verdict is given beforehand. It all depends on the question(s) you want
answered. If it's a reasonable accurate rating (SSDF), a big tournament between
top programs (Cadaques) or an attempted simulation of a real computer chess
tournament (Thorsten).

All of them have strengts and weaknesses. The SSDF can't update programs
frequently due to the fact that they need established ratings. Furthermore, the
number of programs are limited, ie. very few amateur programs and some
commercial programs are missing. These limitations are mainly due to lack of
time/manpower (quarrels) I imagine. Some might want to complain about their
methods as well. The strength is the most accurate assessment of relative
computer strength available. However, the number of games doesn't promote
analysis or program improvement IMO unless Uri finds a problem.

The other two are "just" tournaments, using two different formats. The Cadaques
is narrow when it comes to participants and not very accurate due to the limited
number of games. But the increased number of games limits luck to a certain
degree. The human analysis was a helping hand to the authors and very useful.
Thorsten's tournament offers a lot of programs and a limited number of games.
Both of which can be an advantage. No details drowning in 23-17 game reports.
But some may go unnoticed due to the these limitations. Both tournament formats
are not about rating.

So what was the question? :-)

Mogens.



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.