Main
Date: 31 Dec 2006 08:52:40
From: Dave (from the UK)
Subject: Measuring engine strength
I'd like to estimate the playing strength of Phalanx

http://sourceforge.net/projects/phalanx

on some of its weaker settings. There is a parameter one can set between
0 and 100 which sets its strength. I've not looked at the docs
carefully, but I believe the strength should be independent of CPU time
- it will just take longer to move on slower hardware.

Is there any logical way of testing the strength of such an engine?

Perhaps I could put it on FICS (do they support chess computers?) and
get it a rating. (I have an ICC account, but not one for computers). Not
sure if there is any better method.

Thought about asking the authors of some of the chess engines on ICC if
they would share the code. Then I could play Phalanx against those
engines to get an idea where it ranks when its strength is changed as
per the docs.

I'm well aware computers are not people and so the ratings are not
directly comparable, but to get an idea of strength would be useful.

Someone has suggested Phalanx can play from about 1200 to 2200, so that
is the sort of range I am looking at.
--
Dave (from the UK)

Please note my email address changes periodically to avoid spam.
It is always of the form: [email protected]
Hitting reply will work for a few months only - later set it manually.

http://witm.sourceforge.net/ (Web based Mathematica front end)




 
Date: 31 Dec 2006 17:24:21
From: David Richerby
Subject: Re: Measuring engine strength
Dave (from the UK) <[email protected] > wrote:
> Perhaps I could put it on FICS (do they support chess computers?)
> and get it a rating. (I have an ICC account, but not one for
> computers).

FICS does support computer accounts, yes. ("help computers".) But
that would only tell you how well the computer performs against other
accounts on FICS. You could approximate a FIDE rating from that but
no more.


> I'm well aware computers are not people and so the ratings are not
> directly comparable

(The fact that humans are not computers is not the reason that the
ratings are not directly comparable.)


Dave.

--
David Richerby Flammable Hi-Fi (TM): it's like a
www.chiark.greenend.org.uk/~davidr/ music system but it burns really
easily!


  
Date: 31 Dec 2006 22:15:38
From: Simon Waters
Subject: Re: Measuring engine strength
> FICS does support computer accounts, yes.

They have fairly stringent rules on setting them up, which mean
wandering in and trying to get one to see how good an engine is isn't
likely to get someone new an account.

Phalanx XXII has played in a lot of computer tournaments, at a fair few
time controls, and so a quick search finds plenty of data to assess
relative strength against computers.

Substantially stronger than GNUChess 5.07, weaker than the leading free
software chess engines (even when XXII was released I believe Crafty had an
edge).

http://wbec-ridderkerk.nl/

I wouldn't like to hazard a guess at its performance against humans
(other than the almost ubiquitous "better that most humans").

And Google found this;

http://www.vanheusden.com/cchess/Debian/20061201.php

Which says pretty much the same thing, but covers all the free software
chess engines in the next release of the Debian operating system.


   
Date: 01 Jan 2007 17:46:50
From: David Richerby
Subject: Re: Measuring engine strength
Simon Waters <[email protected] > wrote:
> Phalanx XXII has played in a lot of computer tournaments, at a fair
> few time controls, and so a quick search finds plenty of data to
> assess relative strength against computers.

But that will be Phalanx playing at full strength. Dave wants to know
the strength of the dumbed-down modes.


Dave.

--
David Richerby Impossible Carnivorous Whisky (TM):
www.chiark.greenend.org.uk/~davidr/ it's like a single-malt whisky but it
eats flesh and it can't exist!