Author: Dann Corbit
Date: 11:12:09 12/16/98
Go up one level in this thread
On December 16, 1998 at 00:48:43, Richard A. Fowell (fowell@netcom.com) wrote:
[snip]
>Hmmm ... what HIARCS calls "monitor mode"
>and what MacChess calls "Analyze Player vs. Player"
>
>Where are they these days on:
>- analyze EPD?
>- Next best move?
>- ability to copy analysis to clipboard/file?
All of those features are very interesting to me. I would like to see every
program have *at least* the EPD analysis ability of Crafty. Especially
interesting is the ability to score a test suite after it has been run. How
many tools can do that automatically? There is one measure I can do without in
Crafty's analysis: acs. I don't care so much how long it took because the epd
record does not contain machine information. Therefore, that measure has no
value whatever. Why report it?
>That reminds me I've been wondering what the "Analysis Features"
>people wanted to see when they ranked "Better Analysis Features"
>as the 2nd most desired improvement (after strength improvement)
>to the question posed in CCC poll question #9:
>
>"Which improvements would you most like to see in
>the next version of your favorite chess playing program?"
>
>What, specifically, were the features you wanted to see added?
>( I suppose it might be informative to include the name of the
>program you'd like to see this improvement in.)
I would like to see more focused, two-sided analysis. Analysis tends to be
centered on one player. Why not show white's best move and why. Then show
Black's best refutation and the reason.
My biggest problem with analysis from computer programs is that they do not
follow the EPD standard -- but branch off and do their own thing whenever they
feel like it {Crafty is the best adhering -- but the analysis code was written
by Steven J. Edwards so we should not be surprised.}. The sad part is that all
their little extensions could be handled by proper use of the standard anyway.
On the other hand, some programs offer little or no analysis. The Chessmaster
series, in particular is a pain in the butt to analyze with. I would love to be
able to utilize the King engine for some top quality analysis in the Chess
Analysis Project, but it is quite impossible to do so. I would also like the
programs to divulge all the information they have. Some programs, for instance,
do not divulge the number of plys ahead they are looking. I think this very
valuable. I think node count is of less value -- especially since slow
searching programs are at least on a par with fast searchers -- but the relative
differences for a single program could be useful.
As far as analysis -- one enormous feature would be to improve the display of
the results. For instance, they could animate the pv.
Finally, they should store the analysis in a database.
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.