Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Why use opening books in machine-machine competitions?

Author: Claude Le Page

Date: 05:45:15 11/25/03

Go up one level in this thread


Hello Uri!
I thought of scientific experiments , to understand better some points of IA
without any inference of marketing topics that POLLUTE everything
Moreover , since the matter is with commercial engines any usr may repeat the
test and be sure that there was no cheat
Now the fact of falling into the fegatello trap is only a symptom of a much
broader failure ; correcting it punctually would not avoid to fall in all the
traps of the same kind
it's also possible to use specialized opening books: so , one can know the
general mood of the game, not prepare against it: else , it would be to change
the personality of the engine (what could be good), for instance one could
choose king's gambit for which there are excellent opening books : a program
"corrected" for all king's gambits would be a quite different program
I wonder if it is possible with some of the top programs (hiarcs9 ?)
 for the fegatello test , it would be of no use to correct only it
IMO , the failure to this test results either in a search not deep enough ,
or in a too drastic pruning , so a local correction would not prevent  to fall
in another trap of the same kind
you see, we are much beyond the SSDF tests, that IMO have no significance:
a program which is worth 2800 with King's gambit may be worth only 2000 with QGD
Orthodox and vice versa : where is the true ELO??
Friendly Yours
Cude Le Page



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.