Computer Chess Club Archives


Search

Terms

Messages

Subject: knowledge replacement?

Author: Mike S.

Date: 17:22:58 01/06/00


Dear chess programmers & fans,
since the use of larger and larger opening books and endgame databases with
"dynamic" access increases, I would like to ask if I have to fear that knowledge
(within the engine) is in danger to be replaced by databases?
In other words, do programmers rely for example that an opening book is always
used so that some sophisticated opening knowledge could be left away to gain
speed for the search? The struggle for speed and ply depth is of course
important for good chances in (direct) competitions with other programs, but
from the viewpoint of a fan and customer, I wouldn't like the knowledge to be
reduced.
I have no evidence of that, but it's also hard to say because the programs are
much different in knowledge anyway. I.e., in the opening, basics like not
bringing the queen out too early or to avoid blocking the d2-pawn with Bd3 etc.,
would surely be kept in the program but advanced things like center or tempo
related matters are also implemented often I think. At least this is my
impression, as the programs seem to have been improved much for the opening when
they are "out of book" early. It would be sad if this would be sacrificed for a
few percent more in the results of computer matches.
It's similar in the endgame. I think most programs know that KNN-K is a draw.
But when such built-in knowledge was removed, and the user doesn't have or use 4
piece tablebases (maybe because he doubts if they won't reduce calculating speed
too much, as I do), the program would be "silly" again... I'm afraid of that.

Hopefully there is no such danger? Please tell me.

Regards,
M.Scheidl



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.