Author: Christophe Theron
Date: 20:59:39 04/09/03
Go up one level in this thread
On April 09, 2003 at 01:52:24, Mike S. wrote:
>On April 09, 2003 at 01:30:04, Kurt Utzinger wrote:
>
>>On April 08, 2003 at 20:11:47, Christophe Theron wrote:
>>(...)
>>>I'm convinced that the book learning of the Chessbase GUI is much better than
>>>the book learning of the Chess Tiger engine itself. It is aggressive and has
>>>proven its efficiency, at least in comp-comp games.
>>>(...)
>> Yes, this would indeed be a very intersting experiment. Needed are however
>> many hundred games. My feelings go in the direction that a better tuned
>> book [the Lokasoft book by Jeroen Noomen] brings more than a normal book
>> [ChessBase] with aggressive book learning.
>
>If we have the SSDF list in mind, I guess book learning may be somewhat
>overestimated in discussion of the past. Because, it's most effective in
>
>- *very* many games (when long opening variants are *repeated at all*),
>- against the *same opponent*,
>- and not to forget: on the *same computer* (SSDF uses more comps for the same
>engine AFAIK, so the book learnig effect would be *split*)
>
>Furthermore I tend to see that element of a chess engine's performance (the Elos
>gained or lost due to the book and book learning) as something entirely for the
>"sport" aspekt of the competion. Interesting to watch, but not really important
>for the practical usage of the average home user. For him, a solid big book, as
>up-to-date as possible and not missing unusual lines, is sufficient.
>
>Unfortunately, it's more or less unknown how big the book "influence" on common
>rating lists really is (except those which don't use the program's books of
>course*). I guess it's less than 20 Elo (but that still can mean 3 ranks up or
>down though).
>
>*) Even if "standard" books or opening databases are used, you could argue that
>there may be some book influence still, if some engines are more dependant on an
>individually designed book than others (theoretically).
>
>Regards,
>M.Scheidl
Maybe it's not important for the practical usage, and I tend to agree, but it
might very well be that some engine(s) get overestimated because of the effect
of book learning.
And as people use the results of rating lists to decide which program they are
going to use for their "pratical usage", I think this question becomes of
tremendous importance.
Christophe
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.