Author: Dann Corbit
Date: 12:46:34 07/29/99
Go up one level in this thread
On July 29, 1999 at 15:29:20, KarinsDad wrote:
>On July 29, 1999 at 14:57:36, Dann Corbit wrote:
>[snip]
>>Here is the entire quotation from Amir:
>>"A correct evaluation is one that matches the winning percentages of the
>>position. I think white has about 54% in serious play, and if so the evaluation
>>should be about +0.20."
>>
>>I disagree that winning percentage is the correct and only variable to map to
>>centipawn evaluation.
>
>Yes, as winning percentages are limited to the skill of the players involved
>whereas a correct evaluation of a position (if it was perfectly done) would be
>based on the possibilities of the position and not upon what some players may or
>may not be able to accomplish with the position.
With complete information, either technique works. Even with such a technique,
a normal ce approach is better.
For instance, if I have every game possible in an imaginary database, I can
calculate the "real" win/loss/draw value of a starting position. However, many
games {probably 99.9999999999999999999%) would be idiotic. Hence, the true
value in actual play against top level opponents would be masked.
On the other hand, we could filter the book to include only the best choices at
each junction, forming a game tree as it were.
In such a case, we would end up with conventional ce's by simply scanning the
leaves of the tree.
Similarly, we could exhausively search each position with an 'ideal' computer
clear to the end {win/loss/draw}. If we stored each of these into a database,
we would form the same game tree mentioned above.
Both ideas have merit {human choices/computer choices -- not the
gedankenexperiment above...}. Computers perform tactical evaluations of
position. They are better at that aspect than humans are. Humans form
strategic evaluations of position. They are much better at it than computers
are. So, if we consider both values (human evaluation/computer evaluation) we
will have the best of both worlds.
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.