# Computer Chess Club Archives

## Messages

### Subject: Re: likelihood instead of pawnunits? + chess knowledge

Author: Robert Hyatt

Date: 13:34:40 10/25/02

Go up one level in this thread

```On October 25, 2002 at 14:29:59, Uri Blass wrote:

>On October 25, 2002 at 13:11:44, Robert Hyatt wrote:
>
>>On October 25, 2002 at 12:39:38, Ingo Lindam wrote:
>>
>>>Hello,
>>>
>>>I repost my former post under this new title just hoping to encourage
>>>more people to join the discussion:
>>>
>>>I am new at the Computer-Chess Club and would like to discuss some
>>>suggestions for (a new generation of) chess knowledge using (and
>>>generating?) chess engines. During my time at the university and at my
>>>first job after making my exams in computer science I was involved in
>>>statistical speech/pattern recognition and machine translation. That
>>>might atleast a reason for some of my ideas.
>>>
>>>I am not sure whether these suggestions have never been made or just
>>>named to be impossible to implement. (I am sure they are not.)
>>>
>>>I would really like to see the computers measure a position rather in a
>>>set of probabilities e.g. (P+,P=), where
>>
>>
>>I think that if you look at what chess programs do, this is the essence of the
>>evaluation.  The larger the number, the greater the probability that side will
>>win.  The smaller the number, the greater the probability that side will lose.
>>Scores near zero imply draw, of course...
>
>Not exactly.
>
>You can translate pawn to expected result but not to probabilities.
>
>The expected result is the same in the following 2 cases:
>probability 1% win for white and 98% draw
>probability 40% win for white and 20% draw.
>
>The probabilities are not the same.
>
>Uri

And I would hope the scores are not the same for those two cases either.

The 40% win probability should (IMHO) have a higher score than a position
with a 1% win/98% draw score...

The latter should be near to 0.00

```