Re: Advice on evaluation

[David Fotland <fotland@xxxxxxxxxxxxxxx>]

At 08:49 AM 5/19/99 +0800, Chen Zhixing wrote:
>David Fotland wrote:
>>
>>Thanks.  I think Goemate only uses a single value for group
>>strength, right?  The score doesn't change when the color to
>>move changes.
>
>   Handtalk uses a single value indeed. So far Goemate likes this.  I want
>to change it, but I have not designed the new scheme.  All go programmers
>can use two values just as you said, but how to utilize the values would be
>problematical.  I don't have enough idea on it.

I like Mark's idea of averaging the two values.  I think will give a more
accurate evaluation.

>
>>  One of the things that impresses me about Goemate
>>is that its evaluation is so stable from move to move.
>>
>   I know that you have this question before I contact you directly.  I
>think that you misunderstand it.  In fact, Handtalk/Goemate shows
>'superiority' data only after opponent's move.  As a result, you can see
>that it is so stable.

I knew this, but it is still impressive.  Many Faces is not so accurate.

David

>
>
>Chen Zhixing
>                        
>www.wulu.com
>
>
>