Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Deep Sjeng - Ruffian really draw?

Author: Uri Blass

Date: 04:56:38 10/20/03

Go up one level in this thread


On October 20, 2003 at 07:55:29, Uri Blass wrote:

>On October 20, 2003 at 07:50:21, Gian-Carlo Pascutto wrote:
>
>>On October 20, 2003 at 07:14:59, martin fierz wrote:
>>
>>>this certainly doesn't look like -2.22 to me. white has the exchange, that's
>>>true, but e.g. an exchange of queens probably wins for black thanks to the
>>>monster pawn on b3. i'm not at all surprised that white does not win this...
>>>he'd have to mate black, and with only two pieces left that's not so easy!
>>
>>I agree - your last remark is actually also exactly the bug I fixed in DS 1.7
>>the same night.
>
>bug?
>
>A bug is when the program behaves different than you expect.
>If you did not implement some knolwedge it is not a bug because you cannot teach
>it everything.
>
>>
>>But every computer seems to evaluate this wrongly.
>
>In that case maybe correcting the wrong the evaluation to right evaluation can
>cause correcting right evaluation to wrong evaluation in different positions.
>
>Uri

The point is that there are position when the queen and rook are enough for mate
attack and you may evaluate them wrong.

Uri



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.