Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Computer-Chess Curiosities!

Author: Uri Blass

Date: 22:54:53 07/31/01

Go up one level in this thread


On July 31, 2001 at 17:26:54, Miguel A. Ballicora wrote:

>On July 31, 2001 at 12:44:24, Uri Blass wrote:
>
>>On July 31, 2001 at 10:38:39, Robert Hyatt wrote:
>>
>>>On July 30, 2001 at 18:33:50, Bruce Moreland wrote:
>>>
>>>>On July 30, 2001 at 18:05:56, Uri Blass wrote:
>>>>
>>>>>I believe that it is also possible to beat grandmasterrs by lines like 1.d4 h5
>>>>>2.e4 e5 with black if you try enough games.
>>>>>Does it mean that they do not "play chess".
>>>>>
>>>>>Uri
>>>
>>>I doubt you would beat any GM by playing h5 after d4, unless he fell over
>>>dead, laughing so hard...
>>>
>>>This is exploiting a known hole in an engine's understanding, and then doing
>>>it over and over.  Forget that nonsense with a GM.
>>
>>It may be interesting to do a mtach between the best chess program of 2010 and
>>GM's from the position after the moves 1.d4 h5
>>
>>I believe that the program has good chances to win the match.
>
>This has not been the point, the point is that somebody that is not even an
>FM can eventually beat a program with 1.d4 f5. Can somebody of that level
>beat a GM in those conditions? No way. That is showing the weakness of the
>program, hidden very well by an opening book.
>
>Regards,
>Miguel

After trying again and again for many games it is possible.

I believe that if the same person tries it again and again against a GM and
reset the GM after every game then it is also possible and even if the program
can learn you cannot prevent humans to delete the learning files after every
game in order to show you the game that they won.


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.