Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: program improvements based on weaknesses

Author: Robert Hyatt

Date: 20:24:45 12/15/01

Go up one level in this thread


On December 15, 2001 at 18:07:08, José Carlos wrote:

>On December 15, 2001 at 17:34:28, Robert Hyatt wrote:
>
>>On December 15, 2001 at 12:30:07, Will Singleton wrote:
>>
>>>On December 15, 2001 at 11:44:53, Robert Hyatt wrote:
>>>
>>>>On December 15, 2001 at 03:12:43, K. Burcham wrote:
>>>>
>>>>>
>>>>>
>>>>>if some here say that they dont want to play against someone that
>>>>>is moving manually (changes time to move during game),
>>>>>then it seems that the program time control could be improved for an increase in
>>>>>strength.
>>>>
>>>>
>>>>That isn't the issue.  It is "human interference" that is the issue.  A fully
>>>>automatic program (on a chess server) has to make _all_ decisions by itself.
>>>>How much time to use, when to think longer, when to play quicker, etc.  A
>>>>manually operated program will have help.  And with a good human operator,
>>>>that help can be the "edge" needed to beat an automatic program.  IE when
>>>>the program finds a move that the human is sure is "ok" he can hit "move
>>>>now" and save time.  Or even prevent the program from thinking longer and
>>>>switching to a "worse" move.
>>>>
>>>>time usage isn't really the issue, it is "outside help".
>>>>
>>>
>>>That's true, but both the "move now" button and the "don't make that stupid
>>>move" button can be used even on a fully automatic program.  Can't control for
>>>that.
>>>
>>>But I do think it's important to kib analysis, for later study.
>>>
>>>Will
>>
>>
>>How would you say "don't make that move"??
>
>  If you use an automatic interface you wrote yourself, you could implement
>anything you want, for example 'next best during search' button. Even under
>winboard you can send commands to the engine, so it is easy to cheat, I'm
>afraid. You can have in your program (not you of course, yours is open source,
>but anyone else) some commands for 'move now', 'add n seconds to search this
>move', 'pv move is bad, try something else', ... And just send the right command
>trough winboard during the search.
>
>  José C.
>



That's not quite the same, for this discussion.   IE as a manual operator,
I can either manually let it search longer if I don't like a move, or I can
force it to try something else and if the analysis shows it isn't a blunder,
that move gets played.

Just saying "play second best" seems dangerous and certainly less effective.



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.