Author: Christophe Theron
Date: 11:51:10 05/03/01
Go up one level in this thread
On May 03, 2001 at 01:51:37, Uri Blass wrote:
>On May 02, 2001 at 23:04:05, Christophe Theron wrote:
>
>>On May 02, 2001 at 12:28:09, Robert Hyatt wrote:
>>
>>>On May 02, 2001 at 11:29:27, Christophe Theron wrote:
>>>
>>>>On May 02, 2001 at 02:26:08, James B. Shearer wrote:
>>>>
>>>>> I was playing a Gambit Tiger 2.0 clone on ICC a couple of days ago. I lost
>>>>>my queen (for 2 pieces) before move 20 but Gambit Tiger failed to convert the
>>>>>win allowing the position to become totally blocked leading to a draw by the 50
>>>>>move rule at move 192. It seems there must be problems with the evaluation
>>>>>function for blocked positions (particularly when up material).
>>>>
>>>>
>>>>
>>>>The antihuman mode was not turned on I think.
>>>>
>>>>With antihuman mode it would not have happened.
>>>>
>>>>
>>>
>>>
>>>Why is this considered "anti-human"? IE do you want to draw such positions
>>>when playing a computer, but not a human???
>>
>>
>>
>>I do not want to use the kind of wild stuff I do to open the position if I'm
>>playing against a computer.
>>
>>My time management and evaluation are working together in the case of antihuman
>>mode. One of the behaviour I try to address is the case where the guy
>>1) tries hard to close the position
>>2) when he is done, move a piece back and forth very quickly until the program
>>runs out of time and LOSES
>>
>>The antihuman mode is able to recognize this and to take extremely strong
>>actions to avoid it. One possible action is for example to sacrifice a knight on
>>a pawn in order to open the position. I know I can do this safely against a
>>human player who has been using the close+shuffle strategy, because when I do it
>>I know the guy has not enough time left to compute correctly. So even if I have
>>sacrificied a minor piece in order to open the position, I'm almost sure to win
>>the game by outsearching the human player in the resulting open position.
>>
>>Of course it's out of question to use this against a computer.
>
>I think it is dependent in the level of the computer.
>
>If the computer opponent is at least 300-400 elo weaker than tiger then I
>suspect that it may be a good idea to use it against the computer and there are
>amateurs that are at that level.
>
>Uri
That's right, but at this time the only knowledge that Tiger has about his
opponent is "is it a computer or not".
The right way to do it would be to ask the kind of opponent (human or computer)
and the strength level (stronger or weaker then Tiger, in average).
But then you can go on by asking about the playing style of the opponent, then
ask for some games of him in order to find weaknesses in his play...
Christophe
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.