Author: Christophe Theron
Date: 15:32:04 03/29/02
Go up one level in this thread
On March 29, 2002 at 15:06:51, K. Burcham wrote:
>
>
>Chris here is the point I was making.
>
>I have found that most of the time, if the program is + in eval then it seems
>some programs move very fast. most of the time if the move shredder expects is
>played then shredder will move instantly.
>all programs are not like this. and i am not saying tiger time management is as
>bad as shredder.
>what i was trying to say, is that it seems programs do not allow for opponents
>clock. for example, if you see the opponent is taking 3 minutes per move playing
>an equal player and the program is taking 1 minute per move, this always worries
>me. other things being equal, I look at strength in depth. not always. not in
>every posiiton. but sometimes there is strength in depth. either due to
>hardware, or time allowed. (not talking about pruning for depth here).
>
>it seems to me that programs always think every position is very difficult.
>that programs think every position is very deep.
>programs will keep thinking deeper, even in a won position.
>GM's dont.
>GM's in some games will take 5-8 minutes per move. sometimes this means position
>is difficult. if i let my program make a 1 to 2 minute move, four times in a
>row, i could be in trouble. then five moves later the program eval starts to
>drop because the program played into a deeper -eval position then it allowed in
>its time control to see.
>most here have noticed that the time control, when a program sees eval go from
>+eval to -eval, will start to look for more +eval line, and take more time to do
>so. we do know some things about the programmers time control, and how you want
>to make sure time is allowed for (possible later played line) to show +eval.
>
>there are many others that also are aware of this. some players at ICC will not
>play an opponent that is making manual moves, because of their disadvantage with
>the auto play software.
>also there are several at chess.net and ICC that i can get points from about any
>time I want, only because of the manual moves, but i do not do this.
>this is really useful in a 5/5 game, because most programs play so fast in a 5/5
>game, they play into losing position if opponent uses more time.
>
>i always feel that if there is a draw in long standard game, and both players
>have small but equal time left on clock, then it was a good game for time
>control.
>
>if i lose a game using a program, and i have lots of time left on clock, then my
>program game could have been stronger if i would have used more time.
>i dont want my program to wait until it sees its eval dropping, or until it sees
>a -eval to start taking more time per move.
>program time management is a skill that many gamers have learned at the game
>servers. i have lost several games because i let my program manage the time,
>while my opponent used up his clock and won.
>but----i have also lost several times in drawn or won positions because i ran
>out of time. this is the price you pay hoping eval will increase.
>
>anyway I know there are lots of people that agree with this, mostly gamers that
>know nothing about code and programming, and program time mangement.
>but i assume that there are also many that have not played this way that will
>disagree.
>
>so i will use Gambit Tiger as you recommend.
>and also I will set time control to 120 minutes. for this game I will not change
>clock in any way. Tiger will manage own time.
>agressive and anti-GM.
>
>kburcham
Tiger's time management has been tuned on thousands of games. While you might be
right sometimes in what you have noticed, the opposite also happens.
I prefer you to use the program's time management. As I told you, if you
interfere in any way in the game as an operator, I consider that it is not my
program who has played and I am not interested in the game.
If you want to interfere, then you are playing advanced chess, and that's a
different thing.
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.