Author: Tina Long
Date: 20:31:04 03/08/00
Go up one level in this thread
On March 08, 2000 at 22:41:26, Dave Gomboc wrote: >On March 08, 2000 at 22:36:46, Tina Long wrote: > >>On March 08, 2000 at 16:06:47, Chessfun wrote: >> >>>On March 08, 2000 at 15:21:34, KarinsDad wrote: >>> >>>>On March 08, 2000 at 07:45:03, Thorsten Czub wrote: >>>> >>>>>On March 08, 2000 at 07:23:27, Chessfun wrote: >>>>> >>>>>> >>>>>>Hi, >>>>>> In game 1 of my Easter tourney >>>>>>CStal runs itself out of time. >>>>> >>>>>the "bug" is known. >>>>>chris wanted to make cstal very human-like, >>>> >>>> >>>>Don't you mean very non-human-like? >>>> >>>>A human would (usually) know that it is playing a computer and that he/she must >>>>make good tactical decisions in order to win. CSTal does not know who it is >>>>playing, hence, it plays non-human, non-computer like chess (i.e. some form of >>>>quasi programatic approximation in between human and computer like behavior). >>>> >>>>Does CSTal have a mechanism to tell it that it is playing against a computer and >>>>hence, it should play tactically smarter and time control smarter? Or maybe a >>>>way to indicate the approximate ELO of it's opponent. Human-like play also >>>>consists of playing different based on perceived strengths and weaknesses of >>>>your opponent; not just playing randomly stupid. >>>> >>>>I mean, what good is playing sort of like a human against a computer? >>>> >>>> >>>>>so it plays incorrect sacs and also oversteps time-controls. >>>>>should not happen in 40/120. but in blitz-controls it happens... >>>>>if you would play in tournament-time-control, the problem would >>>>>IMO not occur. i have played in many tournament, and also >>>>>at home, and at home it uses it's own clock, on championships >>>>>i normally give myself an extra time between 3-5 minutes for >>>>>operating. >>>>> >>>>> >>>>>>It counts down from 60 secs and >>>>>>just ignores the countdown and moves >>>>>>20 secs after it. This was after 59 moves, >>>>>>I played the game out but am puzzled. >>>>> >>>>>:-)) >>>>> >>>>>so one point for chessmaster ! >>>>>consider: only the result counts. >>>> >>>> >>>>I'm glad to hear you say that! ;) >>> >>> >>>The problem is. It also timed out in game two although on >>>playing it out lost. But I don't see how in Didzis 1 hour >>>games it could have made any time control like that. >>>It make no attempt to even blitz as the 60 mins is up. >>> >>>I truly assume if I count losses on time in this manual >>>tourney "Easter" CSTal will lose a lot. >>> >>>Thanks. >>> >>A very unfortunate situation. >>- You can't give it draws for losses on time, that's not fair. >>- You can't adjudicate the positions, as Tal has used an unfair amount of time >>compared to the opposition who moved quicker to avoid time loss. >>- If you give Tal a loss & it loses many games on time, but not all, then the >>results are biased. > >Biased... towards the favour of programs that stay within the time limit? > >Um, isn't that the purpose of having a time limit? I think in computer v computer (home) tournaments, that the table of results is usually used to tentitivly rank program strength. The program that is being outplayed by Tal & then wins on time, is not necesarily better than the program that loses to Tal within the time limit. > >> >>I can only offer one solution for THIS time control: Reject Tal, & it's results >>so far, & treat it's position as a bye. > >I think just giving it a "loss" for a "loss on time" is quite fair too. I agree it's fair, but I still think that a win for "surviving" until Tal loses on time, biases the results. So I just wouldn't include CSTal in that type of time limit match. I'd also EMail the author requesting a Fix for this (in my opinion) Bug. Tina > >>Let it play next time you play with incremental time. >> >>Thanks for keeping us informed, >>Tina Long > >Dave
This page took 0.01 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.