Author: Chessfun
Date: 06:01:34 12/13/99
Go up one level in this thread
On December 13, 1999 at 01:21:11, Christophe Theron wrote: >On December 12, 1999 at 21:08:43, Chessfun wrote: > >>On December 12, 1999 at 18:44:27, Christophe Theron wrote: >> >(snip) >>>Have you counted the percentage of games lost on time? >>> >>>I did not think it could even occur, and it looks like it changes the face of >>>the "Rebel-Tiger losing against a lower rated human player" debate. >>> >>>Too bad for some people that tried to destroy Tiger with this argument only in >>>their hands... :) >> >>I started this thread and have never tried to destroy Tiger, I have posted in >>fact that when it comes out I will buy it, even giving the opportunity for the >>reply that hopefully that will be before christmas, thereby allowing some to >>postpone purchases and wait. > >My post was not directed at you specifically. > That should say "at all" as stated I am more than interested in tiger being as good as possible since I intend to purchase it. > > >>It is true most of shutka's wins appear to have been on time but there were >>others where maybe due to Tiger's lack of time shutka clearly was winning. >> >>Personally I would prefer that when I play a computer any time control it can >>keep up and realize as a human does that the object of lightning or blitz is to >>win that the final position when the flag falls is irrelevant. > >Tiger loses on time probably because the GUI takes between 0.1 and 0.5 second >per move and that the engine is not aware of that. > >If the naked engine was playing it would not happen. For example I have never >seen the DOS version losing a game because of this under these conditions. > >One simple cure would be to enter a 0.5 second operator time, and it would not >happen anymore, or it would happen after maybe 200 moves. > >>I also do not see how this changes any debate on humans beating computers, the >>time controls 3/0 were posted from the very first thread when I posted I think >>two games. The argument could be used that the computer didn't have enough >>either cpu speed or time, but the fact remains that the final position is >>irrelevant. > >The final position is interesting when you know that the engine did not know >that for every move it was losing half a second... > >If we find out that this player was mostly winning on time by closing the >position and doing nothing until the engine ran out of time, then you can be >sure that with anti-human ON and a 0.5s operator time it will probably not be >the case anymore... > Clearly that is what he does, in fact it is what a lot of humans do at faster time controls. I agree those changes will help, it will still meet others that try to find a way around it, by using different time controls like 1/1 or 0/3 yesterday "fever" was trying 0/5 so they will always try to find a different time control. Again though that is one of the advantages of these tests finding out what can happen and how to make tiger as strong as possible under any circumstance. Thanks. >My opinion is: I think that Rebel-Tiger was not correctly configured and that >somebody just took advantage of this. That does not mean that Tiger is >unbeatable, but at least that it is not such an easy client. > > > > 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.