Author: Sune Fischer
Date: 13:50:49 11/12/03
Go up one level in this thread
On November 12, 2003 at 16:40:32, Matthew Hull wrote: >On November 12, 2003 at 16:14:45, Sune Fischer wrote: > >>On November 12, 2003 at 14:51:11, Robert Hyatt wrote: >> >>>On November 12, 2003 at 01:36:03, Hans Meiser wrote: >>> >>>>Hi, >>>> >>>>http://www.vrichey.de/cct6/ >>>> >>>>The Problem of the tie-break: >>>> >>>>Suggestion: >>>>(1) two programs tied for the first place >>>> a normal game 45 15 between the two leaders. >>>> if draw two games 5 3 round robin >>>> >>>>(2) three programs tied for the first place >>>> Six games 5 3 round robin >>>> >>>>best regards >>>>volker >>> >>> >>>My suggestion: >>> >>>two tied, a normal 45/15 game. If draw, we just have "co-champions". >>> >>>three tied. three rounds of 45/15. a plays b, b plays c c plays a. >>>If any ties are left, we have co-champions. >>> >>>Otherwise, let's play blitz from the start. I see no valid reasoning to >>>use blitz to tie-break a "standard time-control" event. Having co-champions >>>is (IMHO) better than a crap-shoot blitz match, where quick searches often >>>lead to wins/losses that have little to do with the actual playing skill of >>>the programs being used.. >> >>What is the fundamental difference between searching to ply 11 and ply 15? >>IIRC you have sad you _don't_ believe in "tactical barriers" :) >> >>Anyway, I see no difference and I don't believe blitz would be any more of a >>crap-shoot than longer time controls > > >Then just make it a blitz tourney, right? If it makes no difference, then make >it a blitz. Then you could do a double round-robin and make it more >statistically significant or something. > >But that's not the point, is it. It must be slow time controls for a reason. Yes and there are a couple of reasons. One of them is that the games become of higher quality, another is that one has time to follow the tension, read the kibitzed info and chat with the opponent. However, for a tie-breaker all you need is to break the tie. I'd hate for the tie-breaker to be played the next weekend, with blitz we have a winner the same evening and we don't have to suffer this anti-climax. >That reason is that search speed differences get evened out, and quality of >evaluation becomes more important that it would otherwise. > >Am I right or wrong? IMO, you are wrong. You could even argue the opposite, with longer TC the engine doesn't need hand-coded eval because the search will resolve a lot of the stuff automaticly. -S. >MH > > > >>(unless you can prove variance is smaller >>at longer TC?). >> >>-S.
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.