Author: Stephen A. Boak
Date: 12:21:21 08/26/00
Go up one level in this thread
On August 26, 2000 at 04:15:38, Ricardo Gibert wrote: >On August 26, 2000 at 04:05:23, Uri Blass wrote: > >>On August 26, 2000 at 02:16:24, Ricardo Gibert wrote: >> >>>On August 26, 2000 at 02:11:13, Ricardo Gibert wrote: >>> >>>>On August 25, 2000 at 23:54:08, Jorge Pichard wrote: >>>> >>>>>On August 25, 2000 at 22:05:19, Ricardo Gibert wrote: >>>>> >>>>>> PROGRAM CPU MHZ 1 2 3 4 5 6 7 8 9 >>>>>>------------------------------------------------------------- >>>>>> 1 Shredder K7 1000 *1 *2 *2½ *3 *3½ *4 *5 *6 *7 >>>>>> 2 Fritz PIII 1000 *1 *2 *2½ *3 *3½ *4 *5 5½ 6½ >>>>>> 3 Chess Tiger PIII 800 *1 1 1½ 2½ 3 3½ 4½ 5½ 6 >>>>>> 4 Rebel PIII 800 *1 *2 2 2½ 3 3 4 5 6 >>>>>> 5 Junior PIII 700 ½ 1½ *2½ *3 *3½ *4 4 5 5½ >>>>>> 6 SOS PIII 667 ½ 1 1 2 3 *4 4 4½ 5½ >>>>>> 7 Nimzo PIII 1000 *1 1 2 *3 *3½ *4 *5 5 5 >>>>>> 8 Incomniac PIII 933 0 ½ 1½ 1½ 2 3 4 4 5 >>>>>> 9 Zchess K7 800 0 1 1 2 3 3½ 3½ 4½ 4½ >>>>>>10 Diep PIII 800 0 0 1 1 1½ 2½ 2½ 3½ 4 >>>>>>11 Crafty Alpha 500 *1 1 2 2 2½ 2½ 3½ 3½ 3½ >>>>>>12 Francesca Cel 600 0 1 1½ 1½ 2 3 3 3 3½ >>>>>>13 XiniX Cel 500 0 0 0 1 1 1 1 1 1 >>>>>>14 Pacque Cel 333 0 0 0 0 0 0 0 0 0 >>>>>> >>>>>>* 1st or tied for 1st >>>>>> >>>> >>>>I think all the computer programs can be accomodated. The problem I think is 9 >>>>rounds is overkill in a swiss when there are only 14 participants. If they had >>>>played only 5 rounds, the top programs would probably have never been paired >>>>with the bottom 2 finishers and the pairing problems you have alluded to would >>>>never have occurred. If you look at the pairings actually made, none of the top >>>>6 were paired with the bottom 2 until the 7th round. >>> >>>Oops! make that the top 4 rather than 6. >> >>Wrong >>Tiger played against the weakest program in the first round. >> >>Uri > >You are correct. That one slipped past me...which kinda kills the point I was >trying to make. Curses! ;-) Unless you play a round robin tournament, or restrict entry to only top programs of approx equal rating (controversial at best!), or have a runaway winner, there will *always* be pairing differences that likely cause debate in analyzing for any particular pair programs which one performed relatively the best. Merely keeping the top 4-6 programs from ever playing the bottom-worst 2 programs doesn't mean the number and quality of played 'intermediate' programs (depending on individual pairing schedules) will not also lead to controversy in analyzing the results of the top several programs relative to each other. Better is to come up with some system that allows some sort of 'normalization' in order to compare the normative figure of each individual program with each other individual program. An ELO-based system can do this (see my prior post), subject to having 'well-established' starting ratings for the participants. And we are not talking about proving absolutely which particular program is best of any two selected programs under comparison--certainly not after a relatively short tournament! We are talking about determining which program performed best in the tournament under discussion--which is suggestive of actual program strength (some evidence) but is not normally enough in the way of evidence to draw the 'absolute' conclusion with any reasonable amount of confidence. --Steve
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.