Author: Thorsten Czub
Date: 14:07:47 12/18/01
Go up one level in this thread
On December 18, 2001 at 16:54:54, Louis Fagliano wrote: >Well, I got a suggestion as to how to make it more interesting: Make this a >full round robin of 25 rounds! Why not? :-)))) no - after 15 rounds it is over. i will maybe one day start another tournament, after i have done some other projects, but - THIS tournament will end after 15 rounds. as i said. why not more ? i began this tournament november 2000. Now it is december 2001. and it seems the tournament will be finished in january or february 2002. > It's already a 15 round swiss and >that's way too many rounds for a swiss tournament of 26 participants because >you'll eventually run out of programs (or players) with similar scores to pair >up. yes. but you will not convince me. the tournament finishes after 15 rounds. the idea is not to be precise as possible . the idea was to have 40/120 games, and have all kinds of programs play serious games. so serious that it makes sense for a human beeing to replay the games. i want to come away from this kind of fast food chess, where people let play engines against each other over night, where they produce so much data that no human beeing can follow in other way than study the results. i want to come away from LISTS with results. because this is not chess. it is statistics. my idea was to present games, and to find ways to make the event ONE big event, to FOLLOW , to make it an EVENT. Its important that the people have time enough to follow such a tournament. i want to go away from this fast food computer-chess. the main content is the game the chess program plays and what we can learn out of it. after a while, after a few rounds, a ranking exists and people begin to find out that some programs are stronger and play better than others. i think this tournament has shown this. it has shown this with quality and with ranking. >After playing 15 rounds you're 60% of the way to a full round robin and you >might as well complete the round robin if you've gotten that far. > >The reason behind this is that I've heard that the number of rounds to determine >a statistically significant winner in a swiss tournament is 2 to the power of >how many rounds you need equals the number participants you have. :-))) good that you have heard about. this makes you wiser. instead of continuing this tournament , i will maybe do something else. there is enough to do. >Thus with 26 participants, 2 to the 5th power equals 32, so five rounds was >enough to determine a winner. Any more rounds than that and you'll start >running into pairings where the opponents' scores won't match. > >If anybody has a different opinion, let me know. i do have a different opinion. don't get me wrong. if the day would have 48 hours or more, i would maybe consider different.
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.