Author: Roger Brown
Date: 12:10:43 08/08/02
Go up one level in this thread
On August 08, 2002 at 13:34:59, Dann Corbit wrote: >The point I was trying to make is that *any* championship that relies on a short >series of games *cannot* determine accurately who is the best. Agreed. My question is what is short? Since we are not likely to have hundreds of games to statistically determine it, best either has to leave the vocabulary of the sports fan (Brazil is the best football team this year, Kasparov is the best chessplayer alive, Shredder is (I mean Junior!!)etc)or there needs to be an appreciation of the context of a short (not statistically large) series to determine who or what is best. >I remember one year when UCLA won the college basketball championship with a >16-11 record (IIRC). They probably were not the best team, but they got hot and >won all the marbles anyway. I am going to put it this way; if an underrated team with less talent than another team beats that "better" team, then by coming to the match overconfident, out of condition etc. the "better" team proved that they lacked the qualities (mental) necessary to win. > >The Lakers *may* be the best team in US basketball. Hmmmmmm.... However, the 7 game series >did not prove superiority. It only crowned a champion. Which is fine, after >all. We don't want to wait for 1000 games to decide who is best. And if Mr. >Oneil should get hurt on game 200 what then? You don't want to even think about it! I suppose they would suddenly start to get incredibly lucky.... > >Short contests discover a winner. But they do not really demonstrate >superiority. The longer the series, the greater the confidence. No argument here Dann. I guess this come full circle to an issue I raised in another forum. I am quite a fan of the statistical approach to things. However you raise a fascinating point: precisely what is the statistical minimum number of wins/games required to confer the title of best on any person/team/performance? Computerchess I suppose affords the chance for a statistical determination of which program is best. Unlike mortal chessplayers, they do not tire, get annoyed, sick, crash and burn....well, maybe here is a similarity! Thousands of games can be played day and night to determine which program is best. Or is the entire debate pointless as are these competitions where GM's and computers square off over far fewer than ten games? >So a match between a computer and a human of 6 games will decide a winner. But >it will not show who is better. Agreed. I guess apart from giving the people a circus, this is a meaningless series...well I guess I can hope that there is an opening innovation or stroke of brilliance in a game to excite me instead of anti-computer play. :) Later.
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.