Author: Dann Corbit
Date: 13:26:22 07/11/02
Go up one level in this thread
On July 11, 2002 at 15:23:58, Julian Morley wrote: > What exactly do these results prove? > > > Program Elo + - Games Score Av.Op. > > 1 Deep Junior 7 : 2959 197 277 11 81.8 % 2698 > 2 Shredder : 2882 211 181 11 72.7 % 2711 > 3 Brutus V099.007 : 2872 245 272 9 77.8 % 2654 > 4 Quest (Deep Fritz 7) : 2757 266 169 9 66.7 % 2637 > 5 Diep : 2651 291 165 9 55.6 % 2612 > 6 Warp : 2640 291 229 9 55.6 % 2602 > 7 IsiChess X : 2595 229 229 9 50.0 % 2595 > 8 Sjeng 12.14 : 2537 229 291 9 44.4 % 2576 > 9 Ikarus : 2525 178 178 9 50.0 % 2525 > 10 ParSOS : 2509 280 280 9 50.0 % 2509 > 11 Goliath : 2428 280 280 9 50.0 % 2428 > 12 Chinito 2.1 : 2387 280 280 9 50.0 % 2387 > 13 XiniX : 2383 229 291 9 44.4 % 2422 > 14 Insomniac : 2314 165 291 9 44.4 % 2352 > 15 SpiderChess : 2255 201 278 9 38.9 % 2333 > 16 Postmodernist : 2219 169 266 9 33.3 % 2339 > 17 NoonianChess : 2127 272 245 9 22.2 % 2344 > 18 Sharky : 1774 0 0 9 0.0 % 2374 > > My, look at those error bars. > > Jules They don't prove anything. It's a championship. Like *any* sporting event, there are a finite number of games allowed. Since the top ranking opponents are fairly close in ability, we give a championship to whoever performs best for the event. We'll never know who is strongest anyway, so this method is probably as good as any to make a decision. Besides, the enourmous error bars make for good banter about 'what if' and the like. It's just as valid (or invalid) as the NBA championship, the World Cup, or any other recognized result for a crown.
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.