Author: Owen Lyne
Date: 08:47:53 09/17/99
Go up one level in this thread
One additiomnal classic example - playing different versions of the same code against each other. You can get very lopsided results yet against all other opposition the two versions perform almost exactly the same. It is easy to think of lots of ways this could happen. Conversely, two programs even against each other and very different against the rest of the worldis not hard to believe either. Of course, if one set of results is totally at odds with lots of alternative evidence, it may be a good idea to look closelty, but it cannot prove there unusual results are false. And one thought on cheating/checking the moves people post of games. If I only post the games one program won, they can all be real games, perfectly duplicated by everyone else, and yet they were only 1% of the games I played, the rest were lost, but not posted. Yes, moves not being able to be duplicated can again give rise to suspicion, but we know programs aren't quite deterministic even on identical hardware and the oddest little thing can change some time allocation say, hence a key move choice from good to bad, or vice-versa, not because the computer could see that either was good or bad, but just from random oddities. Owen
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.