Author: Bruce Moreland
Date: 11:03:57 03/17/98
Go up one level in this thread
On March 17, 1998 at 13:06:04, Ed Schröder wrote: >As said, remove the double games. >Result, clean engine-engine fights. >That's what we want isn't it? If program A is equal to program B, and neither use learning, they should have the same SSDF rating, sure. The SSDF rating should reflect how the programs would do against the general community of players. It's debatable whether both programs would do the same against the general pool of players, since they are only tested against computers, but let's say for sake of argument that they would. Now assume that learning is added to A. It would do better against the general pool of players, right? Then why shouldn't its rating be higher. If anything, blame the programs that don't have learning, for being such incredible suckers by allowing repeated games. They are the ones screwing up the rating list. Imagine thwo humans, A and B, both the same strength. But there is a problem. Playing black, A lost a game to B in his favorite Sicilian Defense, long ago. This has consumed and enraged him, so now whenever he plays black against B (and he plays B often, they belong to the same club), he always plays the Sicilian, if he can, and he always loses. B has figured this out, and now always plays 1. e4 against A, and he always wins. Does B have a moral obligation to play something other than 1. e4 now and again against A? Of course not. He can play 1. e4 until they both die of old age. Does this screw up the rating pool? Of course. A has an artificially low rating against anyone other than B, and B has an artificially high rating against anyone other than A. Who is to blame for this situation? Player A is. Certainly not player B. The SSDF list should be a ranking of a pool of chess players, so let the players play the best chess they can, and if some of the players are vulnerable to silly tricks, that is bad for *everyone*, but it is the fault of those players, not those who set traps that the less complete players fall into ad nauseum. bruce
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.