Author: Uri Blass
Date: 18:54:17 06/18/04
Go up one level in this thread
On June 18, 2004 at 21:26:10, Ed Trice wrote: > >>But this is exactly the point we have been arguing over and over with >>programming a capablanca chess engine. > > >Incorrect. > >Selling a Capablanca Chess set is not the same as designing a program. You have >to DESCRIBE THE METHOD OF PLAY when you sell the game. > >They cannot ship with the Capa set the text "Oh, by the way, put the chancellor >on e1, th Archbishop on g1, etc...' and then let the players play the game of >Gothic Chess. > >They cannot discuss playing the game of Capablanca Chess. > >>It seems you've been saying such an engine with a position setup feature would >>be illegal, or did I misunderstand that? > >There is more to a chess engine than a setup feature. The king safety code in >Capablanca Chess is very, very different from Gothic Chess. good king safety code can be good both for capablanca chess and gothic chess. > >If the program has internal heuristics to deal with Gothic patterns, it is in >violation of the patent. The whole point of the CapaGNU Modified match was to >demonstrate that a Capa program, designed only to play Capa's chess, would get >pulverized by one with Gothic-specific algorithms. It proves nothing. CapaGNU is not the strongest possible program to play capablanca chess and it is possible that in the future some program will be able to calculate good evaluation for every game including gothic chess not based on previous specific game human knowledge. I also think that it is possible that search is underestimated and even a program with the same evaluation as CapaGnu may beat your program if it searches some plies deeper and it is possible to do it by better search techniques. Uri
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.