Author: Thorsten Czub
Date: 07:36:02 10/15/00
Go up one level in this thread
On October 15, 2000 at 08:56:16, Enrique Irazoqui wrote: >On the other hand (damn, there is so often another side to things), what does it >tell you the fact that Gambit loses to Tiger, while crushing programs with a >weak king-safety code? thats normal. i do not believe testing a chess program against its predecessors makes ANY sense at all. you will not find out that the NEW engine is stronger. e.g. christophe would not have been able to find out that gambit-tiger is stronger by letting it play against normal-tiger. only playing with the program against: a) all other programs b) humans and doing c) test positions gives you an overview. this - of course - needs time. but let version x play against x+1 definetely NOT a method to find out about the strengh of a chess program. the reason is easy: the amount of similar algorithms that fight each other is too high. it is like a human beeing who should be able to judge himself. this is also not very good possible, we need friends with DIFFERENT point of views to correlate our own opinion well. same in computerchess as in human-life. instead of analysing yourself, you should better talk with friends. > Speculative programs will be great, to some extent they >already are, but do they need further refinement, suppleness, increased >knowledge! For the moment, when I watch their games sometimes the word >"brilliant", sometimes "fireworks" come to my mind. you always talk about "speculative programs". but it is more than that. the programs shall not only have a speculative evaluation function, they shall also PLAN something within the game. it is one plan to mate the king. >Enrique
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.