Author: Pete Galati
Date: 13:05:01 05/17/00
Go up one level in this thread
On May 17, 2000 at 11:07:03, Fernando Villegas wrote: >Hi: >That best programs are IM strenght is not a novelty. I think I have seen many >post by Bob saying that, but in any case if a shadow of a doubt still existed, >the tournament in Holland has given another proof of it. Now IM lose >systematically to best programs or get a draw or win just once in a while. Now >an IM winning a top program IS the new. Respect GM, that's another history as >much as the "GM" expression encloses a very wide category. There are worlds of >difference between a normal GM and one of the top ten or one of the top five. >Seems like differences between players get bigger the highest his capacities. >Between a 1600 and a 1500 elo player there is not really a big gap, but between >a 2500 and a 2600 there are parsecs of distance. So the discussion should be: >which kind of GM top programs already are or are near to be? My guess: they are >or near to be member of the pack of low level GM's. Another guess: they will be >top 50 GM's class in 5 years at most. And ten top in 10 years, excepto if >something unexpected happens, like the fall of an asteroid on Earth. >Fernando I just have another question. How do we determine when a Chess program has reached a certain strength (in relationship to humans)? This has probably been gone over allot at CCC and I've ignored it. My feeling is that if a program can beat the average IM 50% of the time, then it's equal in strength to that person. AND, if a program can beat the average GM _when_ he has just acheaved becoming GM, then it's low GM strength. I doubt that it's close to this because allot of average Joes win against programs, maybe not in 50% of the games, but... So my feelings are that when programs can win 50% of the games against low GMs, then they're GM strength. Pete
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.