Author: Amir Ban
Date: 07:12:46 01/11/00
Go up one level in this thread
On January 11, 2000 at 09:14:57, Stefan Meyer-Kahlen wrote: > > >Please take a closer look at these four games, especially compare game 1+3 and >2+4. It is a very good example of a program with book learning ability playing a >program without book learning. S2 has no book learning and I am the only one who >can be blamed for that. I don't want to complain too much, but in my opinion >those matches are useless. I have only checked those 4 games but I bet you can >fing more examples in the other games. Please correct me if I am wrong. > >Don't get me wrong. I think that Junior6 is probably very strong, but has also >an excellent book learing algo, particularly against programs without any form >of learning. > >Stefan > Shredder 2 is yesterday's program run on yesterday's hardware, but it has a rating and it serves here a useful function of being a yardstick against which a new program is measured. If Shredder 2 had learning, its rating would be higher. If Junior 6 didn't have learning it's results would be worse. So it can't be right to factor out learning because J6 would get a rating that is unrealistically low, compared to the other learners, including Shredder 4. A different way to look at those matches is to note that J6 needs to win in the ratio of about 3:1 to get a decent rating. Anything less is not good enough. If you look at it this way, you realize that in fact the NEWER program has the formidable task here. The older program just needs to scrape a few points to maintain its rating. Junior 4.0 has been at the receiving end of this learning feature and has been taken to the cleaners by Fritz, Rebel and others on this account. What is new here ? Amir
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.