Author: José de Jesús García Ruvalcaba
Date: 16:29:57 11/18/99
Go up one level in this thread
On November 18, 1999 at 19:15:48, Christophe Theron wrote: >On November 18, 1999 at 18:42:42, Charles Unruh wrote: > >>Theories on Tiger's success. No book learning, no anti-prog x play(?)? It is >>Amazing that it appears that tiger 12.0 could end up being 30 to 40 points >>higher rated than any other program, without book learning or optimizations >>against the vatious programs. I haven't got around to testing my tiger 12, but >>it's the results are good enough evidence for me. > > >Charles, I have said this several times already: > >* I don't tune my program against other commercial programs. I almost never play >Autoplayer games. I have only one autoplayer cable at home, I use it sometimes >to make sure that Auto232 still works in my DOS version. > >* I don't have many of the top programs. I don't have Nimzo, I don't have >Hiarcs, I don't have ChessMaster 6000, I don't have Shredder. When I play games >against commercial programs, it's always manual games, never more than 4 in one >day, and it's always against Genius5 and Fritz2. > >* Tiger's book learning is almost nonexistent. Tiger just avoid repeating lost >games. Tiger is unable to repeat a won or a draw game, except by pure luck. I >admit this is maybe a mistake, I might change my mind in a future version. > >* Chess Tiger has been developped by playing against human players at my chess >club here in Guadeloupe. I have always used slow computers like 286 12MHz and >386sx 20MHz. > > > > Christophe Hi Chistophe, I think you have developed a strong chess program by adhering to sound principles. Non-aggresive book learning is one of them. While repeating opening lines which have won games increases the score when playing against a single computer opponent (specially if it does not have book learning), let me state clearly that it does NOT increase overall playing strength. Avoiding losing lines is good because there are mainly to posibilities: a) It is a poor line. Do not bother playing it again. b) Your engine does not understand the resulting middle-game positions. In this case, while the position migth be objetively good, it makes sense to avoid it. It is just like when a human player avoids the positions she/he does not understand. I think most grandmasters have great technique, and also know themselves very well and try lead the game to position which suit their respective style the best. It might seem that repeating winning lines makes sense, even in "equal" positions, pretending your engine understands them quite well. But if it is playing against only one opponent, it only means that your Tiger understands those positions better than that specific player, not neccesarily "well". It does not make sense to repeat them against other player. Cristophe, you decide the future of Chess Tiger, and it is up to you to switch later to a more aggresive book learning. But I ask you, please stick to the sound principles that have led you to develop a strong and sound engine that is overall strong (and will keep hopefully getting stronger), and not specialized in outscoring single computer opponents in long autoplayed matches. José.
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.