Author: Keith Hyams
Date: 13:56:30 11/14/05
Go up one level in this thread
On November 14, 2005 at 10:03:25, Jouni Uski wrote: >I was really surprised, when testing book bin. I played 50 games against F8 >with Fritz8.ctg and then with own book: > >1) both used Fritz8.ctg > > 12345678901234567890123456789012345678901234567890 >1 Fruit 2.2.1 01½0½½1½½½½½½11½½0½½1100½½01100½½½01½0½½01½½½½1½½1 25.5/50 >2 Fritz 8 10½1½½0½½½½½½00½½1½½0011½½10011½½½10½1½½10½½½½0½½0 24.5/50 > >2) Fruit used book.bin > > 12345678901234567890123456789012345678901234567890 >1 Fruit 2.2.1 1111½½½01½111101½½111½11½11101110110011011101100½1 35.5/50 >2 Fritz 8 0000½½½10½000010½½000½00½00010001001100100010011½0 14.5/50 > >According to this Fruit's book gives 150 rating points!? > >Jouni I assume that Fritz8.ctg is tuned to play openings that Fritz8 likes. In order to claim 150 points you would have to compare Fruit book to a neutral book. I played Fruit 2.2.1 vs Fritz8 using first Perfect9.0.ctg and then 5moves.ctg with about 100 games in each match and Fruit was slightly better with Perfect but produced spectacular results, similar to your second set with 5moves.ctg. If I had to come to a conclusion, it would be similar to the suggestion that Kurt made below. Fruit seems to have more spacial awareness than its competitors and the short opening sequence in 5moves.ctg allows the engine to exploit this property. The engine produces some remarkably human-like play. I am very glad that the bulk of the source code has been released.If it had always been closed source, I would probably have become obsessed with how he did it.
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.