Author: John Merlino
Date: 15:57:18 07/12/03
In a recent game that I played on ICC with CM9000 (default personality) against Tinker (a very good amateur program that plays with a nice style), Tinker played several moves that would lead me to believe that it is too materialistic in its evaluation. Here are a few positions to test if your engine might have the same problems. In this first position, Tinker is up two pawns and is probably winning: [D]r1b1k2r/p1p2ppp/2p5/4P3/2P2qn1/3B4/1B2Q1PP/R3K2R b KQkq - 0 15 However, it plays 15...Nxh2? Before this move, CM thought it was down about 1.0; after it thought it was up about 0.4. A few moves later and now Tinker is up FOUR pawns (although the knight on h2 will soon be lost), although CM's eval only favored Black by about 1.0. [D]r3k2r/p1p2ppp/2p1b3/8/2P4q/3B2P1/1B2Q2n/R2K3R b kq - 0 18 Tinker now plays 18...Qxg3?, and the eval jumps up to about 0.5 in White's favor. Finally, this position. Material is "even", but White has a clear advantage. CM9000 scored this position about 1.7 in White's favor. [D]r3k1r1/5p1p/2p1b3/8/q1P5/p1BB1Q2/8/R1K4R b q - 0 27 Tinker plays 27...Bxc4? which loses outright. After this move, CM gives an eval of 6.5 for White. How do your engines do at avoiding these moves? Finally, just for fun, here's a position from a few moves later: [D]r1k3r1/5p1p/2p2Q2/8/2q5/p1B5/8/R1K1R3 w - - 0 31 In this position, on my P4-2.4, Chessmaster announces Mate in 14 in 0:15, and upgrades it to Mate in 13 after 0:23. jm
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.