8 | |||||||||
7 | |||||||||
6 | |||||||||
5 | |||||||||
4 | |||||||||
3 | |||||||||
2 | |||||||||
1 | |||||||||
a | b | c | d | e | f | g | h |
r1b2rk1/ppp2ppp/2n2n2/3q4/3P4/3BPP2/P2B2PP/R2QK1NR b KQ - 0 10
What kind of knowledge is present in a chess engine, when it chooses to sacrifice his Knight by playing ...Nxd4 ? This is the type of sacrifice which has no immediate, concrete returns. I do not do chess programming, but I'm curious as to what knowledge is needed for an engine to make such a move.
Among the engines available today,
Glaurung 1.2.1 likes ...Nxd4 and gives a small advantage for Black.
Rybka 2.3.2a likes the move too, as does Fritz 11. Deep Junior 8 likes the move as well. So does Gandalf 6.
Old engines like Junior 6, Fritz 6 and HIARCS 7.32 do not like it at all. Surprisingly even Gambit Tiger 2.0 does not want to play it.
This type of move was unthinkable in the previous decade, but now more and more engines seem to be 'catching up/maturing'. Can these types of moves be used as benchmarks for 'chess understanding'?
PS
Black often needs to play à la Morozevich and sharpen the game, relying upon an initiative to offset his material sacrifice. Here is Rychagov – Morozevich, Moscow 2001 (blitz): 1.d4 d5 2.c4 Nc6 3.e3 e5 4.cxd5 Qxd5 5.Nc3 Bb4 6.Bd2 Bxc3 7.bxc3 Nf6 8.f3 0–0 9.Bd3 exd4 10.cxd4 (note Black’s thematic exchange of his Bishops for White’s Knights, White’s large pawn center, and Black’s delay of …exd4)