Strengths of chess computers
Obviously the biggest strength of the computer is that it never "messes up." It
sees through all the short-term tactical possibilities in a matter of seconds.
Thus, any open-board, tactical middlegame - especially invloving knights -
would be quite hazardous for a human player striving to play against the
machine. Many games against computers are lost by inane little two- and
three-move oversights.
For this reason, even though I have lost many games against computers, it
was difficult to find losses that were particularly interesting. The wins (in
addition to being more pleasant!) involve many aspects of chess beyond the
fundamental matter of precise calculation. Nevertheless, I think the two games
below are entertaining, and also illustrate interesting aspects of the
human's struggle against the computer.
Grandmasters know the pattern well; it isn't hard to use one's superior
knowledge and strategic pattern recognition to get a promising game against
a computer, but that doesn't guarantee winning the game. In practical
human-against-human chess, ugly positions tend to collapse because on every
move the defending player must steer his way clear of all manner of sinister
pitfalls. It requires great tactical vigilance to do this, and few humans can
manage it. However, computers do this very well (as long as it is possible),
and in order to transform a computer's ugly position into a truly lost one,
the utmost strategic consistency and tactical foresight is often required.
Many times the human player will fall short of these requirements, perhaps
missing a concrete win here or there, gradually losing the thread, overlooking
a surprising defensive maneuver, getting frustrated or impatient, or even
grossly miscalculating a sacrificial line. It is psychologically difficult to
handle "the final assault" against a computer, because while one would like to
find sharp, tactical ways to finish off (often, without these an attack
cannot succeed), one is reluctant to play them against a computer. There's
nothing worse for one's chess self-esteem that completely blowing a crushing
attack, but it happens all the time against programs. The following game is an
extreme example. (OK, this is definitely not one of my better games, but it
pretty vividly describes the frustrations I am describing. I will be
embarrassing the computers in another section.)
|
(1) scriabin - parrot [A67]
FICS game 15 0, May 16, 1998
1.d4 Nf6 2.c4 c5 3.d5 e6 4.Nc3 exd5 5.cxd5 d6 6.e4 g6 7.f4 Bg7 8.Bb5+ Nfd7 9.a4 Qh4+ 10.g3 Qe7? The computer thinks the queen is active here but doesn't realize how awkwardly it stands. 11.Nf3 a6 I've had some games with programs that tried 11...Bxc3+ 12.bxc3 Qxe4+ in positions like this, and the amazing thing is, they sometimes even got away with it! 12.Bd3 Nf6 13.0-0 Bh3? Black should castle and get out of dodge. 14.Re1 Nbd7 15.e5 Of course this is absolutely crushing.
15...dxe5 16.fxe5 Nh5 17.Bg5 Qf8 Only against a computer could one ever get such an overwhelming game positionally, but in blitz games one frequently misses a key point or two, loses the thread, and goes down in a crushing defeat! It's a humiliating experience. 18.e6? Very tempting, but I forgot Tarrasch's advice: when you see a good move, look for a better one! This is mistake number one. [18.Ne4! is clearer.; 18.d6 Be6 stems the tide.] 18...fxe6 19.dxe6 Ndf6 20.e7 Qg8
An unbelievable position; pretty much all I have to do is play Bc4 or put it in check somehow, but one must still be alert to possibilities and very careful with tactics! 21.Qe2 Bg4 Annoying - I figured - now Bc4 is met by ..Bxf3, but aha! we can deflect the bishop .. 22.h3? Mistake number two. [22.Bc4! Bxf3 23.Qxf3 Qxc4 24.Qxb7 wins anyway.] 22...Bxh3 23.Bc4? Game over, right? [23.Kh2! puts White back on track.] 23...Nxg3 24.Qd3 Bf5 Unbelievable. Now the game starts over anew. 25.Qd8+
Rxd8 26.exd8Q+ Kxd8 27.Bxg8 Rxg8 28.Nd5 Nge4 29.Rad1? Here I was obviously frustrated and with little time left and so many pieces hanging, another mistake can be expected. [29.Rxe4! Bxe4 A) 30.Bxf6+ Kd7! (30...Bxf6? 31.Nxf6 Rf8 32.Nxe4 Rxf3 33.Nxc5) 31.Bxg7 Bxd5 32.Rd1 Rxg7 33.Rxd5+ Kc6; B) 30.Nxf6 Bxf6 31.Bxf6+ is unclear.] 29...Nxg5 30.Nxf6+ Kc8 31.Nxg8 Nxf3+ 32.Kf2 Nxe1 33.Ne7+ Kc7 34.Nxf5 gxf5 35.Rxe1
Bxb2 36.Re7+ Kc6 37.Rxh7 The dust has settled, but unfortunately the ending is duck soup for Black. 37...b5 38.axb5+ axb5 39.Rf7 c4 40.Ke2 b4 41.Rxf5 b3 42.Kd1 c3 43.Rf2 Ba3 44.Rg2 Kb5 etc. and wins. 0-1
|
The defensive strength of the computer is particularly dangerous when material sacrifice is involved to achieve the initiative. In general I think such material sacks against computers are underrated; the computer will always take the material if it can't calculate its way to the trouble, and if successful, the gamble will pay off with interest. However, it is a double-edged policy, because sometimes if you don't play the attack just right, you just end up material down. The following game is an extremely impressive defensive performance by Fritz5. It also illustrates that computers can play "weird" positions very well; they aren't prejudiced by "principles" and can see directly to the concrete requirements of the position.
|
(2) Fritz5 (2510) - Bengtson,M (2245) [D91]
Casual game 25 12, Sep 11, 1998
1.d4 Nf6 2.c4 g6 3.Nc3 d5 4.Nf3 Bg7 5.Bg5 Ne4 6.cxd5 Nxg5 7.Nxg5 c6 8.dxc6 Nxc6 9.d5 Ne5 10.e4!? Nobody ever has ventured this interpretation, undoubtedly worrying about the dark squares. 10...0-0 Now White must do something remarkable or ..e6 will give Black great play. [10...e6? 11.Bb5+!] 11.f4!? Only a computer could have the chutzpah to play such a weakening move with the king in the center and very
little development! Probably the move isn't objectively good, but it puts Black in a very uncomfortable position for a human player: play perfectly, or be a pawn down with a cramped game. 11...Ng4 12.Rc1 Qb6 13.Qd2 Bd4 14.Nd1 Yes this is quite ugly. But, Nf3 is coming and Black must quickly try and do some damage. 14...Bd7 [14...Ne3? 15.Nf3 Bc5 16.Rxc5 Nxf1 17.Qd4 wins.; 14...e6!? 15.Nf3 Bc5 looks like the most ambitious, but Bc4 and h3 are coming and Black
is still cramped.] 15.Nf3 Bc5 16.h3 Ne3 With g3 weakened, this is now playable. 17.Ne5 [17.Rxc5 Nxf1 18.Qd4 Ng3 19.Rg1 Nxe4 20.Rc2 Qxd4 21.Nxd4=] 17...Ba4 I can't deny that I was pretty optimistic here. 18.b3 [18.Rxc5? Nxf1 19.Qc3 Ng3; 18.Nd3 Bd4 19.Nc3 Be8 keeps good compensation.] 18...Nxd1 [18...Nxf1!? 19.Rxf1 Bb4 20.Nc3 Bb5 with compensation.] 19.Nc4! I overlooked this move, which is absolutely forced, but puts a damper on Black's initiative.
[19.Kxd1? Be3 20.Nc4 Bxd2 21.Nxb6 Bxc1 22.Nxa8 Bxf4] 19...Qf6 It's frustrating how so few things work for Black. [19...Bb4 20.Nxb6 Bxd2+ 21.Kxd2+-; 19...Qb4 20.Qxb4 Bxb4+ 21.Kxd1±; 19...Nf2 20.Nxb6 Bxb6 21.Rh2 Nxe4 22.Qd3 Nc5 23.Rxc5 Bxc5 24.bxa4+-] 20.Kxd1 Be8 21.e5 Qh4 A truly bizarre situation; if Black's rooks and light bishop ever get out the game is over, but he is terribly cramped. Here all my pattern recognition is thrown off, and after all the tactics we
just came through I didn't have enough time or energy to figure out the position's eccentricities. 22.Na5!! A staggering concept. b6 23.Nb7 Of course, only a computer would be willing to put a knight on a square like this; otherwise, though, Black would get a crushing attack with ..Rd8 and ..Bc6. 23...Bf2 24.Be2 Bg3 25.Rf1 g5? It is natural to make an agressive undermining stab like this, but the weird fact is that White's pieces are much more effectively placed! 25...Bd7!
26.Rc7 Bf5 27.d6 exd6 28.Nxd6 Be6 is still a weird game with compensation for Black; not 25...f6? 26.e6 f5 27.Rc7±] 26.fxg5 Bxe5 27.Qe3± Strangely enough, the e-pawn is weak and Black's game is a mess. 27...Qd4+ 28.Qxd4 Bxd4 29.Rc7+- That damned knight on b7 ends up being the best piece on the board! 29...f6 30.Rxe7 Bg6 31.d6 fxg5 32.d7 Rxf1+ 33.Bxf1 Bf6 34.Re6 Kf7 35.Bc4 Kg7 36.Rxf6 Kxf6 37.d8Q+ Rxd8+ 38.Nxd8 Be4 39.g4 Bg2 40.Ke2 Bxh3 41.Kf3 1-0
|
Back to my chess page
Generated with ChessBase 7.0
