Peter Heine Nielsen (2638) - ChessBrain
ChessBrain World Record Attempt (Copenhagen Denmark, Symbion Science Park.), 2004
[GM Peter Heine Nielsen]

Hosted by the Danish UNIX Users Group (DKUUG) 1.d4 g6 I doubt this the best choice for a computer. The risk of ending up in closed strategic positions is huge. If one considers the last matches with Fritz and Junior the basically went for some lines in the Slav/meran leading to lively piece-play. 2.c4 Bg7 3.e4 d6 4.Nc3 Nf6 5.Nf3 O-O 6.Be2 e5 7.O-O a5 This surprised me. It`s a little known variation favored by now German Grandmaster Glek. Apparently its mentioned in ChessBrain`s opening book, but only with a 4% chance of being played.Normally I would go for 7...Nc6 8 d5 Ne7 which is the main line. I intended 9 Kh1!? not maybe the best move, but a late favorite of my Compatriot Bent Larsen. It would have thrown ChessBrain out of its book, probably answering 9...b6? according to the programmers. This shows why computers shouldn`t go for closed positions. Obviously black should prepare f5 in such positions. Not weaken his queenside. 8.Re1 An excellent move, still however the clever choice was 8 d5! Closing the position. I however was curious, and decided to play the move I Thought was best, not going for anti computer strategies. 8...exd4 Not giving me the chance to change my mind. Actually the move is dubious, the structure now arising is definitely better for white, but a though and patient defender might do ok as black. ChessBrain is certainly such. 9.Nxd4 Bd7 10.Bg5 Nc6 10...h6 11 Be3 followed by Qd2 would just help white. 11.Nxc6 Bxc6 12.f3 Qd7 13.Qd2 Rfe8 14.Rac1 White is for sure better here. Blacks position is quite without any prospect, the only advantage being its solidity. However Computers can do well in such. A human would probably try and get some counter play preparing f5, here it would just make whites task much easier. ChessBrain does well in the coming phase. [14.Nd5!? Might have been the right moment. I wasn’t sure I had enough after: 14...Nxd5 15.cxd5 Bb5 Though. Here f5 is actually a relevant way of getting counter play.] 14...h5 I like this move. It weakens a bit yes, but still nothing fatal. Nh7 is now possible now and then, and a white Nd5 can be answered with Nh7 exactly. 15.Kh1 Nh7 16.Bh6 Bxh6 17.Qxh6 Re5 18.Nd5 [18.f4 Re6 19.f5 gxf5 20.Qxh5 Rg6 Is pretty unclear. I might be better in such concrete lines, but it`s not what you want against the machine.] 18...Rae8 19.Qd2 b6 20.Bd3 Qd8 21.Rf1 Signaling my intensions to run forward the f-pawn. Here in general I was hoping ChessBrain would take on d5. Then having taken back with the c-pawn I would generate huge pressure against c7, being clearly better in a dry position. Actually it might have been a mistake placing the rook on c1 if that`s been my hope. Then it becomes less attractive for black to take on d5. Such tricks might work against a machine. 21...Nf6 22.b3 Bb7 23.Qc2 Nd7 As usual I think white has a stable edge, but its not getting bigger. Normally it needs a bit of cooperation to win such positions, but typically a player who ended up like this, as black will continue making mistakes! ChessBrain just sits tight, the clever strategy, so difficult for humans though. 24.f4 Not really the way to go, but again curiosity played a role. Playing a3+b4 was sensible and safe. 24...R5e6 25.e5 c6 A cold shower, still it was too naive hoping for 25...dxe5 26 f5! Especially as white gets his material back more or less immediately. 26.f5 Realizing this leads to a draw, however I felt I passed the point of no return. Interesting was 26 Nc3 where I feared 26...Nc5! [26.Nc3 Nc5 what I missed was 27 Ne4 Nxe4 28 Bxe4 f5! this rules out my f5 pushes and next it will just take on e5. 27 Rcd1!? is suggested by many, its a good move, but still I find the position unclear. 27...f5!? being an good answer. In general its possible white can find an edge here, but playing a position opening up like this, is playing against odds with a machine.] 26...gxf5 27.Bxf5 cxd5 28.Bxe6 Rxe6 29.Rxf7 Sweet, unfortunately only giving me a draw. 29...Kxf7 30.Qh7+ Ke8 31.Qxh5+ Ke7 32.Qg5+ Ke8 33.Qh5+ Ke7 34.Qh7+ Here I offered a draw. Normally I would have repeated once more, making it a threefold repetition, however ChessBrain was low on time, and this was the correct result, instead of some nonsense win on time in a dead drawn position. As usual the computer both impresses as well as plays some awful moves. Strategically its pretty weak, but in defense and calculations is horribly strong. This mixture means humans versus machines to my mind always will be a tense and rather equal struggle. I for one at least enjoyed being part of one. 1/2-1/2