LC0 bad search and poor tactics

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

yanquis1972
Posts: 1766
Joined: Wed Jun 03, 2009 12:14 am

Re: LC0 bad search and poor tactics

Post by yanquis1972 »

t30 seems to have issues with basic QP opening positions, & overly optimistic pawn grabs at b2/g7...in a 20 game match at 10'+10s this was the only game lost https://lichess.org/study/CpmlY6g4/VtKMT2ow#0

absolutely smashed with no effort. what's more the slav was at one point a strength for t30, but i've seen these patterns repeatedly in bullet games, really didn't expect to see it at a much longer TC.

attached zipped pgn for anyone interested; leela's win was again spectacular.
yanquis1972
Posts: 1766
Joined: Wed Jun 03, 2009 12:14 am

Re: LC0 bad search and poor tactics

Post by yanquis1972 »

another example, probably lost before move 8 --

[pgn][Event "Lc0 20 start"] [Site "PC"] [Date "2019.01.29"] [Round "4"] [White "SFDEV"] [Black "Lc0 T30 id32952"] [Result "*"] [BlackElo "3000"] [ECO "D10"] [Opening "Slav"] [Time "09:53:05"] [Variation "Exchange, 4.Bf4"] [WhiteElo "3000"] [TimeControl "600+10"] [Termination "unterminated"] [PlyCount "59"] [WhiteType "program"] [BlackType "program"] 1. d4 d5 2. c4 c6 3. cxd5 (cxd5 cxd5 Bf4 Nc6 e3 Nf6 Bd3 Bg4 Qb3 Na5 Qa4 Bd7 Qd1 Nc4 Bxc4 dxc4 Nf3 e6 O-O Be7 Nbd2 Rc8 Ne5 b5 a4 O-O axb5 Bxb5 Rxa7 c3 bxc3 Rxc3 Rb7 Bxf1 Qxf1) {+0.61/30 45} cxd5 (cxd5 Nc3 Nf6 Nf3 Nc6 Bf4 Bf5 e3 e6 Qb3 Bb4 Bb5 Bxc3 Qxc3 O-O Bxc6 Ne4 Qa3 bxc6 Ne5 g5 Bg3 h5 f3 Nxg3 hxg3 f6) {+44.49/13 9} 4. Bf4 (Bf4 Nc6 e3 Nf6 Nc3 a6 Bd3 Bg4 Nge2 e6 O-O Bh5 Rc1 Bd6 Bxd6 Qxd6 Na4 O-O Qb3 Rab8 Nf4 Nxd4 Qb6 Nc6 Bxa6 bxa6 Qxc6 Qxc6 Rxc6 Rfc8 Rxa6) {+0.55/29 8} Qb6 (Qb6 Nc3 Nf6 e3 Qxb2 Bb5 Bd7 Bxd7 Nbxd7 Nge2 Qa3 Rb1 b6 O-O a6 Bc7 e6 Bxb6 Nxb6 Rxb6 Be7 Qa4 Qxa4 Nxa4 O-O Rb7 Bd6 Rfb1 Rac8 R7b6 Ba3 R6b3 Bd6 Rb6 Ba3 R6b3 Bd6) {+45.43/17 16} 5. Nc3 (Nc3 Nf6 e3 Qxb2 Bb5 Bd7 Bxd7 Nbxd7 Nge2 Qa3 Rb1 b6 f3 a6 e4 e6 O-O Rc8 Rc1 Rc4 exd5 Be7 dxe6 fxe6 Ne4 b5 Qb3 Nxe4 fxe4 O-O Rxc4 Qxb3 axb3 bxc4 bxc4 e5 Bxe5 Nxe5 Rxf8 Kxf8 dxe5 Bc5 Kh1 Ke7) {+0.89/28 11} Nf6 (Nf6 e3 Qxb2 Bb5 Bd7 Bxd7 Nbxd7 Nge2 Qa3 Rb1 b6 O-O a6 Bc7 e6 Bxb6 Nxb6 Rxb6 Be7 Qa4 Qxa4 Nxa4 O-O Rb7 Bd6 Rfb1 Rac8 R7b6 Ba3 R6b3 Bd6 Rb6 Ba3 R6b3 Bd6) {+43.56/17 12} 6. e3 (e3 Bg4 f3 Bd7 Bd3 e6 a3 Nc6 Nge2 Nh5 Na4 Qd8 Bg3 Nxg3 hxg3 g6 Kf2 Ne7 b4 b6 g4 Bg7 Nac3 O-O Qg1 h6 Qh2 Rc8 Nb5 Bxb5) {+0.96/29 11} Qxb2 (Qxb2 Bb5 Nc6 Nge2 Bf5 O-O Rc8 a3 Ne4 Bd3 Nxc3 Nxc3 Qxc3 Bxf5 e6 Bd3 Be7 Rb1 O-O Rxb7 Qxa3 g3 Nb4 Bb5 Na2 Qg4 Bf6 Bd7 Rc1 Bxe6 Qa6 Rxc1 Nxc1) {+43.74/21 12} 7. Bb5+ (Bb5 Nbd7 Nge2 e6 O-O Qa3 Rb1 Qa5 Bd3 a6 Qc2 Nh5 Be5 Nhf6 Bg3 Nh5 Bh4 Nhf6 e4 dxe4 Nxe4 Be7 Rfc1 h6 Bg3 O-O Bd6 Qd8 Bc7) {+1.21/29 18} Nc6 (Nc6 Nge2 Bf5 O-O Rc8 a3 Ne4 Bd3 Nxc3 Nxc3 Qxc3 Bxf5 e6 Bd3 Be7 Rb1 O-O Rxb7 Qxa3 g3 h6 h4 Rfd8 Kg2 Bd6 Bxh6 gxh6 Qg4 Kf8 Qh5 Rc7 Qxh6 Ke7 Qg5 Kd7) {+42.97/22 26} 8. Nge2 (Nge2 Bf5 O-O a6 Ba4 b5 g4 Bxg4 f3 Bf5 e4 e5 dxe5 Nh5 Be3 Rd8 exf5 d4 Bg5 Be7 Nxb5 axb5 Rb1 Qa3 Bxb5 O-O f6 gxf6 exf6 Bxf6 Qd2 d3 Bxf6 dxe2 Qxe2 Nxf6 Bxc6) {+1.46/27 9} Bf5 (Bf5 O-O Rc8 a3 Ne4 Bd3 Nxc3 Nxc3 Qxc3 Bxf5 e6 Bd3 Be7 Rb1 O-O Rxb7 Qxa3 g3 h6 h4 Rfd8 Kg2 Bd6 Bxh6 gxh6 Qg4 Kf8 Qh5 Rc7 Qxh6 Ke7 Qg5 Kd7) {+43.06/22 0} 9. O-O (O-O Rc8 f3 a6 Bxc6 bxc6 e4 dxe4 fxe4 Nxe4 Na4 Qb4 Rb1 Nf2 Rxb4 Nxd1 Rb7 Bd3 Rxd1 Bc2 Nb2 Bxd1 Nxd1 f6 Be3 h5 Nb2 g5 Nc4 h4 Ra7 h3 g3 Rb8) {+1.52/28 11} Rc8 (Rc8 a3 Ne4 Bd3 Nxc3 Nxc3 Qxc3 Bxf5 e6 Bd3 Be7 Qb1 Nd8 Ra2 Qa5 Rc2 Rxc2 Qxc2 h6 Rb1 Qxa3 Bb5 Kf8 Qc7 Kg8 Be5 Kh7 Qc2 Kg8 g3 Bf6 Bc7 g6 h4 h5 Kg2 Kg7 Bd7 Qe7) {+43.27/23 18} 10. f3 (f3 a6 Ba4 Qb6 e4 dxe4 fxe4 Nxe4 Nxe4 Bxe4 Nc3 Bf5 Kh1 Qa5 Bd2 e6 d5 Bb4 dxc6 bxc6 Bc2 O-O Bxf5 exf5 Rc1 Rfe8 a3 Bf8 Qf3 Rcd8 Bg5 Be7 Bxe7 Rxe7 Qxf5 Qxf5 Rxf5) {+1.59/29 12} a6 (a6 Ba4 Qb6 e4 Be6 Bg5 Qd8 Nf4 b5 Bb3 Qb6 Nxe6 fxe6 Ne2 Na5 e5 Nd7 Kh1 g6 Qd3 Nb8 f4 Kd7 Qh3 Nxb3 axb3 Nc6 f5 gxf5 Nf4 Nb4 Nxe6 Qxe6) {+32.84/17 40} 11. Ba4 (Ba4 Qb6 e4 dxe4 fxe4 Bg4 e5 Nd7 e6 fxe6 h3 Bh5 Be3 Nf6 Qe1 Bf7 Nf4 Qa5 Nd3 g6 d5 Nxd5 Nxd5 Qxe1 Raxe1 exd5 Ne5 Bg7 Nxf7 Rf8 Ng5 Rxf1 Rxf1 d4 Bf4 Bh6 Bb3 d3) {+1.79/30 10} Qb6 (Qb6 e4 Be6 Bg5 Qd8 Nf4 b5 Bb3 Qb6 Nxe6 fxe6 Ne2 Na5 e5 Nd7 f4 g6 Kh1 Nb8 g4 Rg8 Bh4 Nbc6 f5 Nc4 Qd3 Bh6 fxe6 Nd8 g5 Bg7 Nf4 Nxe6) {+32.09/20 13} 12. e4 (e4 dxe4 fxe4 Bg4 e5 Nd7 e6 fxe6 h3 Bh5 Be3 Nf6 Qe1 Bf7 Nf4 Qc7 Nd3 b5 Bb3 Qd6 Rc1 Bg6 Nc5 e5 dxe5 Qxe5 Qf2 Na5 Ne2 Nxb3 axb3 e6 Nf4) {+1.69/29 13} Be6 (Be6 Bg5 Qd8 Nf4 b5 Bb3 Qb6 Nxe6 fxe6 Ne2 Na5 e5 Nd7 f4 g6 Kh1 Nb8 Qd3 Nxb3 axb3 Nc6 Rac1 Kd7 Qh3 Bg7 Rc5 Nb4 Rfc1 Rxc5 dxc5 Qb7 Nd4) {+31.92/19 15} 13. Rc1 (Rc1 h5 Bb3 Qd8 Be3 h4 Nf4 Bd7 e5 Ng8 Ncxd5 e6 Nc3 Ba3 Rb1 Nge7 d5 Nf5 Bf2 Nxe5 dxe6 fxe6 Nxe6 Bxe6 Bxe6 Rxc3 Qxd8 Kxd8 Bxf5 Rf8 Bxh4 Ke8) {+1.96/29 20} g6 (g6 Bb3 Rd8 Na4 Qa7 Nc5 Bg7 Kh1 O-O e5 Nd7 Nxe6 fxe6 g3 Nb6 Rc3 Rc8 h4 Nb4 Rxc8 Rxc8 h5 a5 hxg6 hxg6 Qb1 Qa6 Qxg6 Qxe2) {+28.29/14 52} 14. Kh1 (Kh1 Qd8 Qb3 Qd7 e5 Ng8 Bxc6 Rxc6 Na4 Bh6 Nc5 Qc8 g4 Bxf4 Nxf4 Bd7 Rb1 b5 Qe3 e6 Nfd3 h5 Qf4 Kf8 Kg1 Kg7 Rbc1 Qd8 Nb4 Rb6 Nbxa6) {+2.06/29 11} Bg7 (Bg7 Bb3 Qd8 Bg5 Na5 e5 Nd7 Bxd5 Bxd5 Nxd5 Rxc1 Qxc1 Nc6 Rd1 O-O Qb2 b5 Qa3 f6 Be3 fxe5 Qxa6 Nxd4 Nxd4 exd4 Bxd4 Bxd4 Rxd4 e5 Rd1 e4 Qe6 Rf7 Qxe4) {+30.09/15 27} 15. Be5 (Be5 Qa5 exd5 Bxd5 Nxd5 Qxd5 Nf4 Qa5 d5 O-O Bxf6 Bxf6 dxc6 e6 Nd3 Rfd8 Qc2 bxc6 Bb3 c5 Bc4 Rc6 f4 Qc7 Qe2 Rcd6 Ne5 Bg7 Qf3 h5 Rb1 Rb6 Rxb6 Qxb6 Qe2 Rd4 Bxa6) {+2.52/30 11} Qa5 (Qa5 Nxd5 Bxd5 exd5 Qxd5 Rc5 Qd8 d5 b5 dxc6 bxa4 Qxd8 Rxd8 Nc3 O-O c7 Rc8 Rd1 Ne8 Nd5 Bxe5 Nxe7 Kg7 Nxc8 Bxc7 Rc6 a5) {+16.96/15 37} 16. exd5 (exd5 Bxd5 Nxd5 Qxd5 Rc5 Qxa2 d5 O-O dxc6 Rcd8 Nd4 bxc6 Bb3 Qb2 Rc2 Qa3 Qa1 Qxa1 Rxa1 Nd7 Bxg7 Kxg7 Nxc6 Rc8 Kg1 e6 Rxa6 Nb8 Nxb8 Rxc2 Bxc2 Rxb8 Kf2 Rb2 Rc6 g5 Ke2 h5 Kd3 Kf6 h3 Rb5 Rc7 Rd5 Ke2) {+3.29/34 25} Bxd5 (Bxd5 Nxd5 Qxd5 Rc5 Qxa2 d5 O-O dxc6 bxc6 Bb3 Qa3 Rc4 Rfd8 Qc2 Qa5 Bc3 Qb5 Rb1 Nd5 Bxg7 Ne3 Qc3 Nxc4 Bh6 Qe5 Qxc4 e6 h3) {+15.68/15 12} 17. Nxd5 (Nxd5 Qxd5 Rc5 Qxa2 d5 O-O dxc6 bxc6 Bb3 Qa3 Qc2 Nd7 Rc3 Qa5 Bxg7 Kxg7 f4 c5 Bc4 Nb6 f5 Nxc4 Rxc4 Rcd8 Qe4 Qd2 fxg6 Qd5 gxh7 Qxe4 Rxe4 Kxh7 Rxe7 Rde8 Rexf7 Rxf7 Rxf7 Kh6 Ng1 Rc8 Rf3 c4 Rc3 Kg5 Nf3 Kf6 g3 a5 Kg2 a4 Kh3) {+3.33/32 10} Qxd5 (Qxd5 Rc5 Qxa2 d5 O-O dxc6 bxc6 Bb3 Qa3 Qc2 Rfd8 Nd4 Nd7 Rxc6 Nc5 Bxf7 Kxf7 Qc4 Kf8 Rxc5 Qxc5 Ne6 Kg8 Nxc5 Kh8) {+14.24/15 8} 18. Rc5 (Rc5 Qxa2 d5 O-O dxc6 bxc6 Bb3 Qa3 Qc2 e6 Ra1 Qb4 Nd4 Nd7 Rc4 Qe7 Bxg7 Kxg7 Nxc6 Qf6 Rd1 Nb6 Rc5 h5 Qa2 Rfe8 Kg1 Nd5 Bxd5 exd5 Qxd5 Re2 Kf1 Rce8 Qd4 Qxd4 Nxd4 Ra2 Rc7 Kg8 Nc2 Rb8 Ke2 Rbb2 Rd2 Kg7 Kd3 Rb3 Ke4 Rbb2 Kd4 Rb5 Kc3 Rab2 h3 R2b3 Kc4) {+2.99/36 25} Qxa2 (Qxa2 d5 O-O dxc6 bxc6 Bb3 Qa3 Qc2 Nd7 Rc3 Qa5 Bxg7 Kxg7 Re3 e6 Ng3 Nf6 Ne4 Qe5 Rfe1 Nd5 R3e2 Rfd8 Nc5 Qc3 h3 Rb8 Rc1 Qxc2 Nd7) {+13.41/16 18} 19. d5 (d5 O-O dxc6 bxc6 Bb3 Qa3 Qc2 e6 Ra1 Qb4 Nd4 Nd7 Rc4 Qe7 Bxg7 Kxg7 Nxc6 Qf6 Rxa6 Ne5 Nxe5 Qxe5 h3 Rxc4 Qxc4 Rd8 Rc6 Rb8 Qc3 Qxc3 Rxc3 Rb4 Bc4 g5 Kh2 h5 h4 gxh4 Kh3 Rb2 Bd3 Rd2 Bb5 Rb2 Be8 Kf8 Ba4 Rb1 Kh2 Rb4 Bc2 Rd4 Rc7 Kg7 Kh3 Kf8) {+2.99/36 14} O-O (O-O dxc6 bxc6 Bb3 Qa3 Qc2 Nd7 Rc3 Qa5 Bxg7 Kxg7 Re3 e6 Ng3 Nf6 Ne4 Qe5 Rfe1 Nd5 R3e2 Rfd8 Nc5 Qc3 h3 Rb8 Rc1 Qxc2 Rexc2) {+14.00/16 11} 20. dxc6 (dxc6 bxc6 Nd4 Nd5 Nxc6 Rxc6 Rxd5 Rc1 Qxc1 Qxd5 Bg3 Qa5 Bd7 Qc3 Qxc3 Bxc3 Rd1 Kg7 Kg1 e5 Rb1 f5 Kf1 Rf7 Ba4 Re7 Ke2 e4 Bd6 Re6 Rb7 Kf6 Bc5 exf3 Kxf3 a5 Bb5 Re5) {+2.96/35 13} bxc6 (bxc6 Bb3 Qa3 Qc2 Nd7 Rc3 Qa5 Bxg7 Kxg7 Re3 e6 Ng3 Nf6 Ne4 Qe5 Rfe1 Nd5 R3e2 Rfd8 Nc5 Qc3 h3 Rb8 Rc1 Qxc2 Rexc2) {+14.26/14 10} 21. Nd4 (Nd4 Nd5 Nxc6 Rxc6 Rxd5 Rb6 Bxg7 Kxg7 Qd4 Kg8 Qxb6 Qxd5 Qxa6 Rd8 Qe2 Qg5 Rd1 Rxd1 Bxd1 h5 g3 h4 Kg2 hxg3 hxg3 Qd5 Qc2 e6 Be2 Qd4 Qc4 Qxc4 Bxc4 Kf8 Kf2 Ke7 Ke3 Kd6 Kd4 g5 Bb3 Ke7) {+2.90/35 25} Nd7 (Nd7 Bb3 Qa3 Rc3 Qb4 Nc2 Qa5 Bxg7 Kxg7 Qa1 Qxa1 Rxa1 Nf6 Kg1 Rfd8 Rxa6 Rd1 Kf2 Rcd8 Re3 R8d2 Re2 e6 Rxc6 Nh5) {+12.93/14 43} 22. Bb3 (Bb3 Qa3 Rc3 Qb4 Rc4 Qb6 Bxg7 Kxg7 Re1 Nf6 Rxe7 Kg8 Ba4 c5 Nb3 Nd5 Re2 Rfd8 Qg1 Qa7 Nxc5 Nb6 Rce4 Nxa4 Nxa4 Qxg1 Kxg1 Rc1 Kf2 Kg7 Re5 Rdc8 Nb6 R8c2 Rxc2 Rxc2 Kg3 a5 Rxa5) {+3.29/33 15} Qa3 (Qa3 Rc3 Qb4 Nc2 Qa5 Bxg7 Kxg7 Qa1 Qxa1 Rxa1 Nf6 Kg1 Rfd8 Rxa6 Rd1 Kf2 Rd2 Kf1 Rb8 Raxc6 Rd1 Ke2 Rb1 Bc4) {+12.69/15 11} 23. Rc3 (Rc3 Qb4 Rc4 Qb7 Bxg7 Kxg7 Ba4 Kg8 Nxc6 Nb8 Qb3 Qc7 Rfc1 Nxc6 Bxc6 Qa7 h3 Rfd8 Qb2 Rd3 Qe5 Rcd8 Kh2 Qb8 Qxb8 Rxb8 Be4 Rd2 R1c2 Rxc2 Rxc2 Kg7 Rc7 e6 Ra7 Rb6 Kg3 g5 h4 gxh4 Kxh4 Rb2 Kg3 Ra2 Bd3) {+3.24/35 16} Qb4 (Qb4 Nc2 Qa5 Bxg7 Kxg7 Qa1 Qxa1 Rxa1 Nf6 Kg1 Rfd8 Rxa6 Rd1 Kf2 Rd2 Kf1 Rb8 Raxc6 e6 Rc7 Nd5 Bxd5 Rb1 Ne1 exd5 Rc1) {+12.73/13 20} 24. Rc4 (Rc4 Qb7 Bxg7 Kxg7 Ba4 Kg8 Nxc6 Nb8 Qb3 Qc7 Rfc1 Nxc6 Bxc6 Qa7 h3 Rfd8 Qb2 Qe3 Qa1 Rd2 f4 Rcd8 Re4 Qa7 Ra4 Qf2 Rf1 Qxf1 Qxf1 Rd1 Qxd1 Rxd1 Kh2 Kg7 Bb7 Rd6 Bxa6 h6 Kg3 g5 fxg5 hxg5 Be2 e6 Ra7 Kg6 Kf3 f5 Ke3) {+3.09/36 24} Qb7 (Qb7 Bxg7 Kxg7 Ba4 Nb6 Qa1 Kg8 Bxc6 Qc7 Rcc1 Qd6 Be4 Rxc1 Rxc1 Nd7 Nc6 Nf6 Qe5 Qxe5 Nxe5 Nxe4 fxe4 Rb8 Kg1 Rb4 Rc4 a5 Rxb4 axb4) {+11.35/13 58} 25. Bxg7 (Bxg7 Kxg7 Ba4 Kg8 Nxc6 Nb8 Qb3 Qc7 h3 Nxc6 Bxc6 Qe5 Qa3 Rfd8 Qa1 Qe2 Re4 Qd2 Qxa6 Qc3 Rc4 Qe5 Qa1 Qxa1 Rxa1 Rd6 Ra6 g5 Kh2 Kg7 g4 e6 Kg3 f5 h4 gxh4 Kxh4 Rc7 Bb5 Rxc4 Rxd6 Rb4 Be2 Kf6 Rc6 Rd4 Bc4 fxg4 fxg4 Re4) {+3.27/37 11} Kxg7 (Kxg7 Ba4 Nb6 Qa1 Kg8 Bxc6 Qc7 Rcc1 Qd6 Be4 Rxc1 Rxc1 Nd7 Nc6 Nf6 Qe5 Qxe5 Nxe5 Nxe4 fxe4 Rb8 Kg1 Rb4 Rc4 a5 Rxb4 axb4) {+11.47/14 0} 26. Ba4 (Ba4 Kg8 Nxc6 Nb8 Qb3 Qc7 Rfc1 Nxc6 Bxc6 Qd6 Qa4 Rfd8 h3 Qf6 Qa1 Rb8 Qxa6 Rd2 Qa3 Rbd8 Qc3 R8d4 Rxd4 Qxd4 Qxd4 Rxd4 g3 Rd2 Ra1 Rc2 Be4 Rd2 f4 f5 Bf3 Kf7 Ra6 h6 Kg1 e6 h4 Kf6 Rc6 g5 fxg5 hxg5 hxg5 Kxg5) {+3.22/39 20} Nb6 (Nb6 Qa1 Kg8 Bxc6 Qc7 Rcc1 Qd6 Be4 Rxc1 Rxc1 Nd7 Nc6 Nf6 Qe5 Qxe5 Nxe5 Nxe4 fxe4 Rb8 Nc6 Rb7 Kg1 Rc7 Rc4 Kg7 Na5 Rxc4) {+10.78/13 14} 27. Qa1 (Qa1 Kg8 Rc5 Qb8 Nxc6 Qd6 Rc2 Nd5 Rd1 e6 Ne5 Rxc2 Bxc2 Qc7 Bb3 Qc3 Bxd5 Qxa1 Rxa1 exd5 Rxa6 Re8 f4 g5 g3 gxf4 gxf4 Re6 Ra8 Kg7 Kg2 Re7 Rd8 f6 Ng4 h5 Nf2 d4 Rxd4 Re3 Rd8 Kg6) {+3.35/31 17} Kg8 (Kg8 Bxc6 Qc7 Rcc1 Qd6 Be4 Rxc1 Rxc1 Nd7 Nc6 Nf6 Rd1 Qe6 Qe5 Qxe5 Nxe5 Nxe4 fxe4 Rc8 Ra1 Rc5 Nf3 a5 Kg1 f6 Kf2 Rc2 Kg3 Rc5) {+9.97/13 26} 28. Rc5 (Rc5 Qb8 Nxc6 Qd6 Rc2 Rfe8 Rfc1 Qf6 Bb3 Qxa1 Rxa1 e6 Rxa6 Ra8 Rca2 Rxa6 Rxa6 Ra8 Rxa8 Nxa8 h4 Nb6 g4 Kg7 Kg2 h5 Kg3 hxg4 fxg4 Nd7 Kf4 Nc5 Bc2 Nd7 Bd3 Kf6 g5 Kg7 Ke3 Nc5) {+3.45/35 21} Qd7 (Qd7 Nxc6 Qd6 Rc2 Nxa4 Qxa4 Rc7 Rfc1 Ra8 Qa5 Kg7 h3 Rac8 Qxa6 h5 Qb5 f6 Rc4 Qd7 Qb6 Qd6 Qc5) {+11.03/11 54} 29. Nxc6 (Nxc6 Qd6 Rc2 Nd5 Rd1 e6 Ne5 Rxc2 Bxc2 Qc5 Be4 Nf6 Bc6 a5 h3 Rc8 Bb7 Rf8 Nc6 Nd5 Qxa5 Qf2 Qa1 h5 Qd4 Qxd4 Rxd4 f6 Kg1 Kf7 Kf2 Rh8 Ra4 Nc7 Ra3 h4 Rd3 Nd5 Ba6 Ra8 Bc4 Rc8) {+3.21/37 37} Qd6 (Qd6 Rc2 Nxa4 Qxa4 Rc7 Rfc1 Ra8 Qa5 Kg7 h3 f6 Rc4 Rac8 Qxa6 Qe6 Qb5 Qd7 Qb6 Qd6 f4 Kh6 Qb5) {+10.95/12 10} 30. Rc2 (Rc2 Rc7 Rd1 Qf6 Ne5 Rxc2 Bxc2 e6 Be4 Rd8 Rb1 Nd7 Nc6 Qxa1 Rxa1 Rf8 Rxa6 Nb8 Nxb8 Rxb8 h4 h6 g3 Rb2 Kg1 Rd2 Ra7 Kf8 Kf1 g5 h5 f5 Bc6 Rh2 Rh7 Rxh5 Bd7 f4 g4 Rh2 Bxe6 Ke8 Bc4) {+3.27/37 18} * [/pgn]
MikeGL
Posts: 1010
Joined: Thu Sep 01, 2011 2:49 pm

Re: LC0 bad search and poor tactics

Post by MikeGL »

M ANSARI wrote: Mon Jan 28, 2019 7:49 am I just did a 50 game match with SF latest Dev against Lc0 and although Lc0 won that match easily (+90 ELO), I can easily point to many one move howlers. I mean where a position would move from +5 for Lc0 (totally winning agreed by both sides) to -5 in one move. Now of course you could say a 3 min and 2 sec increment is very fast time control and those one move tactical misses are to be expected .... BUT ... I took those positions and put Lc0 on it and let it sit for a long time. Even after 10 minutes in one position, it could not see much difference between 2 moves ... one that is totally 0.00 and the other that SF and Komodo immediately goes around +4 and losing. It took SF and Komodo less than 1 second to see the tactic. In another position where Lc0 was also up +5 it missed a simple tactic where it would finally break through and win. It actually ended up missing that tactic and moving from +5 to losing the entire game. I am putting together a post with these things I saw and I must say I was incredibly impressed. I have been around computer chess long enough to realize that the small things that cause the engine to play some bad moves will be easily fixed soon enough, so that part doesn't really concern me. What got my attention more was that in basically every game Lc0 got out of the opening with a nice advantage. Sure it didn't win many of those games and even lost some winning positions, but that seems like it will be easy to patch up.

I think there is something that can be improved in the way it searches. I don't know much how the engine works, but if it is following AI then it should mimic how a strong GM plays chess. I wouldn't change how it plays the opening or middle game, but then there should be a trigger where it starts to look for tactics in a position ... sort of like the old Rybka Winfinder. If you look at how a human plays chess, they will play positionally and then they will feel that there is a tactical shot in the position ... they might not be able to calculate it quickly in time, but they know it is there and they will start looking for the working tactic. If Lc0 could use the unused CPU power it has at its disposal to identify and look for these tactics it will be an even more formidable chess engine. Not only would it be stronger, but it will also be a much more useful chess engine for analysis.

By the way, if you want to see something incredible ... watch this tourney of Magnus Carlsen playing in a Lichess 1 minute bullet tourney just a few weeks back. Amazing positional acumen with incredible tactical awareness ... what a potent combination. If you enjoy the game of chess you will really enjoy this video

https://www.youtube.com/watch?v=8RO1YrQMBQ8
Would be very nice to see and analyze those 50 games of SF dev and LC0_xxxxx or kindly advise in case I missed the link from other thread.
I told my wife that a husband is like a fine wine; he gets better with age. The next day, she locked me in the cellar.
User avatar
M ANSARI
Posts: 3707
Joined: Thu Mar 16, 2006 7:10 pm

Re: LC0 bad search and poor tactics

Post by M ANSARI »

I have done analysis on some of the 50 games and will post something on it. I think many will find it interesting. No doubt Lc0 is an extremely strong engine, but even more than that ... there is no doubt that it will become much stronger very quickly. At the moment it is still not a useful stand alone engine for analysis but if used with another engine to check that no instant tactics are missed, it could be very useful.
MikeGL
Posts: 1010
Joined: Thu Sep 01, 2011 2:49 pm

Re: LC0 bad search and poor tactics

Post by MikeGL »

M ANSARI wrote: Wed Jan 30, 2019 8:11 am I have done analysis on some of the 50 games and will post something on it. I think many will find it interesting. No doubt Lc0 is an extremely strong engine, but even more than that ... there is no doubt that it will become much stronger very quickly. At the moment it is still not a useful stand alone engine for analysis but if used with another engine to check that no instant tactics are missed, it could be very useful.
thanks, would be ineresting to review the said games.
I told my wife that a husband is like a fine wine; he gets better with age. The next day, she locked me in the cellar.
User avatar
M ANSARI
Posts: 3707
Joined: Thu Mar 16, 2006 7:10 pm

Re: LC0 bad search and poor tactics

Post by M ANSARI »