Will computer soon be playing chess strong as humans?

[AdminX]

Komodo 9.3 doesn't like Rg7 as Stockfish does.
It takes a bit for Komodo to recognize Rg7 as quite good move (peak eval was +0.89), but never best

Code: Select all

D=26, 1 715 MN
+0,90	17.a3 Na4 18.Nxa4 bxa4 19.c3 Qb6 20.Bc4 Rb8
+0,81	17.b4 Na4 18.Nxa4 bxa4 19.a3 a5 20.Bb5 Bxb5
+0,66	17.e5 d5 18.Qh5 0-0-0 19.Kb1 b4 20.Nce2 Be8
+0,63	17.Rg7 h5 18.a3 Rb8 19.b4 Bxg7 20.fxg7 Rg8	(+0.77 @ D=28)

17.Rg7, D=27, 4 469 MN
+0,77	17. ... Qb7 18.Re1 e5 19.Nb3 Bxg7 20.fxg7 Nxb3+
+1,08	17. ... b4 18.Nd5 Qa5 19.Kb1 exd5 20.exd5 Bxg7
+1,18	17. ... h5 18.Kb1 Qb7 19.Qe3 Bxg7 20.fxg7 Rg8
+1,45	17. ... Bxg7 18.fxg7 Rg8 19.e5 0-0-0 20.exd6 Qb6

It does at much deeper depths.

Analysis by Komodo 9.42 64-bit Depth=37:

1. +/- (0.98): 17.Rg7 b4 18.Nd5 exd5 19.exd5 Qa5 20.Kb1 Bxg7 21.fxg7 Rg8 22.Qf6 b3 23.axb3 Ne4 24.Qf4 Rxg7 25.Qxe4+ Kf8 26.Ne6+ Bxe6 27.dxe6 Re8 28.Bc4 Qe5 29.Qf3 f6 30.Qf2 Rg4 31.Qb6 Rg7 32.Rxd6 Qxh2 33.Rd1 Rge7 34.Rg1 Qe5 35.Qf2 Rg7 36.Re1 Qg5 37.Qb6 Rge7 38.Rf1 Qe5 39.Qf2 Rg7 40.Qh4 Kg8 41.Bxa6 Kh8 42.Bc4 Qg5 43.Qe1 Qe5 44.Qf2

2. +/- (0.97): 17.b4 Nb7 18.Be2 0-0-0 19.a4 Kb8 20.axb5 a5 21.bxa5 Rc8 22.a6 Qxc3 23.Qxc3 Rxc3 24.axb7 Bh6+ 25.Kb2 Re3 26.Bh5 Bf4 27.Bxf7 Be5 28.Kc1 Ra3 29.Nb3 Kxb7 30.Bg8 Bxf6 31.Rxd6 Kc7 32.Rgd1 Bxb5 33.Bxe6 Rd8 34.Rxd8 Bxd8 35.Kb1 Be7 36.e5 Bc6 37.Kb2 Ra4 38.Bd5 Bxd5 39.Rxd5 Kc6 40.Rd3 Re4 41.Nd4+ Kd5 42.Nf5+ Ke6 43.Nxe7

3. +/- (0.73): 17.a3 Na4 18.Nxa4 bxa4 19.c3 Qb6 20.Bc4 Rb8 21.Qe2 h5 22.Rg2 Qc5 23.Bxa6 h4 24.Rg4 Rh6 25.Rf1 e5 26.Nf5 Rxf6 27.Rxh4 Be6 28.Bd3 Kd7 29.Rh7 Rg6 30.Rd1 Rf6 31.Bc2 Bxf5 32.Bxa4+ Ke7 33.exf5 Bh6+ 34.Kb1 Qxc3 35.Rh8 Qc7 36.Rxb8 Qxb8 37.Bc2 Bf4 38.Qg2 Qc7 39.Bd3 Rh6

PS: I have no idea when it first liked 17. Rg7 as I went to bed and let my systems run a deep cloud analysis on the position.

[Waschbaer]

Komodo 9.42 on my system:

• [+1.43] d=38 17.b4 Na4 18.Nxa4 bxa4 19.a3 e5 20.Nf5 a5 21.b5 Rb8 22.Rd5 h5 23.Rg5 Be6 24.Ng7 Bxg7 25.fxg7 Rg8 26.Rd2 h4 27.Qe3 Ke7 28.Rg1 Qc5 29.Qxc5 dxc5 30.c4 Rbd8 31.Rb2 Rb8 32.Be2 Kf6 33.Bd3 Rbd8 34.Rf2 Ke7 35.Kc2 Rd6 36.Kc3 f6 37.Rd2 Rd4 38.b6 Rd7 39.Rb2 Rb7 40.Be2 Bd7 41.Bh5 Bc6 42.Kd3 Kd8 43.Rg6(1:49:33) 62461582kN

[yurikvelo]

At virtual D=38, 17.a3 Komodo give 0.99 for 17.a3

17. a3 Na4 18. Nxa4 bxa4 19. c3 Qb6 20. Bc4 Rb8 21. Qe2 h5 22. Rg2 Qc5 23. Bxa6

D=25, 204 MN
+0,99 23. ... h4 24.Bc4 h3 25.Rg3 Rh4 26.Rf1 Qe5
+1,05 23. ... Rh6 24.Rf1 e5 25.Nc2 Bh3 26.Nb4 Bxg2

[Vinvin]

Will computer soon be playing chess strong as humans

Dear Thomas, computer score around 50% or more against top humans on a home computer for more than 10 years. The watershed was around 2004, see history here : http://chessprogramming.wikispaces.com/ ... ournaments .

[AdminX]

At virtual D=38, 17.a3 Komodo give 0.99 for 17.a3

17. a3 Na4 18. Nxa4 bxa4 19. c3 Qb6 20. Bc4 Rb8 21. Qe2 h5 22. Rg2 Qc5 23. Bxa6

D=25, 204 MN
+0,99 23. ... h4 24.Bc4 h3 25.Rg3 Rh4 26.Rf1 Qe5
+1,05 23. ... Rh6 24.Rf1 e5 25.Nc2 Bh3 26.Nb4 Bxg2

I am getting the same.

[Thomas Lagershausen]

Will computer soon be playing chess strong as humans

Dear Thomas, computer score around 50% or more against top humans on a home computer for more than 10 years. The watershed was around 2004, see history here : http://chessprogramming.wikispaces.com/ ... ournaments .

Dear Vincent,

i know that computers were very sucessfull against human players. But the reason for this was not that computers find outstanding moves. The main reason for this is that they avoid all this tactical weak moves human players make.

So statistical computer play better chess as humans. But, and this is my point, computers are finding less great moves in chessgames as strong humans did.

I know too that computers have passed the humans in chess, but they still play the game not so deep humans can do.

So let us enjoy this situation and let the programers keep improving the understanding of chess in the strongest chessprograms.

Have a good time.

[yurikvelo]

At virtual D=44, 17.a3 Komodo give 0.90 for 17.a3

17. a3 Na4 18. Nxa4 bxa4 19. c3 Qb6 20. Bc4 Rb8 21. Qe2 h5 22. Rg2 Qc5 23. Bxa6


D=31, 4 005 MN
+0,90 23. ... Rh6 24.Rf1 e5 25.Nc2 Bh3 26.Nb4 Bxg2
+1,18 23. ... h4 24.Bc4 Rh6 25.Rf1 Rh8 26.Kc2 h3

At virtual D=49, 17.a3 Komodo give 1.02 for 17.Rg7

17.Rg7 b4 18.Nd5 exd5 19.exd5 Qa5 20.Kb1 Bxg7 21.fxg7 Rg8
22.Qf6 b3 23.axb3 Ne4 24.Qf4 Rxg7 25.Qxe4+ Kf8 26.Ne6+ Bxe6
27.dxe6 Re8 28.Bc4 Qe5 29. Qf3 f6

D=23, 112 MN
+1,02 30.Qf2 Rg4 31.Qb6 Rg7 32.Rxd6 Qxh2 33.Rd1 Qe5
+0,96 30.Qf1 Rg6 31.Qh1 Rg7 32.Re1 Qa5 33.Qf1 Kg8

[zullil]

Louis,
which SF version did you use?

Well, I'm not exactly sure. I compiled it on March 18, 2016:

Code: Select all

Stockfish 180316 64 POPCNT by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

16 threads and I think I used just 1 GB hash.

Perhaps I was simply very lucky with the non-deterministic search. Don't have time to test more right now.

[yurikvelo]

But the reason for this was not that computers find outstanding moves.

Main reason for moves to be considered outstanding is its psycological power to make opponent blunder

In this game fenomenal success of Rg7 was due to 20. ..b4?? blunder (forced Mate in 19)

If Chandler played 20. .. Rc8, than 17.Rg7 was just one of good moves.


Anyway, good success of 17.Rg7 even without b4 blunder is thanks to Chandler eating poisoned Rook, which strong engines cannot do


17.Rg7, D=27, 4 469 MN
+0,77 17. ... Qb7 18.Re1 e5 19.Nb3 Bxg7 20.fxg7 Nxb3+
+1,08 17. ... b4 18.Nd5 Qa5 19.Kb1 exd5 20.exd5 Bxg7
+1,18 17. ... h5 18.Kb1 Qb7 19.Qe3 Bxg7 20.fxg7 Rg8
+1,45 17. ... Bxg7 18.fxg7 Rg8 19.e5 0-0-0 20.exd6 Qb6 <--- Chandler

[yurikvelo]

[pgn][Event "Hoogovens A Tournament"]
[Site "Wijk aan Zee NED"]
[Date "1999.01.20"]
[EventDate "?"]
[Round "4"]
[Result "1-0"]
[White "Garry Kasparov"]
[Black "Veselin Topalov"]
[ECO "B06"]
[WhiteElo "2812"]
[BlackElo "2700"]
[PlyCount "87"]

1. e4 d6 2. d4 Nf6 3. Nc3 g6 4. Be3 Bg7 5. Qd2 c6 6. f3 b5
7. Nge2 Nbd7 8. Bh6 Bxh6 9. Qxh6 Bb7 10. a3 e5 11. O-O-O Qe7
12. Kb1 a6 13. Nc1 O-O-O 14. Nb3 exd4 15. Rxd4 c5 16. Rd1 Nb6
17. g3 Kb8 18. Na5 Ba8 19. Bh3 d5 20. Qf4+ Ka7 21. Rhe1 d4
22. Nd5 Nbxd5 23. exd5 Qd6 24. Rxd4 cxd4 25. Re7+ Kb6
26. Qxd4+ Kxa5 27. b4+ Ka4 28. Qc3 Qxd5 29. Ra7 Bb7 30. Rxb7
Qc4 31. Qxf6 Kxa3 32. Qxa6+ Kxb4 33. c3+ Kxc3 34. Qa1+ Kd2
35. Qb2+ Kd1 36. Bf1 Rd2 37. Rd7 Rxd7 38. Bxc4 bxc4 39. Qxh8
Rd3 40. Qa8 c3 41. Qa4+ Ke1 42. f4 f5 43. Kc1 Rd2 44. Qa7 1-0[/pgn]

Another example of "brilliant" move 24.Rxd4!!?

Engines never select this move. Not because they are blind, but because they do not blunder

Code: Select all

Stockfish 171114, D=42, 100 BN
01. -0.43: 24...Kb6 25.b4 Qxf4 26.Rxf4 Nxd5 27.Rxf7 cxb4 28.axb4 Nxb4
02. =0.00: 24...Rhe8 25.Rxe8 Nxe8 26.Qxf7+ Kb6 27.Nc6 Bxc6 28.dxc6 Qxd4
03. =0.00: 24...Rhf8 25.c4 cxd4 26.Qxd4+ Kb8 27.Re7 Nd7 28.Bxd7 Rxd7
04. =0.00: 24...Rhg8 25.c4 cxd4 26.Qxd4+ Kb8 27.c5 Qc7 28.Qxf6 Rxd5
05. =0.00: 24...Bxd5 25.Rxd5 Nxd5 26.Qxf7+ Nc7 27.Re6 Qd1+ 28.Ka2 Rd7  
06. =0.11: 24...g5 25.Qxd6 Rxd6 26.Rdd1 Rxd5 27.Rxd5 Bxd5 28.c4 bxc4
07. =0.15: 24...Nxd5 25.Qxf7+ Kb8 26.Rdd1 c4 27.Re6 Qc7 28.Qxc7+ Nxc7
08. +0.42: 24...Qxf4 25.Rxf4 Nxd5 26.Rxf7+ Kb6 27.Nb3 Rhe8 28.Rxe8 Rxe8
09. +0.70: 24...Kb8 25.Qxd6+ Rxd6 26.Rf4 Bxd5 27.c4 Nh5 28.cxd5 Nxf4
10. +0.86: 24...Bb7 25.Qxd6 Rxd6 26.Rdd1 Re8 27.Rxe8 Nxe8 28.Nxb7 Kxb7
11. +1.47: 24...h5 25.c4 g5 26.Qxd6 Rxd6 27.Re7+ Kb8 28.Rd2 g4
12. +1.95: 24...h6 25.c4 bxc4 26.Nxc4 Nxd5 27.Qxf7+ Qc7 28.Qxc7+ Nxc7
13. +2.85: 24...c4 25.Rxc4 Qxf4 26.Rxf4 Nxd5 27.Rxf7+ Kb8 28.c4 bxc4
14. +3.62: 24...Ne8 25.Qxf7+ Nc7 26.Re6 Rd7 27.Rxd6 Rxf7 28.Rd2 Ne8
15. +3.87: 24...b4 25.axb4 cxb4 26.Rxb4 Qxf4 27.Rxf4 Rhe8 28.Rxe8 Nxe8
16. +5.30: 24...cxd4 25.Re7+ Kb8 26.Qxd4 Nd7 27.Bxd7 Rxd7 28.Rxd7 Qxd7     <----- Topalov
17. +6.18: 24...Ng4 25.Nc6+ Bxc6 26.dxc6 Qxf4 27.Rxf4 Nh6 28.Re7+ Ka8