Chess Genius 3 Beat Kasparov on a Pentium 90?

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

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
BrendanJNorman
Posts: 1271
Joined: Sun Feb 07, 2016 11:43 pm
Full name: Brendan J Norman

Re: Chess Genius 3 Beat Kasparov on a Pentium 90?

Post by BrendanJNorman » Wed May 31, 2017 1:17 pm

Uri Blass wrote:
BrendanJNorman wrote:
Guenther wrote:
BrendanJNorman wrote:
Vinvin wrote:+ In AubervilliersRapid Open 1993 and 1994, Chess Genius made a perf of 2530 and 2540 (hardware was probably 486-66 MHz) :
http://chessprogramming.wikispaces.com/ ... tions-1993
Must have been the magical opening book. :wink:
All he (Uri) wanted to say is, that 1.e4 Nf6 probably NEVER was played by any program for the Black side in serious Human vs. Comp games.
(Aljechin will be blocked for such a match for good reasons... the link below verifies it for Genius, which means he is right)

http://www.chessgames.com/player/chess_genius.html

Interesting though that they once allowed Genius to play a Pirc defense, which is also an opening C players judge as bad.
Hey Guenther,

I don't think the Alekhine (or any "inferior" opening) was blocked, otherwise CG3's opening book would be consistent across the game range.

Instead we see Pirc (known as a "come and get me" opening among strong players), Caro Kann (another "come and get me" opening for computers, but more solid than the Pirc), and 1...e5 (Petrov, Ruy Lopez and Two Knights).

I'd say that it's opening book is/was weighted to play 1...e5 more often, but throw in other openings based on some statistical distribution (as in, play the Caro Kann 40% of the time, Pirc 25% of the time or whatever), and that the only reason we don't see Alekhine's in the chessgames list is due to a relative lack of 1.e4 games.

As I alluded to before, the MAINLINE Alekhine (where white plays 3.d4 d6 4. c4 Nb6 5.f4 dxe5 6.fxe5 Nc6 etc isn't so bad for black - especially against a human (who might have more trouble holding the center together against a computer's tactical powers.

My playing an offbeat idea 3.Nc3 e6 4.d4 Bb4 5.Bd2 brought the computer out of its book and provoked a strategic mistake immediately since 5...Nxc3 6.bxc3 Be7 (6...Bf8 MIGHT be better) 7.Qg4 is already much better for white.

So would you say the opening book was still the problem if I'd played the mainline?
I think that 1...Nf6 is relatively a rare move so it is easier to find good sidelines after 1.e4 Nf6 to get the computer out of book.
So I was reeeeeeellly lucky to get that opening, huh? Otherwise I'd have no chance. :lol: :roll:

Lyudmil Tsvetkov
Posts: 6037
Joined: Tue Jun 12, 2012 10:41 am

Re: Chess Genius 3 Beat Kasparov on a Pentium 90?

Post by Lyudmil Tsvetkov » Wed May 31, 2017 2:58 pm

BrendanJNorman wrote:
Uri Blass wrote:
BrendanJNorman wrote:
Guenther wrote:
BrendanJNorman wrote:
Vinvin wrote:+ In AubervilliersRapid Open 1993 and 1994, Chess Genius made a perf of 2530 and 2540 (hardware was probably 486-66 MHz) :
http://chessprogramming.wikispaces.com/ ... tions-1993
Must have been the magical opening book. :wink:
All he (Uri) wanted to say is, that 1.e4 Nf6 probably NEVER was played by any program for the Black side in serious Human vs. Comp games.
(Aljechin will be blocked for such a match for good reasons... the link below verifies it for Genius, which means he is right)

http://www.chessgames.com/player/chess_genius.html

Interesting though that they once allowed Genius to play a Pirc defense, which is also an opening C players judge as bad.
Hey Guenther,

I don't think the Alekhine (or any "inferior" opening) was blocked, otherwise CG3's opening book would be consistent across the game range.

Instead we see Pirc (known as a "come and get me" opening among strong players), Caro Kann (another "come and get me" opening for computers, but more solid than the Pirc), and 1...e5 (Petrov, Ruy Lopez and Two Knights).

I'd say that it's opening book is/was weighted to play 1...e5 more often, but throw in other openings based on some statistical distribution (as in, play the Caro Kann 40% of the time, Pirc 25% of the time or whatever), and that the only reason we don't see Alekhine's in the chessgames list is due to a relative lack of 1.e4 games.

As I alluded to before, the MAINLINE Alekhine (where white plays 3.d4 d6 4. c4 Nb6 5.f4 dxe5 6.fxe5 Nc6 etc isn't so bad for black - especially against a human (who might have more trouble holding the center together against a computer's tactical powers.

My playing an offbeat idea 3.Nc3 e6 4.d4 Bb4 5.Bd2 brought the computer out of its book and provoked a strategic mistake immediately since 5...Nxc3 6.bxc3 Be7 (6...Bf8 MIGHT be better) 7.Qg4 is already much better for white.

So would you say the opening book was still the problem if I'd played the mainline?
I think that 1...Nf6 is relatively a rare move so it is easier to find good sidelines after 1.e4 Nf6 to get the computer out of book.
So I was reeeeeeellly lucky to get that opening, huh? Otherwise I'd have no chance. :lol: :roll:
good game, but nothing special.

the losing move is already 1...Nf6

I do not know why people do not believe it is possible to score wins against opponents of similar strength. of course, it is.

for example, nowadays I have a score above 90%, when I play against Comet.

I also hate bad graphics, with eye-scratching colours and small pieces, I definitely play some 1000 elo weaker.

Uri Blass
Posts: 8434
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: Chess Genius 3 Beat Kasparov on a Pentium 90?

Post by Uri Blass » Wed May 31, 2017 3:12 pm

Lyudmil Tsvetkov wrote:
BrendanJNorman wrote:
Uri Blass wrote:
BrendanJNorman wrote:
Guenther wrote:
BrendanJNorman wrote:
Vinvin wrote:+ In AubervilliersRapid Open 1993 and 1994, Chess Genius made a perf of 2530 and 2540 (hardware was probably 486-66 MHz) :
http://chessprogramming.wikispaces.com/ ... tions-1993
Must have been the magical opening book. :wink:
All he (Uri) wanted to say is, that 1.e4 Nf6 probably NEVER was played by any program for the Black side in serious Human vs. Comp games.
(Aljechin will be blocked for such a match for good reasons... the link below verifies it for Genius, which means he is right)

http://www.chessgames.com/player/chess_genius.html

Interesting though that they once allowed Genius to play a Pirc defense, which is also an opening C players judge as bad.
Hey Guenther,

I don't think the Alekhine (or any "inferior" opening) was blocked, otherwise CG3's opening book would be consistent across the game range.

Instead we see Pirc (known as a "come and get me" opening among strong players), Caro Kann (another "come and get me" opening for computers, but more solid than the Pirc), and 1...e5 (Petrov, Ruy Lopez and Two Knights).

I'd say that it's opening book is/was weighted to play 1...e5 more often, but throw in other openings based on some statistical distribution (as in, play the Caro Kann 40% of the time, Pirc 25% of the time or whatever), and that the only reason we don't see Alekhine's in the chessgames list is due to a relative lack of 1.e4 games.

As I alluded to before, the MAINLINE Alekhine (where white plays 3.d4 d6 4. c4 Nb6 5.f4 dxe5 6.fxe5 Nc6 etc isn't so bad for black - especially against a human (who might have more trouble holding the center together against a computer's tactical powers.

My playing an offbeat idea 3.Nc3 e6 4.d4 Bb4 5.Bd2 brought the computer out of its book and provoked a strategic mistake immediately since 5...Nxc3 6.bxc3 Be7 (6...Bf8 MIGHT be better) 7.Qg4 is already much better for white.

So would you say the opening book was still the problem if I'd played the mainline?
I think that 1...Nf6 is relatively a rare move so it is easier to find good sidelines after 1.e4 Nf6 to get the computer out of book.
So I was reeeeeeellly lucky to get that opening, huh? Otherwise I'd have no chance. :lol: :roll:
good game, but nothing special.

the losing move is already 1...Nf6

I do not know why people do not believe it is possible to score wins against opponents of similar strength. of course, it is.

for example, nowadays I have a score above 90%, when I play against Comet.

I also hate bad graphics, with eye-scratching colours and small pieces, I definitely play some 1000 elo weaker.
I tried to show that 1...Nf6 is losing with no success.

Assuming 2.e5 Nd5 is the correct line for white
Here is some analysis by stockfish after it.

score of only 0.33 does not suggest a forced win.


[D]rnbqkb1r/pppppppp/8/3nP3/8/8/PPPP1PPP/RNBQKBNR w KQkq - 1 3

Stockfish_17052608_x64_modern:
1/1 00:00 96 96k +1.46 3.d4
2/2 00:00 187 187k +2.96 3.c4 c6
3/3 00:00 291 291k +2.96 3.c4 c6 4.cxd5
4/4 00:00 427 214k +2.20 3.c4 d6 4.cxd5 dxe5
5/5 00:00 777 389k +2.25 3.c4 d6 4.cxd5 dxe5 5.Qb3
6/6 00:00 4k 1,350k +1.08 3.d4 d6 4.c4 Nb4 5.Bd2 N8c6
7/8 00:00 11k 1,905k +0.74 3.c4 Nb4 4.d4 d5 5.Qb3 N8c6 6.Bd2
8/8 00:00 25k 2,065k +0.90 3.d4 Nb6 4.h3 d6 5.Nf3 dxe5 6.Nxe5 Nc6 7.Nxc6
9/12 00:00 46k 1,981k +1.27 3.d4 d6 4.Nf3 Nd7 5.Bg5 dxe5 6.dxe5 Nc5 7.Qd4
10/13 00:00 108k 2,078k +1.01 3.d4 d6 4.Nf3 dxe5 5.Nxe5 Bf5 6.Bc4 c6 7.0-0 Nd7 8.Re1 Be6
11/16 00:00 231k 2,312k +1.02 3.d4 d6 4.Nf3 Nb6 5.Bd3 c5 6.0-0 cxd4 7.exd6 Qxd6 8.Bb5+ Bd7 9.Nxd4
12/16 00:00 321k 2,359k +0.83 3.Nf3 d6 4.d4 Bf5 5.Bd3 Bxd3 6.Qxd3 Nc6 7.exd6 Ncb4 8.Qd1 exd6 9.0-0
13/16 00:00 419k 2,279k +0.79 3.d4 d6 4.Nf3 Bf5 5.Bd3 Bxd3 6.Qxd3 Nc6 7.0-0 dxe5 8.Qb5 exd4 9.Nxd4
14/19 00:00 775k 1,973k +0.74 3.d4 d6 4.Nf3 dxe5 5.Nxe5 e6 6.Be2 f6 7.Nf3 c5 8.0-0 Nc6 9.Bb5 Bd7 10.Re1
15/22 00:00 1,211k 1,922k +0.74 3.d4 d6 4.Nf3 dxe5 5.Nxe5 e6 6.Be2 f6 7.Nf3 c5 8.0-0 Nc6 9.Bb5 Nc7 10.Nc3 Nxb5 11.Nxb5 Qd5 12.Nc3
16/24 00:01 2,969k 1,945k +0.64 3.c4 Nb6 4.d4 d6 5.Nf3 dxe5 6.Nxe5 Nc6 7.Nxc6 bxc6 8.c5 Nd5 9.Bc4 e6 10.Nc3 Be7 11.Ne4 Nf6
17/27 00:01 3,372k 1,956k +0.75 3.c4 Nb6 4.d4 d6 5.Nf3 dxe5 6.Nxe5 N6d7 7.Nf3 e5 8.dxe5 Bb4+ 9.Nbd2 Nc6 10.a3 Bxd2+ 11.Bxd2 0-0 12.Be2 Ncxe5 13.Nxe5 Nxe5 14.Bb4 Re8 15.0-0 Qxd1 16.Rfxd1
18/30 00:02 4,174k 1,909k +0.68 3.c4 Nb6 4.d4 d6 5.Nf3 dxe5 6.Nxe5 Nc6 7.Nxc6 bxc6 8.c5 Nd5 9.Bc4 e5 10.0-0 Be7 11.Nd2 exd4 12.Ne4 0-0 13.Qxd4
19/30- 00:03 5,883k 1,941k +0.61 3.c4 Nb6
19/30 00:03 6,435k 1,968k +0.59 3.c4 Nb6 4.d4 d6 5.Nf3 dxe5 6.Nxe5 N6d7 7.Nf3 e5 8.Nc3 exd4 9.Nxd4 Be7 10.Be2 0-0 11.0-0 Bf6 12.Be3 Nc5 13.Rc1
20/30+ 00:03 6,567k 1,971k +0.66 3.c4
20/30 00:03 7,118k 1,990k +0.60 3.c4 Nb6 4.d4 d6 5.Nf3 dxe5 6.Nxe5 N6d7 7.Nxd7 Nxd7 8.Be2 e5 9.Nc3 Bd6 10.Be3 exd4 11.Qxd4 0-0 12.Rd1 Be5 13.Qd2
21/34+ 00:05 11,077k 2,145k +0.67 3.c4
21/34 00:05 11,430k 2,153k +0.61 3.c4 Nb6 4.d4 d6 5.f4 Bf5 6.Nc3 e6 7.Nf3 Be7 8.Be2 0-0 9.0-0 c5 10.d5 Na6 11.Ne1 Bg6 12.Nc2 Bxc2 13.Qxc2 Nb4 14.Qe4 dxe5 15.dxe6 Qd4+ 16.Be3 Qxe4
22/34- 00:06 15,265k 2,241k +0.54 3.c4 Nb6
22/34 00:07 16,716k 2,257k +0.58 3.c4 Nb6 4.d4 d6 5.Nf3 dxe5 6.Nxe5 N6d7 7.Nxd7 Nxd7 8.Nc3 e5 9.dxe5 Bb4 10.f4 0-0 11.a3 Bxc3+ 12.bxc3 Qe7 13.Be2 Nc5 14.0-0 Bf5 15.Be3
23/34+ 00:09 23,384k 2,349k +0.65 3.d4
23/35- 00:11 26,536k 2,373k +0.58 3.d4 d6
23/35- 00:13 31,927k 2,401k +0.47 3.d4 d6
23/35+ 00:15 37,791k 2,424k +0.56 3.d4
23/35 00:15 38,697k 2,429k +0.62 3.d4 d6 4.c4 Nb6 5.exd6 exd6 6.Nc3 Be7 7.Bd3 Nc6 8.Nge2 0-0 9.0-0 Bf6 10.Be3 Nb4 11.b3 Nxd3 12.Qxd3 g6 13.d5 Be5 14.Bd4 Bf5 15.Qd2
24/35- 00:19 46,904k 2,453k +0.55 3.d4 d6
24/35- 00:20 50,270k 2,461k +0.47 3.d4 d6
24/38+ 00:24 59,638k 2,452k +0.55 3.d4
24/38 00:24 61,221k 2,449k +0.63 3.d4 d6 4.c4 Nb6 5.exd6 exd6 6.Nc3 Be7 7.Bd3 Nc6 8.Nge2 0-0 9.b3 Bg4 10.h3 Bxe2 11.Nxe2 Re8 12.0-0 Bg5 13.Bb2 Nb4 14.Bb1 g6 15.Re1
25/38- 00:25 63,301k 2,444k +0.56 3.d4 d6
25/38 00:28 70,139k 2,424k +0.53 3.d4 d6 4.c4 Nb6 5.exd6 exd6 6.Nc3 Be7 7.Bd3 0-0 8.Nge2 Nc6 9.0-0 Bf6 10.b3 Bg4 11.Be3 Bxd4 12.Bxh7+ Kxh7 13.Qd3+ Kh8 14.Bxd4 Qh4 15.f3 Bd7 16.Bf2 Qh7 17.Qd2 Rae8 18.h3 Kg8
26/39+ 00:31 75,803k 2,433k +0.60 3.d4
26/39+ 00:31 77,361k 2,438k +0.68 3.d4
26/41- 00:34 85,343k 2,449k +0.60 3.d4 d6
26/41 00:52 122,943k 2,362k +0.63 3.c4 Nb6 4.d4 d6 5.Nf3 dxe5 6.Nxe5 Nc6 7.Nxc6 bxc6 8.c5 Nd5 9.Bc4 e5 10.Nc3 Be6 11.Qa4 Qd7 12.0-0 Nxc3 13.bxc3 exd4 14.Bxe6 fxe6 15.cxd4 Be7 16.Rb1 0-0 17.Rd1 Bf6
27/41- 00:55 130,436k 2,356k +0.56 3.c4 Nb6
27/41- 00:59 138,051k 2,325k +0.48 3.c4 Nb6
27/41 01:14 171,163k 2,311k +0.44 3.c4 Nb6 4.Nf3 d6 5.exd6 exd6 6.d4 Be7 7.Bd3 0-0 8.0-0 Nc6 9.Nc3 Bf6 10.Be3 Re8 11.b3 Nb4 12.h3 c5 13.Rc1 d5 14.dxc5 dxc4 15.Bxh7+ Kxh7 16.cxb6 Qxd1 17.Rfxd1 cxb3 18.axb3 axb6 19.Bxb6
28/41+ 01:25 194,338k 2,285k +0.51 3.c4
28/41- 01:37 219,313k 2,243k +0.44 3.c4 Nb6
28/41+ 01:45 236,895k 2,247k +0.51 3.c4
28/41 01:50 248,470k 2,257k +0.46 3.d4 d6 4.c4 Nb6 5.exd6 exd6 6.Nf3 Be7 7.Bd3 0-0 8.0-0 Nc6 9.Nc3 Bg4 10.h3 Bxf3 11.Qxf3 Bf6 12.Be3 g6 13.Qf4 Nb4 14.Be2 Nc2 15.Rad1 Nxe3 16.fxe3 Bg7 17.e4 Qe7 18.Nd5 Nxd5 19.cxd5 c6 20.dxc6 bxc6
29/45 02:09 291,041k 2,256k +0.44 3.d4 d6 4.c4 Nb6 5.exd6 exd6 6.Nc3 Nc6 7.Nf3 Be7 8.d5 Ne5 9.Nxe5 dxe5 10.Be3 0-0 11.Bd3 f5 12.0-0 e4 13.Be2 Bb4 14.Qd2 c6 15.dxc6 Qxd2 16.Bxd2 bxc6 17.Nxe4 Be7 18.Bg5 Re8 19.Bxe7 Rxe7
30/45- 02:25 329,222k 2,268k +0.37 3.d4 d6
30/45 02:46 379,781k 2,279k +0.36 3.c4 Nb6 4.Nc3 d6 5.exd6 exd6 6.d4 Nc6 7.Nf3 Be7 8.d5 Ne5 9.Nxe5 dxe5 10.Be3 c6 11.d6 Bg5 12.Qd2 Bxe3 13.Qxe3 0-0 14.Be2 Be6 15.c5 Nd7 16.0-0 f5 17.f4 e4 18.a3
31/45+ 02:58 408,660k 2,286k +0.43 3.c4
31/45 03:06 427,463k 2,293k +0.37 3.c4 Nb6 4.Nc3 d6 5.exd6 exd6 6.d4 Nc6 7.Nf3 Be7 8.d5 Ne5 9.Nxe5 dxe5 10.Be3 0-0 11.Bd3 f5 12.0-0 e4 13.Be2 Nd7 14.Qd2 Bd6 15.Bg5 Be7 16.Bf4 Bd6 17.Rad1 Qe7 18.Nb5 Bxf4 19.Qxf4 Ne5 20.d6 cxd6 21.Nxd6
32/45+ 03:19 456,139k 2,290k +0.44 3.c4
32/45- 03:46 521,131k 2,297k +0.37 3.c4 Nb6
32/45- 05:02 702,757k 2,322k +0.26 3.c4 Nb6
32/45+ 05:38 788,559k 2,327k +0.35 3.c4
32/45 05:46 806,622k 2,329k +0.34 3.c4 Nb6 4.d4 d6 5.exd6 exd6 6.Nc3 Nc6 7.Nf3 Be7 8.d5 Ne5 9.Nxe5 dxe5 10.Be3 0-0 11.Be2 Bf5 12.0-0 Nd7 13.Qd2 Bg6 14.Rad1 h6 15.Nb5 Nf6 16.Bf3 Ne4 17.Bxe4 Bxe4 18.Qe2 Bg6 19.Bxa7 f6 20.Be3 Rxa2 21.Nc3
33/45+ 05:56 830,184k 2,329k +0.41 3.c4
33/45- 06:17 879,359k 2,327k +0.34 3.c4 Nb6
33/45+ 06:28 904,416k 2,329k +0.41 3.c4
33/45+ 06:36 923,145k 2,330k +0.57 3.c4
33/52 07:17 1,015,719k 2,324k +0.48 3.c4 Nb6 4.d4 d6 5.exd6 exd6 6.Nc3 Nc6 7.Nf3 Be7 8.d5 Ne5 9.Nxe5 dxe5 10.Be3 0-0 11.Bd3 f5 12.0-0 f4 13.Bd2 g6 14.Qe2 Bf5 15.Bxf5 Rxf5 16.Ne4 Na4 17.Rad1 Nxb2 18.Rb1 Na4 19.Rxb7 Nc5 20.Rb5 Nxe4 21.Qxe4 a6 22.Ra5 Qf8 23.g4 fxg3 24.hxg3 Bc5 25.Be3
34/52 08:39 1,178,578k 2,267k +0.53 3.c4 Nb6 4.d4 d6 5.exd6 exd6 6.Nc3 Nc6 7.Nf3 Be7 8.d5 Ne5 9.Nxe5 dxe5 10.Be3 0-0 11.Bd3 f5 12.0-0 f4 13.Bd2 g6 14.Qe2 Bf5 15.Rfe1 Nd7 16.Bxf5 Rxf5 17.Rad1 a5 18.Ne4 Nc5 19.g4 Rf8 20.Bc3 f3 21.Qe3 Nxe4
35/52- 09:00 1,214,737k 2,245k +0.45 3.c4 Nb6
35/52- 09:29 1,280,315k 2,249k +0.38 3.c4 Nb6
35/52 12:12 1,608,660k 2,195k +0.31 3.c4 Nb6 4.Nc3 d6 5.exd6 exd6 6.d4 Nc6 7.Nf3 Be7 8.d5 Ne5 9.Nxe5 dxe5 10.Be3 0-0 11.Bd3 f5 12.0-0 e4 13.Be2 Bb4 14.Qc1 Nd7 15.Nb5 Ne5 16.Bd4 a6 17.Nc3 Ng6 18.a3 Bxc3 19.Qxc3 Qe7 20.c5 f4 21.Bc4 f3
36/52+ 18:03 2,323,962k 2,145k +0.38 3.d4
36/52 20:15 2,613,997k 2,151k +0.34 3.d4 d6 4.exd6 exd6 5.Nf3 Nf6 6.Bd3 Be7 7.0-0 0-0 8.Re1 c5 9.c4 cxd4 10.Nxd4 Nc6 11.Nxc6 bxc6 12.Nc3 d5 13.cxd5 Nxd5 14.Nxd5 cxd5 15.Bf4 Be6 16.Rc1 Bf6 17.Rc7 Bxb2 18.Rb7 Bc3 19.Re3 Qf6 20.Bg3 h6 21.Bc2 Rfd8
37/52- 21:57 2,836,856k 2,153k +0.27 3.d4 d6
37/52+ 23:27 3,035,580k 2,157k +0.34 3.d4
37/52- 28:29 3,692,549k 2,160k +0.27 3.d4 d6
37/55 41:18 5,285,724k 2,133k +0.23 3.d4 d6 4.exd6 cxd6 5.c3 Bf5 6.Qf3 e6 7.Bc4 Nf6 8.Qxb7 Be4 9.Bb5+ Nbd7 10.Bc6 d5 11.Qxa8 Qxa8 12.Bxa8 Bxg2 13.Bf4 Bxh1 14.Nd2 g5 15.Bxg5 Rg8 16.Ndf3 Ne4 17.Bf4 Ke7 18.Bc6 Ndf6 19.Ke2 Ng4 20.Nd2 Bh6 21.Bxh6 Nxh6 22.Nxe4 Bxe4 23.Bb5 Rg2 24.Nf3
38/55+ 52:40 6,943,884k 2,197k +0.31 3.c4
38/55+ 53:03 7,001,278k 2,199k +0.38 3.c4
38/55+ 53:58 7,140,648k 2,205k +0.49 3.c4
38/60- 55:14 7,311,586k 2,206k +0.39 3.c4 Nb6
38/60 1:00:17 7,902,327k 2,185k +0.24 3.c4 Nb6 4.d4 d6 5.f4 Bf5 6.Nc3 e6 7.Nf3 dxe5 8.fxe5 Bb4 9.Bd3 c5 10.Bxf5 cxd4 11.Nxd4 exf5 12.Nxf5 Qxd1+ 13.Kxd1 0-0 14.Nd5 Rd8 15.Kc2 Nc6 16.Rd1 Nxc4 17.Bg5 Re8 18.Nxb4 Nxb4+ 19.Kb3 Rxe5 20.Rf1 Re4 21.Kxb4 Nd2+ 22.Kc3 Nxf1 23.Rxf1 Re2 24.Nd6 h6 25.Bf4 Rxg2 26.Kb3 Rd8 27.Nxb7
39/60+ 1:01:59 8,105,409k 2,179k +0.31 3.c4
39/60- 1:05:48 8,627,874k 2,185k +0.24 3.c4 Nb6
39/63+ 1:07:04 8,798,180k 2,186k +0.31 3.c4
39/63- 1:14:55 9,873,304k 2,196k +0.22 3.c4 Nb6
39/63+ 1:16:26 10,093,986k 2,201k +0.34 3.c4
39/63 1:17:58 10,314,205k 2,204k +0.27 3.c4 Nb6 4.d4 d6 5.f4 Bf5 6.Nc3 e6 7.Nf3 dxe5 8.fxe5 Nc6 9.Be3 Bb4 10.Be2 Na5 11.c5 Nd5 12.Bd2 Ne7 13.0-0 0-0 14.Be3 Bxc3 15.bxc3 b6 16.cxb6 axb6 17.Qc1 Be4 18.Bf2 Qd5 19.c4 Qc6 20.Nh4 h6 21.h3 Ng6 22.Nf3 Qb7 23.Qe3
40/63+ 1:27:16 11,460,205k 2,189k +0.34 3.d4
40/63- 1:29:58 11,851,333k 2,195k +0.27 3.d4 d6
40/63 1:37:04 12,883,761k 2,212k +0.33 3.d4 d6 4.Nf3 g6 5.Bc4 Nb6 6.Bb3 Bg7 7.0-0 0-0 8.exd6 cxd6 9.c3 Nc6 10.Re1 Bf5 11.d5 Na5 12.Na3 Rc8 13.Be3 Re8 14.Bd4 Bxd4 15.Qxd4 Qc7 16.Nb5 Qd8 17.Qh4 Bd7 18.Qh6 f6 19.Nbd4 Nxb3 20.axb3 Nxd5 21.Ne6 Bxe6 22.Rxe6 Qb6 23.Nd4 Qc5

Ras
Posts: 1130
Joined: Tue Aug 30, 2016 6:19 pm
Contact:

Re: Chess Genius 3 Beat Kasparov on a Pentium 90?

Post by Ras » Wed May 31, 2017 5:33 pm

With your game, were the time controls comparable? I mean, the effective time relative to the computing speed?

It's difficult for me to think of the whole opening as weak. Not only because of Alekhine himself, but also other masters: https://en.wikipedia.org/wiki/Alekhine's_Defence#Use

However, the opening requires understanding the idea behind it (which is easy for a human) and also implementing it the right way on the board (which is where I struggle ^^), and I can well imagine that the CG3 engine doesn't even know what to do.

With the CT800, I'm seeing something like that with Dutch as Black, which is (nearly) always a desaster. Even against clearly inferior opponents, a draw is already lucky, and I've never seen a win. The reason isn't that Dutch is that bad, it's that the engine is not able to cope with the position. I relegated Dutch to passive kowledge for that reason.

Tested against a dedicated unit 200 ELO below the CT800 on average, the odds with Dutch as Black are so bad that the inferior machine actually has at least a plus of 200 ELO from the winning percentage. That's well in the order of magnitude of 500 ELO.

It's not that the opening book can do wonders here - but it can at least avoid exposing glaring weaknesses on a silver plate.

Lyudmil Tsvetkov
Posts: 6037
Joined: Tue Jun 12, 2012 10:41 am

Re: Chess Genius 3 Beat Kasparov on a Pentium 90?

Post by Lyudmil Tsvetkov » Thu Jun 01, 2017 7:13 am

Uri Blass wrote:
Lyudmil Tsvetkov wrote:
BrendanJNorman wrote:
Uri Blass wrote:
BrendanJNorman wrote:
Guenther wrote:
BrendanJNorman wrote:
Vinvin wrote:+ In AubervilliersRapid Open 1993 and 1994, Chess Genius made a perf of 2530 and 2540 (hardware was probably 486-66 MHz) :
http://chessprogramming.wikispaces.com/ ... tions-1993
Must have been the magical opening book. :wink:
All he (Uri) wanted to say is, that 1.e4 Nf6 probably NEVER was played by any program for the Black side in serious Human vs. Comp games.
(Aljechin will be blocked for such a match for good reasons... the link below verifies it for Genius, which means he is right)

http://www.chessgames.com/player/chess_genius.html

Interesting though that they once allowed Genius to play a Pirc defense, which is also an opening C players judge as bad.
Hey Guenther,

I don't think the Alekhine (or any "inferior" opening) was blocked, otherwise CG3's opening book would be consistent across the game range.

Instead we see Pirc (known as a "come and get me" opening among strong players), Caro Kann (another "come and get me" opening for computers, but more solid than the Pirc), and 1...e5 (Petrov, Ruy Lopez and Two Knights).

I'd say that it's opening book is/was weighted to play 1...e5 more often, but throw in other openings based on some statistical distribution (as in, play the Caro Kann 40% of the time, Pirc 25% of the time or whatever), and that the only reason we don't see Alekhine's in the chessgames list is due to a relative lack of 1.e4 games.

As I alluded to before, the MAINLINE Alekhine (where white plays 3.d4 d6 4. c4 Nb6 5.f4 dxe5 6.fxe5 Nc6 etc isn't so bad for black - especially against a human (who might have more trouble holding the center together against a computer's tactical powers.

My playing an offbeat idea 3.Nc3 e6 4.d4 Bb4 5.Bd2 brought the computer out of its book and provoked a strategic mistake immediately since 5...Nxc3 6.bxc3 Be7 (6...Bf8 MIGHT be better) 7.Qg4 is already much better for white.

So would you say the opening book was still the problem if I'd played the mainline?
I think that 1...Nf6 is relatively a rare move so it is easier to find good sidelines after 1.e4 Nf6 to get the computer out of book.
So I was reeeeeeellly lucky to get that opening, huh? Otherwise I'd have no chance. :lol: :roll:
good game, but nothing special.

the losing move is already 1...Nf6

I do not know why people do not believe it is possible to score wins against opponents of similar strength. of course, it is.

for example, nowadays I have a score above 90%, when I play against Comet.

I also hate bad graphics, with eye-scratching colours and small pieces, I definitely play some 1000 elo weaker.
I tried to show that 1...Nf6 is losing with no success.

Assuming 2.e5 Nd5 is the correct line for white
Here is some analysis by stockfish after it.

score of only 0.33 does not suggest a forced win.


[D]rnbqkb1r/pppppppp/8/3nP3/8/8/PPPP1PPP/RNBQKBNR w KQkq - 1 3

Stockfish_17052608_x64_modern:
1/1 00:00 96 96k +1.46 3.d4
2/2 00:00 187 187k +2.96 3.c4 c6
3/3 00:00 291 291k +2.96 3.c4 c6 4.cxd5
4/4 00:00 427 214k +2.20 3.c4 d6 4.cxd5 dxe5
5/5 00:00 777 389k +2.25 3.c4 d6 4.cxd5 dxe5 5.Qb3
6/6 00:00 4k 1,350k +1.08 3.d4 d6 4.c4 Nb4 5.Bd2 N8c6
7/8 00:00 11k 1,905k +0.74 3.c4 Nb4 4.d4 d5 5.Qb3 N8c6 6.Bd2
8/8 00:00 25k 2,065k +0.90 3.d4 Nb6 4.h3 d6 5.Nf3 dxe5 6.Nxe5 Nc6 7.Nxc6
9/12 00:00 46k 1,981k +1.27 3.d4 d6 4.Nf3 Nd7 5.Bg5 dxe5 6.dxe5 Nc5 7.Qd4
10/13 00:00 108k 2,078k +1.01 3.d4 d6 4.Nf3 dxe5 5.Nxe5 Bf5 6.Bc4 c6 7.0-0 Nd7 8.Re1 Be6
11/16 00:00 231k 2,312k +1.02 3.d4 d6 4.Nf3 Nb6 5.Bd3 c5 6.0-0 cxd4 7.exd6 Qxd6 8.Bb5+ Bd7 9.Nxd4
12/16 00:00 321k 2,359k +0.83 3.Nf3 d6 4.d4 Bf5 5.Bd3 Bxd3 6.Qxd3 Nc6 7.exd6 Ncb4 8.Qd1 exd6 9.0-0
13/16 00:00 419k 2,279k +0.79 3.d4 d6 4.Nf3 Bf5 5.Bd3 Bxd3 6.Qxd3 Nc6 7.0-0 dxe5 8.Qb5 exd4 9.Nxd4
14/19 00:00 775k 1,973k +0.74 3.d4 d6 4.Nf3 dxe5 5.Nxe5 e6 6.Be2 f6 7.Nf3 c5 8.0-0 Nc6 9.Bb5 Bd7 10.Re1
15/22 00:00 1,211k 1,922k +0.74 3.d4 d6 4.Nf3 dxe5 5.Nxe5 e6 6.Be2 f6 7.Nf3 c5 8.0-0 Nc6 9.Bb5 Nc7 10.Nc3 Nxb5 11.Nxb5 Qd5 12.Nc3
16/24 00:01 2,969k 1,945k +0.64 3.c4 Nb6 4.d4 d6 5.Nf3 dxe5 6.Nxe5 Nc6 7.Nxc6 bxc6 8.c5 Nd5 9.Bc4 e6 10.Nc3 Be7 11.Ne4 Nf6
17/27 00:01 3,372k 1,956k +0.75 3.c4 Nb6 4.d4 d6 5.Nf3 dxe5 6.Nxe5 N6d7 7.Nf3 e5 8.dxe5 Bb4+ 9.Nbd2 Nc6 10.a3 Bxd2+ 11.Bxd2 0-0 12.Be2 Ncxe5 13.Nxe5 Nxe5 14.Bb4 Re8 15.0-0 Qxd1 16.Rfxd1
18/30 00:02 4,174k 1,909k +0.68 3.c4 Nb6 4.d4 d6 5.Nf3 dxe5 6.Nxe5 Nc6 7.Nxc6 bxc6 8.c5 Nd5 9.Bc4 e5 10.0-0 Be7 11.Nd2 exd4 12.Ne4 0-0 13.Qxd4
19/30- 00:03 5,883k 1,941k +0.61 3.c4 Nb6
19/30 00:03 6,435k 1,968k +0.59 3.c4 Nb6 4.d4 d6 5.Nf3 dxe5 6.Nxe5 N6d7 7.Nf3 e5 8.Nc3 exd4 9.Nxd4 Be7 10.Be2 0-0 11.0-0 Bf6 12.Be3 Nc5 13.Rc1
20/30+ 00:03 6,567k 1,971k +0.66 3.c4
20/30 00:03 7,118k 1,990k +0.60 3.c4 Nb6 4.d4 d6 5.Nf3 dxe5 6.Nxe5 N6d7 7.Nxd7 Nxd7 8.Be2 e5 9.Nc3 Bd6 10.Be3 exd4 11.Qxd4 0-0 12.Rd1 Be5 13.Qd2
21/34+ 00:05 11,077k 2,145k +0.67 3.c4
21/34 00:05 11,430k 2,153k +0.61 3.c4 Nb6 4.d4 d6 5.f4 Bf5 6.Nc3 e6 7.Nf3 Be7 8.Be2 0-0 9.0-0 c5 10.d5 Na6 11.Ne1 Bg6 12.Nc2 Bxc2 13.Qxc2 Nb4 14.Qe4 dxe5 15.dxe6 Qd4+ 16.Be3 Qxe4
22/34- 00:06 15,265k 2,241k +0.54 3.c4 Nb6
22/34 00:07 16,716k 2,257k +0.58 3.c4 Nb6 4.d4 d6 5.Nf3 dxe5 6.Nxe5 N6d7 7.Nxd7 Nxd7 8.Nc3 e5 9.dxe5 Bb4 10.f4 0-0 11.a3 Bxc3+ 12.bxc3 Qe7 13.Be2 Nc5 14.0-0 Bf5 15.Be3
23/34+ 00:09 23,384k 2,349k +0.65 3.d4
23/35- 00:11 26,536k 2,373k +0.58 3.d4 d6
23/35- 00:13 31,927k 2,401k +0.47 3.d4 d6
23/35+ 00:15 37,791k 2,424k +0.56 3.d4
23/35 00:15 38,697k 2,429k +0.62 3.d4 d6 4.c4 Nb6 5.exd6 exd6 6.Nc3 Be7 7.Bd3 Nc6 8.Nge2 0-0 9.0-0 Bf6 10.Be3 Nb4 11.b3 Nxd3 12.Qxd3 g6 13.d5 Be5 14.Bd4 Bf5 15.Qd2
24/35- 00:19 46,904k 2,453k +0.55 3.d4 d6
24/35- 00:20 50,270k 2,461k +0.47 3.d4 d6
24/38+ 00:24 59,638k 2,452k +0.55 3.d4
24/38 00:24 61,221k 2,449k +0.63 3.d4 d6 4.c4 Nb6 5.exd6 exd6 6.Nc3 Be7 7.Bd3 Nc6 8.Nge2 0-0 9.b3 Bg4 10.h3 Bxe2 11.Nxe2 Re8 12.0-0 Bg5 13.Bb2 Nb4 14.Bb1 g6 15.Re1
25/38- 00:25 63,301k 2,444k +0.56 3.d4 d6
25/38 00:28 70,139k 2,424k +0.53 3.d4 d6 4.c4 Nb6 5.exd6 exd6 6.Nc3 Be7 7.Bd3 0-0 8.Nge2 Nc6 9.0-0 Bf6 10.b3 Bg4 11.Be3 Bxd4 12.Bxh7+ Kxh7 13.Qd3+ Kh8 14.Bxd4 Qh4 15.f3 Bd7 16.Bf2 Qh7 17.Qd2 Rae8 18.h3 Kg8
26/39+ 00:31 75,803k 2,433k +0.60 3.d4
26/39+ 00:31 77,361k 2,438k +0.68 3.d4
26/41- 00:34 85,343k 2,449k +0.60 3.d4 d6
26/41 00:52 122,943k 2,362k +0.63 3.c4 Nb6 4.d4 d6 5.Nf3 dxe5 6.Nxe5 Nc6 7.Nxc6 bxc6 8.c5 Nd5 9.Bc4 e5 10.Nc3 Be6 11.Qa4 Qd7 12.0-0 Nxc3 13.bxc3 exd4 14.Bxe6 fxe6 15.cxd4 Be7 16.Rb1 0-0 17.Rd1 Bf6
27/41- 00:55 130,436k 2,356k +0.56 3.c4 Nb6
27/41- 00:59 138,051k 2,325k +0.48 3.c4 Nb6
27/41 01:14 171,163k 2,311k +0.44 3.c4 Nb6 4.Nf3 d6 5.exd6 exd6 6.d4 Be7 7.Bd3 0-0 8.0-0 Nc6 9.Nc3 Bf6 10.Be3 Re8 11.b3 Nb4 12.h3 c5 13.Rc1 d5 14.dxc5 dxc4 15.Bxh7+ Kxh7 16.cxb6 Qxd1 17.Rfxd1 cxb3 18.axb3 axb6 19.Bxb6
28/41+ 01:25 194,338k 2,285k +0.51 3.c4
28/41- 01:37 219,313k 2,243k +0.44 3.c4 Nb6
28/41+ 01:45 236,895k 2,247k +0.51 3.c4
28/41 01:50 248,470k 2,257k +0.46 3.d4 d6 4.c4 Nb6 5.exd6 exd6 6.Nf3 Be7 7.Bd3 0-0 8.0-0 Nc6 9.Nc3 Bg4 10.h3 Bxf3 11.Qxf3 Bf6 12.Be3 g6 13.Qf4 Nb4 14.Be2 Nc2 15.Rad1 Nxe3 16.fxe3 Bg7 17.e4 Qe7 18.Nd5 Nxd5 19.cxd5 c6 20.dxc6 bxc6
29/45 02:09 291,041k 2,256k +0.44 3.d4 d6 4.c4 Nb6 5.exd6 exd6 6.Nc3 Nc6 7.Nf3 Be7 8.d5 Ne5 9.Nxe5 dxe5 10.Be3 0-0 11.Bd3 f5 12.0-0 e4 13.Be2 Bb4 14.Qd2 c6 15.dxc6 Qxd2 16.Bxd2 bxc6 17.Nxe4 Be7 18.Bg5 Re8 19.Bxe7 Rxe7
30/45- 02:25 329,222k 2,268k +0.37 3.d4 d6
30/45 02:46 379,781k 2,279k +0.36 3.c4 Nb6 4.Nc3 d6 5.exd6 exd6 6.d4 Nc6 7.Nf3 Be7 8.d5 Ne5 9.Nxe5 dxe5 10.Be3 c6 11.d6 Bg5 12.Qd2 Bxe3 13.Qxe3 0-0 14.Be2 Be6 15.c5 Nd7 16.0-0 f5 17.f4 e4 18.a3
31/45+ 02:58 408,660k 2,286k +0.43 3.c4
31/45 03:06 427,463k 2,293k +0.37 3.c4 Nb6 4.Nc3 d6 5.exd6 exd6 6.d4 Nc6 7.Nf3 Be7 8.d5 Ne5 9.Nxe5 dxe5 10.Be3 0-0 11.Bd3 f5 12.0-0 e4 13.Be2 Nd7 14.Qd2 Bd6 15.Bg5 Be7 16.Bf4 Bd6 17.Rad1 Qe7 18.Nb5 Bxf4 19.Qxf4 Ne5 20.d6 cxd6 21.Nxd6
32/45+ 03:19 456,139k 2,290k +0.44 3.c4
32/45- 03:46 521,131k 2,297k +0.37 3.c4 Nb6
32/45- 05:02 702,757k 2,322k +0.26 3.c4 Nb6
32/45+ 05:38 788,559k 2,327k +0.35 3.c4
32/45 05:46 806,622k 2,329k +0.34 3.c4 Nb6 4.d4 d6 5.exd6 exd6 6.Nc3 Nc6 7.Nf3 Be7 8.d5 Ne5 9.Nxe5 dxe5 10.Be3 0-0 11.Be2 Bf5 12.0-0 Nd7 13.Qd2 Bg6 14.Rad1 h6 15.Nb5 Nf6 16.Bf3 Ne4 17.Bxe4 Bxe4 18.Qe2 Bg6 19.Bxa7 f6 20.Be3 Rxa2 21.Nc3
33/45+ 05:56 830,184k 2,329k +0.41 3.c4
33/45- 06:17 879,359k 2,327k +0.34 3.c4 Nb6
33/45+ 06:28 904,416k 2,329k +0.41 3.c4
33/45+ 06:36 923,145k 2,330k +0.57 3.c4
33/52 07:17 1,015,719k 2,324k +0.48 3.c4 Nb6 4.d4 d6 5.exd6 exd6 6.Nc3 Nc6 7.Nf3 Be7 8.d5 Ne5 9.Nxe5 dxe5 10.Be3 0-0 11.Bd3 f5 12.0-0 f4 13.Bd2 g6 14.Qe2 Bf5 15.Bxf5 Rxf5 16.Ne4 Na4 17.Rad1 Nxb2 18.Rb1 Na4 19.Rxb7 Nc5 20.Rb5 Nxe4 21.Qxe4 a6 22.Ra5 Qf8 23.g4 fxg3 24.hxg3 Bc5 25.Be3
34/52 08:39 1,178,578k 2,267k +0.53 3.c4 Nb6 4.d4 d6 5.exd6 exd6 6.Nc3 Nc6 7.Nf3 Be7 8.d5 Ne5 9.Nxe5 dxe5 10.Be3 0-0 11.Bd3 f5 12.0-0 f4 13.Bd2 g6 14.Qe2 Bf5 15.Rfe1 Nd7 16.Bxf5 Rxf5 17.Rad1 a5 18.Ne4 Nc5 19.g4 Rf8 20.Bc3 f3 21.Qe3 Nxe4
35/52- 09:00 1,214,737k 2,245k +0.45 3.c4 Nb6
35/52- 09:29 1,280,315k 2,249k +0.38 3.c4 Nb6
35/52 12:12 1,608,660k 2,195k +0.31 3.c4 Nb6 4.Nc3 d6 5.exd6 exd6 6.d4 Nc6 7.Nf3 Be7 8.d5 Ne5 9.Nxe5 dxe5 10.Be3 0-0 11.Bd3 f5 12.0-0 e4 13.Be2 Bb4 14.Qc1 Nd7 15.Nb5 Ne5 16.Bd4 a6 17.Nc3 Ng6 18.a3 Bxc3 19.Qxc3 Qe7 20.c5 f4 21.Bc4 f3
36/52+ 18:03 2,323,962k 2,145k +0.38 3.d4
36/52 20:15 2,613,997k 2,151k +0.34 3.d4 d6 4.exd6 exd6 5.Nf3 Nf6 6.Bd3 Be7 7.0-0 0-0 8.Re1 c5 9.c4 cxd4 10.Nxd4 Nc6 11.Nxc6 bxc6 12.Nc3 d5 13.cxd5 Nxd5 14.Nxd5 cxd5 15.Bf4 Be6 16.Rc1 Bf6 17.Rc7 Bxb2 18.Rb7 Bc3 19.Re3 Qf6 20.Bg3 h6 21.Bc2 Rfd8
37/52- 21:57 2,836,856k 2,153k +0.27 3.d4 d6
37/52+ 23:27 3,035,580k 2,157k +0.34 3.d4
37/52- 28:29 3,692,549k 2,160k +0.27 3.d4 d6
37/55 41:18 5,285,724k 2,133k +0.23 3.d4 d6 4.exd6 cxd6 5.c3 Bf5 6.Qf3 e6 7.Bc4 Nf6 8.Qxb7 Be4 9.Bb5+ Nbd7 10.Bc6 d5 11.Qxa8 Qxa8 12.Bxa8 Bxg2 13.Bf4 Bxh1 14.Nd2 g5 15.Bxg5 Rg8 16.Ndf3 Ne4 17.Bf4 Ke7 18.Bc6 Ndf6 19.Ke2 Ng4 20.Nd2 Bh6 21.Bxh6 Nxh6 22.Nxe4 Bxe4 23.Bb5 Rg2 24.Nf3
38/55+ 52:40 6,943,884k 2,197k +0.31 3.c4
38/55+ 53:03 7,001,278k 2,199k +0.38 3.c4
38/55+ 53:58 7,140,648k 2,205k +0.49 3.c4
38/60- 55:14 7,311,586k 2,206k +0.39 3.c4 Nb6
38/60 1:00:17 7,902,327k 2,185k +0.24 3.c4 Nb6 4.d4 d6 5.f4 Bf5 6.Nc3 e6 7.Nf3 dxe5 8.fxe5 Bb4 9.Bd3 c5 10.Bxf5 cxd4 11.Nxd4 exf5 12.Nxf5 Qxd1+ 13.Kxd1 0-0 14.Nd5 Rd8 15.Kc2 Nc6 16.Rd1 Nxc4 17.Bg5 Re8 18.Nxb4 Nxb4+ 19.Kb3 Rxe5 20.Rf1 Re4 21.Kxb4 Nd2+ 22.Kc3 Nxf1 23.Rxf1 Re2 24.Nd6 h6 25.Bf4 Rxg2 26.Kb3 Rd8 27.Nxb7
39/60+ 1:01:59 8,105,409k 2,179k +0.31 3.c4
39/60- 1:05:48 8,627,874k 2,185k +0.24 3.c4 Nb6
39/63+ 1:07:04 8,798,180k 2,186k +0.31 3.c4
39/63- 1:14:55 9,873,304k 2,196k +0.22 3.c4 Nb6
39/63+ 1:16:26 10,093,986k 2,201k +0.34 3.c4
39/63 1:17:58 10,314,205k 2,204k +0.27 3.c4 Nb6 4.d4 d6 5.f4 Bf5 6.Nc3 e6 7.Nf3 dxe5 8.fxe5 Nc6 9.Be3 Bb4 10.Be2 Na5 11.c5 Nd5 12.Bd2 Ne7 13.0-0 0-0 14.Be3 Bxc3 15.bxc3 b6 16.cxb6 axb6 17.Qc1 Be4 18.Bf2 Qd5 19.c4 Qc6 20.Nh4 h6 21.h3 Ng6 22.Nf3 Qb7 23.Qe3
40/63+ 1:27:16 11,460,205k 2,189k +0.34 3.d4
40/63- 1:29:58 11,851,333k 2,195k +0.27 3.d4 d6
40/63 1:37:04 12,883,761k 2,212k +0.33 3.d4 d6 4.Nf3 g6 5.Bc4 Nb6 6.Bb3 Bg7 7.0-0 0-0 8.exd6 cxd6 9.c3 Nc6 10.Re1 Bf5 11.d5 Na5 12.Na3 Rc8 13.Be3 Re8 14.Bd4 Bxd4 15.Qxd4 Qc7 16.Nb5 Qd8 17.Qh4 Bd7 18.Qh6 f6 19.Nbd4 Nxb3 20.axb3 Nxd5 21.Ne6 Bxe6 22.Rxe6 Qb6 23.Nd4 Qc5
try deeper the 4 pawns attack line 1.e4 Nf6 2.e5 Nd5 3.d4 d6 4.c4 Nb6 5.f4:

[d]rnbqkb1r/ppp1pppp/1n1p4/4P3/2PP1P2/8/PP4PP/RNBQKBNR b KQkq f3 0 5

should get to around +200cps white advantage some 20 moves later. (possibly too long to play for SF exactly)

Lyudmil Tsvetkov
Posts: 6037
Joined: Tue Jun 12, 2012 10:41 am

Re: Chess Genius 3 Beat Kasparov on a Pentium 90?

Post by Lyudmil Tsvetkov » Thu Jun 01, 2017 7:22 am

Ras wrote:With your game, were the time controls comparable? I mean, the effective time relative to the computing speed?

It's difficult for me to think of the whole opening as weak. Not only because of Alekhine himself, but also other masters: https://en.wikipedia.org/wiki/Alekhine's_Defence#Use

However, the opening requires understanding the idea behind it (which is easy for a human) and also implementing it the right way on the board (which is where I struggle ^^), and I can well imagine that the CG3 engine doesn't even know what to do.

With the CT800, I'm seeing something like that with Dutch as Black, which is (nearly) always a desaster. Even against clearly inferior opponents, a draw is already lucky, and I've never seen a win. The reason isn't that Dutch is that bad, it's that the engine is not able to cope with the position. I relegated Dutch to passive kowledge for that reason.

Tested against a dedicated unit 200 ELO below the CT800 on average, the odds with Dutch as Black are so bad that the inferior machine actually has at least a plus of 200 ELO from the winning percentage. That's well in the order of magnitude of 500 ELO.

It's not that the opening book can do wonders here - but it can at least avoid exposing glaring weaknesses on a silver plate.
the Alekhine is a bust.

evaluation-wise, it is indeed very bad to place your knight to multiple pawn attacks, gaining a couple of tempos. plus, white yielding the white pieces. simply does not make sense.

if you are lost eval-wise, most probably you are also lost search-wise, no matter how long it takes to get to the decisive outcome.

about the Dutch, you are completely right, engines simply fail to undertsand the springs of such closed or semi-closed play. in order to draw, it is sufficient for black to play the Stonewall, place all blck pawns on dark squares, defending each other, and wait. Certainly, no breaking through.

thus, I already know the Dutch provides a forced draw for black, hence score it with 0.0.

On the other hand, 1.d4 d5 still gives white quite some advantage, until some 40 moves later or so, and I really do not know how the game should end, at least no forced outcome.

thus, the Dutch is much better than the Queen's gambit for black, although engines will need couple decades more, until they grasp it.

Ras
Posts: 1130
Joined: Tue Aug 30, 2016 6:19 pm
Contact:

Re: Chess Genius 3 Beat Kasparov on a Pentium 90?

Post by Ras » Thu Jun 01, 2017 4:20 pm

Lyudmil Tsvetkov wrote:simply does not make sense.
The idea is a classic Sun Tsu strategy: "the way to cover a thousand miles and then win a battle is to make the other side do the march." Space is good, usually, but it can also become a liability.

The drawback is that this is a passive-aggressive strategy, and if White just does not grab "too much" space, Black ends up with a cramped position.

In the 2016 dedicated chess computer tournament here, my CT800 chose Alekhine against Fidelity V11 60 MHz - and won.
it is sufficient for black to play the Stonewall, place all blck pawns on dark squares, defending each other, and wait. Certainly, no breaking through.
A more active strategy could be to initiate a kingside attack. That's precisely what the CT800 fails to do with Black Dutch; instead, it tries centre play. But then, the moved f pawn usually just has weakened the centre, that's why it loses.

User avatar
MikeGL
Posts: 862
Joined: Thu Sep 01, 2011 12:49 pm

Re: Chess Genius 3 Beat Kasparov on a Pentium 90?

Post by MikeGL » Thu Jun 01, 2017 6:44 pm

BrendanJNorman wrote:It's very interesting how computer chess progresses over time.

Chess Genius 3 playing in the 1994 Intel Grand Prix on a Pentium 90 defeated Gary Kasparov, and also beat GM Nikolic (a very strong GM) 2-0.
ChessGenius3 is a 1994 chess program. Fritz2.00 in 1992 already won multiple times against Kasparov at Koln Germany.

hammerklavier
Posts: 95
Joined: Mon Nov 28, 2011 9:17 am

Re: Chess Genius 3 Beat Kasparov on a Pentium 90?

Post by hammerklavier » Thu Jun 01, 2017 9:23 pm

ChessGenius3 is a 1994 chess program. Fritz2.00 in 1992 already won multiple times against Kasparov at Koln Germany.
Blitz games... only 5m /all game. Until version Fritz5, Genius was always stronger than Fritz.

BrendanJNorman
Posts: 1271
Joined: Sun Feb 07, 2016 11:43 pm
Full name: Brendan J Norman

Re: Chess Genius 3 Beat Kasparov on a Pentium 90?

Post by BrendanJNorman » Fri Jun 02, 2017 2:42 am

Uri Blass wrote:
BrendanJNorman wrote:
Guenther wrote:
BrendanJNorman wrote:
Vinvin wrote:+ In AubervilliersRapid Open 1993 and 1994, Chess Genius made a perf of 2530 and 2540 (hardware was probably 486-66 MHz) :
http://chessprogramming.wikispaces.com/ ... tions-1993
Must have been the magical opening book. :wink:
All he (Uri) wanted to say is, that 1.e4 Nf6 probably NEVER was played by any program for the Black side in serious Human vs. Comp games.
(Aljechin will be blocked for such a match for good reasons... the link below verifies it for Genius, which means he is right)

http://www.chessgames.com/player/chess_genius.html

Interesting though that they once allowed Genius to play a Pirc defense, which is also an opening C players judge as bad.
Hey Guenther,

I don't think the Alekhine (or any "inferior" opening) was blocked, otherwise CG3's opening book would be consistent across the game range.

Instead we see Pirc (known as a "come and get me" opening among strong players), Caro Kann (another "come and get me" opening for computers, but more solid than the Pirc), and 1...e5 (Petrov, Ruy Lopez and Two Knights).

I'd say that it's opening book is/was weighted to play 1...e5 more often, but throw in other openings based on some statistical distribution (as in, play the Caro Kann 40% of the time, Pirc 25% of the time or whatever), and that the only reason we don't see Alekhine's in the chessgames list is due to a relative lack of 1.e4 games.

As I alluded to before, the MAINLINE Alekhine (where white plays 3.d4 d6 4. c4 Nb6 5.f4 dxe5 6.fxe5 Nc6 etc isn't so bad for black - especially against a human (who might have more trouble holding the center together against a computer's tactical powers.

My playing an offbeat idea 3.Nc3 e6 4.d4 Bb4 5.Bd2 brought the computer out of its book and provoked a strategic mistake immediately since 5...Nxc3 6.bxc3 Be7 (6...Bf8 MIGHT be better) 7.Qg4 is already much better for white.

So would you say the opening book was still the problem if I'd played the mainline?
I think that 1...Nf6 is relatively a rare move so it is easier to find good sidelines after 1.e4 Nf6 to get the computer out of book.
You'r really clutching a straws now, buddy... :lol:

How was I to know that the Alekhine was coming in a random game while TESTING if the program would run on my Macbook?

An experienced enough player can find a decent sideline against ANYTHING...even Sicilian for example, 1. e4 c5 2.Nc3 Nc6 3.Nge2 (followed with g3, Bg2, 0-0, d3, h3, f4, g4, Qe1-h4 ideas) can also be effective against "booked up" opponents with little sense of strategy.

Thats just an example, but you get the point (hopefully).

Post Reply