AlphaZero SF game 10 Does Your Engine Find 29Qh3?

[Eelco de Groot]

The turning point that I found in the game was 32...Re8 in this position
rn2r1q1/p5k1/2p2bp1/1p4p1/2P5/4B1PQ/5PK1/3R3R w - - 0 33

Zero thinks Sf was busted before Re8 was played. At that point several moves were probably winning:

AlphaZero - Stockfish 8, 1 min / 1 move
[d]rn2r1q1/p5k1/2p2bp1/1p4p1/2P5/4B1PQ/5PK1/3R3R w - - 0 1

Analysis by Zero-the-Hero: (Depth 30)

1. +- (2.81): 33.Rd6 Be5 34.Rd3 Bf6 35.cxb5 cxb5 36.Qh6+ Kf7 37.Bxg5 Qh8 38.Rd6 Qxh6 39.Rxf6+ Kg7 40.Rxh6 Nd7 41.Rfxg6+ Kf7 42.Rd6 Nf8 43.Rd5 Rab8 44.Rf5+ Kg7 45.Rh4 Rb6 46.Bc1 Ng6 47.Bb2+ Kg8 48.Rg5 Kf8 49.Rhg4 Ree6 50.Bd4 Rbc6 51.Rxb5 Ke8 52.Rb8+ Kd7 53.Be3 a5 54.Rb5 Ke8 55.Rxa5 Rcd6 56.Rb4 Ne5 57.Rb7 Nc6 58.Ra8+ Rd8
2. +- (2.61): 33.Rd2 Re4 34.Rd6 Be7 35.Rd4 Re6 36.Rd2 bxc4 37.Bd4+ Kf7 38.Qg4 Nd7 39.Rh8 Qxh8 40.Bxh8 Nf6 41.Bxf6 Bxf6 42.Qxc4 Rd8 43.Rc2 a5 44.Qb3 Rd5 45.Rc1 g4 46.Rxc6 Rxc6 47.Qxd5+ Re6 48.Qxa5 Re5 49.Qc7+ Kg8 50.Qc8+ Kh7 51.Qxg4 Bg7 52.Qc4 Rh5 53.Qc6 Rf5 54.Qc7 Kh6 55.Qd7
3. +- (2.12): 33.Rd3 Re4 34.Rd6 Be7 35.Rd4 Re6 36.Rdd1 Bf6 37.Rd2 a5 38.Bd4 Kf7 39.Bxf6 Rxf6 40.Re2 Ra7 41.Qh6 Na6 42.Qh7+ Qxh7 43.Rxh7+ Kf8 44.Rxa7 Nb4 45.Rd2 Rf7 46.Ra8+ Ke7 47.cxb5 cxb5 48.Rxa5 Rf5 49.g4 Rc5 50.Rb2 Rc4 51.f3 Nd5 52.Raxb5 Ne3+ 53.Kh2
4. +- (2.07): 33.Bd4 Bxd4 34.Rxd4 Rd8 35.Qh6+ Kf7 36.Re4 Re8 37.Rxe8 Qxe8 38.Rd1 Qf8 39.Qh7+ Qg7 40.Qh3 Qf8 41.Re1 Kg7 42.Qe6 Qf6 43.Qc8 bxc4 44.Qb7+ Nd7 45.Qxa8 Ne5 46.Qxa7+ Nf7 47.Qc5 c3 48.Re8 Kh6 49.Re3 g4 50.Qxc3 Qxc3 51.Rxc3 Ne5 52.Rc5 Nd3 53.Rxc6 Kg5 54.Rb6 Kf5
5. +- (2.18): 33.Qh6+ Kf7 34.Bxg5 Qg7 35.Qh4 Re5 36.Bxf6 Qxf6 37.Qh7+ Qg7 38.Qh3 Na6 39.Rd4 Nc5 40.Rh4 Rae8 41.cxb5 cxb5 42.Rh7 Rh5 43.Rxg7+ Kxg7 44.Qg4 Rxh1 45.Kxh1 Ne6 46.Qe4 Re7 47.Qe5+ Kf7 48.Qxb5 Nc7 49.Qc4+ Ne6 50.Kg2 Rc7 51.Qb5 Kf6 52.f4 Rc5 53.Qa6 Rc7 54.Kh3 Kf7 55.Kg2 Rc2+ 56.Kh3
6. +/- (1.57): 33.Bc1 Re6 34.Rh2 Re4 35.Rd6 Kf7 36.cxb5 Be7 37.Rd3 g4 38.Qh6 Na6 39.bxa6 Qg7 40.Qxg7+ Kxg7 41.Rd7 Kf6 42.Rc7 c5 43.Rh7 Bd6 44.Bb2+ Ke6 45.Rcg7 Kd5 46.Rxg6 Re2 47.Bc1 Rf8 48.Be3 Re8 49.Rd7 R8xe3 50.Rgxd6+ Kc4 51.Rxa7 Ra2 52.Rg6 Re1

I must be missing something big or was it total trolling from "John" in the first post. Less than 1 minute analysis on one thread shows that 34...Rd8 was yet another terrible blunder from Stockfish 8. What did they feed the poor emaciated, dehydrated thing. Wodka? What am I missing

[d]rn2r1q1/p5k1/2p3p1/1p4p1/2PR4/6PQ/5PK1/7R b - -

Engine: Kaissa (1 thread, Codebase 22-09-2017, 512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

25 0:42 0.00 34...bxc4 35.Qh6+ Kf6 36.Rhd1 Na6
37.Rf4+ gxf4 38.Qxf4+ Ke7 39.Qd6+ Kf7
40.Qf4+ Ke7 (96.358.446) 2260

25 0:42 0.00 34...a5 35.Qh6+ Kf6 36.Rhd1 Ra7
37.Rd6+ Re6 38.Rxe6+ Qxe6 39.Qh8+ Rg7
40.Qf8+ Rf7 41.Qh8+ Rg7 (96.358.446) 2260

24 0:42 +0.62 34...Re7 35.Rd3 Nd7 36.Rxd7 Kf6
37.Rd6+ Re6 38.Rhd1 Rae8 39.Qg4 Qf8
40.Rd7 R6e7 41.R1d6+ Kg7 42.cxb5 cxb5
43.Qxg5 Qf7 44.Qxb5 Rxd7 45.Rxd7 Re7
46.Qb2+ Kg8 47.Rd8+ Re8 (96.358.446) 2260

24 0:42 +1.24 34...Kf7 35.Qg4 Qg7 36.Rd6 a5
37.Qxg5 Re5 38.Qg4 Ra7 39.Rd8 Rae7
40.Rxb8 Re8 41.Rb7+ R8e7 42.Rxe7+ Kxe7
43.Qc8 Kf7 44.Qc7+ Kf8 45.Qxc6 bxc4
46.Rh4 c3 47.Qc8+ Kf7 (96.358.446) 2260

24 0:42 +1.27 34...Rd8 35.Rxd8 Qxd8 36.Qe6 Nd7
37.Rd1 Nc5 38.Rxd8 Nxe6 39.Rxa8 Kf6
40.cxb5 cxb5 41.Rxa7 Ke5 42.Rb7 Nd4
43.Kf1 Kd5 44.Ke1 b4 45.Kd2 Kc4
46.Rb6 Nf3+ 47.Ke3 Ne5 (96.358.446) 2260

24 0:42 +1.30 34...Re6 35.Rd8 Qxd8 36.Qxe6 Nd7
37.Rd1 Nc5 38.Rxd8 Nxe6 39.Rxa8 Kf6
40.cxb5 cxb5 41.Rxa7 Nd4 42.Kf1 Kf5
43.Ke1 Ke4 44.Ra6 b4 45.Kd2 Nb3+
46.Kd1 Nc5 47.Rxg6 b3 (96.358.446) 2260

24 0:42 +1.69 34...a6 35.Rd3 Ra7 36.Rf3 Qh8 37.Qg4 Qxh1+
38.Kxh1 Rd7 39.Qxg5 Rf7 40.Rf4 Nd7
41.Rxf7+ Kxf7 42.Qf4+ Nf6 43.Qc7+ Re7
44.Qxc6 bxc4 45.Qxc4+ Re6 46.Kg2 Ke7
47.Qc7+ Nd7 (96.358.446) 2260

[carldaman]

Reposted with diagram included:

[d]rn3r2/p3b1k1/2p3p1/1p4p1/8/q1P1B1P1/P4P2/3RR1KQ w - - 0 29

Zero-the-Hero:
1/1 00:00 158 158k +6.17 Qh1-e4 Qa3xa2
2/2 00:00 334 334k +14.32 Qh1-e4 Qa3xc3 Qe4xe7+ Kg7-g8 Be3xg5
3/4 00:00 527 527k +11.36 f2-f4 Qa3xc3 f4xg5
4/6 00:00 1k 1,302k +9.69 Be3-d4+ Be7-f6 Bd4xf6+ Kg7xf6 Qh1-e4 Kf6-g7
6/8 00:00 3k 1,473k +10.60 Be3-d4+ Rf8-f6 Re1-e6 Nb8-d7 Rd1-e1 Qa3xa2 Re6xe7+ Qa2-f7
7/10 00:00 8k 1,944k +8.35 f2-f4 Qa3xc3 f4xg5 Be7xg5 Be3-d4+ Qc3xd4+ Rd1xd4 c6-c5
8/12 00:00 20k 2,171k +6.27 Be3-d4+ Be7-f6 Re1-e3 g5-g4 Qh1-h4 Bf6xd4 c3xd4 Rf8-h8 Re3-e7+ Qa3xe7

I have never seen such nonsense before.

What do you mean?

That was the Arena 3.5 output. It is what it is. I can post CB output, too, to compare.

I don't doubt you nor Arena but the (origin of the) engine and what kind of settings were used to produce such ridiculous scores in the early plies.

It is what it is. I'm showing it's possible to produce the same moves that AlphaZero did, using a strong alpha-beta engine. In the OpenTal thread there is an example of how it typically plays - very attractive attacking chess.

And ZTH default would crush ZTH with the settings you used here, not true?

ZtH with its [roughly] current settings has *predated* (by more than a year) the AlphaZero news, so I'm not tweaking settings just to replicate the AZ moves. I'm not claiming it's the strongest engine in the world. It's certainly stronger than the best humans, I can tell you that much.

By the way, Ed, I really liked your approach with Prodeo 2.2, using different cores to run separate searches. I was hoping more programmers would follow suit and improve on that approach, but it didn't quite happen. There seems to be an entrenched resistance to innovation in the whole chess programming field and only the AZ surprise may have awakened some folks.

[carldaman]

The turning point that I found in the game was 32...Re8 in this position
rn2r1q1/p5k1/2p2bp1/1p4p1/2P5/4B1PQ/5PK1/3R3R w - - 0 33

Zero thinks Sf was busted before Re8 was played. At that point several moves were probably winning:

AlphaZero - Stockfish 8, 1 min / 1 move
[d]rn2r1q1/p5k1/2p2bp1/1p4p1/2P5/4B1PQ/5PK1/3R3R w - - 0 1

Analysis by Zero-the-Hero: (Depth 30)

1. +- (2.81): 33.Rd6 Be5 34.Rd3 Bf6 35.cxb5 cxb5 36.Qh6+ Kf7 37.Bxg5 Qh8 38.Rd6 Qxh6 39.Rxf6+ Kg7 40.Rxh6 Nd7 41.Rfxg6+ Kf7 42.Rd6 Nf8 43.Rd5 Rab8 44.Rf5+ Kg7 45.Rh4 Rb6 46.Bc1 Ng6 47.Bb2+ Kg8 48.Rg5 Kf8 49.Rhg4 Ree6 50.Bd4 Rbc6 51.Rxb5 Ke8 52.Rb8+ Kd7 53.Be3 a5 54.Rb5 Ke8 55.Rxa5 Rcd6 56.Rb4 Ne5 57.Rb7 Nc6 58.Ra8+ Rd8
2. +- (2.61): 33.Rd2 Re4 34.Rd6 Be7 35.Rd4 Re6 36.Rd2 bxc4 37.Bd4+ Kf7 38.Qg4 Nd7 39.Rh8 Qxh8 40.Bxh8 Nf6 41.Bxf6 Bxf6 42.Qxc4 Rd8 43.Rc2 a5 44.Qb3 Rd5 45.Rc1 g4 46.Rxc6 Rxc6 47.Qxd5+ Re6 48.Qxa5 Re5 49.Qc7+ Kg8 50.Qc8+ Kh7 51.Qxg4 Bg7 52.Qc4 Rh5 53.Qc6 Rf5 54.Qc7 Kh6 55.Qd7
3. +- (2.12): 33.Rd3 Re4 34.Rd6 Be7 35.Rd4 Re6 36.Rdd1 Bf6 37.Rd2 a5 38.Bd4 Kf7 39.Bxf6 Rxf6 40.Re2 Ra7 41.Qh6 Na6 42.Qh7+ Qxh7 43.Rxh7+ Kf8 44.Rxa7 Nb4 45.Rd2 Rf7 46.Ra8+ Ke7 47.cxb5 cxb5 48.Rxa5 Rf5 49.g4 Rc5 50.Rb2 Rc4 51.f3 Nd5 52.Raxb5 Ne3+ 53.Kh2
4. +- (2.07): 33.Bd4 Bxd4 34.Rxd4 Rd8 35.Qh6+ Kf7 36.Re4 Re8 37.Rxe8 Qxe8 38.Rd1 Qf8 39.Qh7+ Qg7 40.Qh3 Qf8 41.Re1 Kg7 42.Qe6 Qf6 43.Qc8 bxc4 44.Qb7+ Nd7 45.Qxa8 Ne5 46.Qxa7+ Nf7 47.Qc5 c3 48.Re8 Kh6 49.Re3 g4 50.Qxc3 Qxc3 51.Rxc3 Ne5 52.Rc5 Nd3 53.Rxc6 Kg5 54.Rb6 Kf5
5. +- (2.18): 33.Qh6+ Kf7 34.Bxg5 Qg7 35.Qh4 Re5 36.Bxf6 Qxf6 37.Qh7+ Qg7 38.Qh3 Na6 39.Rd4 Nc5 40.Rh4 Rae8 41.cxb5 cxb5 42.Rh7 Rh5 43.Rxg7+ Kxg7 44.Qg4 Rxh1 45.Kxh1 Ne6 46.Qe4 Re7 47.Qe5+ Kf7 48.Qxb5 Nc7 49.Qc4+ Ne6 50.Kg2 Rc7 51.Qb5 Kf6 52.f4 Rc5 53.Qa6 Rc7 54.Kh3 Kf7 55.Kg2 Rc2+ 56.Kh3
6. +/- (1.57): 33.Bc1 Re6 34.Rh2 Re4 35.Rd6 Kf7 36.cxb5 Be7 37.Rd3 g4 38.Qh6 Na6 39.bxa6 Qg7 40.Qxg7+ Kxg7 41.Rd7 Kf6 42.Rc7 c5 43.Rh7 Bd6 44.Bb2+ Ke6 45.Rcg7 Kd5 46.Rxg6 Re2 47.Bc1 Rf8 48.Be3 Re8 49.Rd7 R8xe3 50.Rgxd6+ Kc4 51.Rxa7 Ra2 52.Rg6 Re1

I must be missing something big or was it total trolling from "John" in the first post. Less than 1 minute analysis on one thread shows that 34...Rd8 was yet another terrible blunder from Stockfish 8. What did they feed the poor emaciated, dehydrated thing. Wodka? What am I missing

[d]rn2r1q1/p5k1/2p3p1/1p4p1/2PR4/6PQ/5PK1/7R b - -

Engine: Kaissa (1 thread, Codebase 22-09-2017, 512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

25 0:42 0.00 34...bxc4 35.Qh6+ Kf6 36.Rhd1 Na6
37.Rf4+ gxf4 38.Qxf4+ Ke7 39.Qd6+ Kf7
40.Qf4+ Ke7 (96.358.446) 2260

25 0:42 0.00 34...a5 35.Qh6+ Kf6 36.Rhd1 Ra7
37.Rd6+ Re6 38.Rxe6+ Qxe6 39.Qh8+ Rg7
40.Qf8+ Rf7 41.Qh8+ Rg7 (96.358.446) 2260

24 0:42 +0.62 34...Re7 35.Rd3 Nd7 36.Rxd7 Kf6
37.Rd6+ Re6 38.Rhd1 Rae8 39.Qg4 Qf8
40.Rd7 R6e7 41.R1d6+ Kg7 42.cxb5 cxb5
43.Qxg5 Qf7 44.Qxb5 Rxd7 45.Rxd7 Re7
46.Qb2+ Kg8 47.Rd8+ Re8 (96.358.446) 2260

24 0:42 +1.24 34...Kf7 35.Qg4 Qg7 36.Rd6 a5
37.Qxg5 Re5 38.Qg4 Ra7 39.Rd8 Rae7
40.Rxb8 Re8 41.Rb7+ R8e7 42.Rxe7+ Kxe7
43.Qc8 Kf7 44.Qc7+ Kf8 45.Qxc6 bxc4
46.Rh4 c3 47.Qc8+ Kf7 (96.358.446) 2260

24 0:42 +1.27 34...Rd8 35.Rxd8 Qxd8 36.Qe6 Nd7
37.Rd1 Nc5 38.Rxd8 Nxe6 39.Rxa8 Kf6
40.cxb5 cxb5 41.Rxa7 Ke5 42.Rb7 Nd4
43.Kf1 Kd5 44.Ke1 b4 45.Kd2 Kc4
46.Rb6 Nf3+ 47.Ke3 Ne5 (96.358.446) 2260

24 0:42 +1.30 34...Re6 35.Rd8 Qxd8 36.Qxe6 Nd7
37.Rd1 Nc5 38.Rxd8 Nxe6 39.Rxa8 Kf6
40.cxb5 cxb5 41.Rxa7 Nd4 42.Kf1 Kf5
43.Ke1 Ke4 44.Ra6 b4 45.Kd2 Nb3+
46.Kd1 Nc5 47.Rxg6 b3 (96.358.446) 2260

24 0:42 +1.69 34...a6 35.Rd3 Ra7 36.Rf3 Qh8 37.Qg4 Qxh1+
38.Kxh1 Rd7 39.Qxg5 Rf7 40.Rf4 Nd7
41.Rxf7+ Kxf7 42.Qf4+ Nf6 43.Qc7+ Re7
44.Qxc6 bxc4 45.Qxc4+ Re6 46.Kg2 Ke7
47.Qc7+ Nd7 (96.358.446) 2260

There were quite a few strange moves played by SF, from what I've seen. People have very right to question the match conditions that appeared to cripple SF in tangible ways.

[F. Bluemers]

Hi Eelco
Just curious,
[d] rn2r1q1/p5k1/2p3p1/1p4p1/2PR4/6PQ/5PK1/7R b - -
after 34 ... bc4
what does kaissa think of the black position after 35 g4? (instead of (Qh6+)
Best Fonzy

[F. Bluemers]

Also after 34..a6 kaissa found the idea 35. Rd3 and Rf3 for white.
But not after the move 34..a5 ?!
I guess it should still work?
Best
Fonzy

[carldaman]

You could also check out this interesting post by Ulysses and my reply to him:

http://www.talkchess.com/forum/viewtopi ... 17&t=66042

The problem with speculative engines is that they can find many amazing resources, but they can suddenly stumble on the very next move, and play an unsound sac that loses. (Like the great dribbler who loses the ball after one dribble too many.)

A second brain, running on separate cores, could then provide a second opinion and rule out such unsound moves while retaining the better ones.

The thinking process of the second brain could be paraphrased like this:

"I see this is an unexpected move that I pruned away, but it sure seems to work - now go ahead and play that move!"

"Here's another sac I stupidly pruned away - but it can't be refuted! I'm getting to like this new paradigm! Play the move, fast!"

"Wow, another sac, where do these moves come from? But wait, hold your horses, there's a problem: I see a definite refutation! Sorry, but we can't play this move, or we'll lose. Let's play my stodgy old move instead and live to play more moves, ok?"

Now, Ed, I know you're a great expert in these matters. Maybe you can tell us why this sort of thing is not in wide use, because I don't get it.

[Eelco de Groot]

Also after 34..a6 kaissa found the idea 35. Rd3 and Rf3 for white.
But not after the move 34..a5 ?!
I guess it should still work?
Best
Fonzy

Hi Fonz,

Yes, you saw that absolutely right. Kaissa finds out that 34...a5 also does not work, just a little bit later. I guess searchluck that it finds them for 34...a6 a bit earlier. Kaissa sees all these moves, but just not the incredible 34...bxc4 35. g4 {!!} Not until much later. And so does not know 29. Qh3 is winning.

35. g4 is after 34... bxc4 the only winning move, as far as Kaissa can see here anyway, but it sees not enough. And it is a move that is totally crushing Black who is at this moment three {three!} pawns, AND a full knight up in material. And still it is lost, without Kaissa seeing it or I why. That is incredible...


[d]rn2r1q1/p5k1/2p3p1/6p1/2pR4/6PQ/5PK1/7R w - -

Engine: Kaissa (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

36 52:37 +5.95 35.g4 Qe6 36.Qh7+ Kf6 37.Rh3 Qe7
38.Qh6 Rd8 39.Rf3+ Ke5 40.Re3+ Kxd4
41.Rxe7 Kc5 42.Rb7 Nd7 43.Qxg5+ Kd4
44.Qd2+ Kc5 45.Rxd7 Rxd7 46.Qxd7 a5
47.Qb7 Re8 48.Qa6 (7.860.200.778) 2489

36 52:37 0.00 35.Qh6+ Kf6 36.Rd6+ Re6 37.Rxe6+ Kxe6
38.Rd1 Qf7 39.Qh3+ Qf5 40.Qh8 Qe4+
41.Kg1 Qe5 42.Qc8+ Kf6 43.Qf8+ Ke6
44.Qc8+ (7.860.200.778) 2489

36 52:37 0.00 35.Rd6 Rd8 36.Qh6+ Kf7 37.Rhd1 Rxd6
38.Rxd6 Ke8 39.Qxg5 Nd7 40.Rxc6 Qf7
41.Rxg6 Rc8 42.Rg8+ Nf8 43.Rg7 Qe6
44.Qh5+ Kd8 45.Qg5+ Ke8 (7.860.200.778) 2489

36 52:37 0.00 35.Rdd1 Re7 36.Rd6 Nd7 37.Rxd7 Rae8
38.Qg4 c5 39.Rd6 Re6 40.Rd5 Re5
41.Rxe5 Rxe5 42.Qd7+ Kf6 43.Qd6+ Re6
44.Qxc5 Qa8+ 45.Kh2 a5 46.Rd1 a4
47.Rd7 Qh8+ 48.Kg2 (7.860.200.778) 2489

36 52:37 -0.35 35.Qg4 c5 36.Rdd1 Re5 37.Qf3 Nc6
38.Qxc6 Rae8 39.Rd7+ R8e7 40.Rxe7+ Rxe7
41.Qxc5 Qa8+ 42.Kh2 Qe4 43.Rc1 a5
44.Rxc4 Qe5 45.Qb6 Rf7 46.Ra4 Rf5
47.Kg1 Qe1+ 48.Kg2 (7.860.200.778) 2489

36 52:37 -1.01 35.Rd2 Rd8 36.Rxd8 Qxd8 37.Qh7+ Kf6
38.Qb7 Qd5+ 39.Kh2 Qf3 40.Rf1 Qe2
41.Kg2 c3 42.Qxa8 Qxf1+ 43.Kxf1 c2
44.Qxb8 c1Q+ 45.Kg2 Qc5 46.Qh8+ Ke7
47.Qg7+ Kd8 48.Qg8+ (7.860.200.778) 2489

36 52:37 -1.94 35.f4 (7.860.200.778) 2489

best move: g3-g4 time: 52:37.609 min n/s: 2.489.295 nodes: 7.860.200.778

[F. Bluemers]

Ok thanks
Dirty found g4 and it looked good,I just wondered if there was some defence it might have overlooked.
Best
Fonzy

[Rebel]

By the way, Ed, I really liked your approach with Prodeo 2.2, using different cores to run separate searches. I was hoping more programmers would follow suit and improve on that approach, but it didn't quite happen. There seems to be an entrenched resistance to innovation in the whole chess programming field and only the AZ surprise may have awakened some folks.

The idea using a second (somewhat speculative) search was not meant to gain playing strength but to improve on playing style (hopefully) without ELO loss. That it eventually turned out a small improvement was an unexpected case of collateral damage in reverse.

I think that most programmers would call it a waste of time because a typical doubling should bring 30-35 ELO and not only 10 ELO.

[Rebel]

You could also check out this interesting post by Ulysses and my reply to him:

http://www.talkchess.com/forum/viewtopi ... 17&t=66042

The problem with speculative engines is that they can find many amazing resources, but they can suddenly stumble on the very next move, and play an unsound sac that loses. (Like the great dribbler who loses the ball after one dribble too many.)

A second brain, running on separate cores, could then provide a second opinion and rule out such unsound moves while retaining the better ones.

The thinking process of the second brain could be paraphrased like this:

"I see this is an unexpected move that I pruned away, but it sure seems to work - now go ahead and play that move!"

"Here's another sac I stupidly pruned away - but it can't be refuted! I'm getting to like this new paradigm! Play the move, fast!"

"Wow, another sac, where do these moves come from? But wait, hold your horses, there's a problem: I see a definite refutation! Sorry, but we can't play this move, or we'll lose. Let's play my stodgy old move instead and live to play more moves, ok?"

Now, Ed, I know you're a great expert in these matters. Maybe you can tell us why this sort of thing is not in wide use, because I don't get it.

The ProDeo 2.2 way is indeed one way to do it. Houdart recently stated his tactical version works in a similar way, that is, if I understood his words correctly.