Opening analyses: SF / Komodo to 88% similar, not nice!!

[mjlef]

"same first move or better"? What does the "or better" mean?

if you mean just the same move, then that is a much higher matching rate than I have seen reported before. If these were positions with just one good move then that could be expected, but I would not expect it in varied openings.

If chess is a game where 88% of the time there is just one correct (win or draw preserving) move I would be very surprised.

Thanks for the link correction.

[Frank Quisinsky]

Hi Mark,

if you are looking in Excel workbook v6.46 you can see the results from the others I tested. You can see how many different moves to each of the 26.146 positions engines found out.

With Texel (half ready) 7 Engines so far.
We have 3 positions 7 different engines found 7 different best first moves without 0,00 or a bad eval. Such positions are optimal for a test-set (for an example). With 6 different first moves over 100, with 5 different first moves more as 1.000 and so one.

88% the same first move is really a lot.
OK, many bad positions are in the update 1 database Stockfish and Komodo made the error check. One of the reasons for the 88%. Around 10.000 I reject. 12.229 are new for the FEOBOS project after error checking by Komodo and Stockfish with 30 seconds per position.

67% Shredder / Houdini is the highest results from the 26.146 error checked database with 1-minute analyses in comparing engines.

Stockfish and Komodo have to analyse the 26.146 balanced FEOBOS postions very soon. Sure here that the results will be not 88% ... mabye 70 or 75% ... think so!

All that and much more can be see in Excel workbook.

88,15% means ... the same first move.

Others can thinking I am speaking here about: Stockfish is a clone from Komodo or Komodo is a clone from Stockfish. That is bullshit ... I am speaking more about it that Stockfish and Komodo are very similar with the first calculated moves after the opening book ended.

Should be clear that such strong engines, like Komodo and Stockfish produced more often the same results. But in the past I am thinking that for opening analysis the combination Stockfish / Komodo is not to beat.

Today I am thinking that one of the engines is enough and more interesting is to use here Chiron, Booot (complete other style and very sure with the first moves) in combination with Stockfish or Komodo for more interesting results to opening analyses.

After all ...
The FEOBOS files are more clear if 10 engines analyzed the complete 38.375 (with the update 26.146 + 12.229) positions. A long way ... ready around January 2018.

Best
Frank

But after I added the analyses from Stockfish and Komodo in Excel workbook at first I am very suprised about the stats I posted before (have a look in the grafics).

[Tony P.]

right, reason is both top engines are fully clueless about the opening.

for example, taking the Dutch.

after 1.d4 f5, there is a single move that would give white advantage and winning chances, 2.d5!

[d]rnbqkbnr/ppppp1pp/8/3P1p2/8/8/PPP1PPPP/RNBQKBNR b KQkq - 0 2

unfortunately, SF and Komodo are fully clueless about that.

as well as all current chess theory, which even does not consider d5.

on the other hand, if black wants to enter the favourable stonewall construction, this is only possible with a transposition via the QG, for example 1.d4 d5 2.c4 e6 3.Nc3 f5!

[d]rnbqkbnr/ppp3pp/4p3/3p1p2/2PP4/2N5/PP2PPPP/R1BQKBNR w KQkq f6 0 4

again, neither top engines, nor current chess theory considers 3...f5 for black as the best move or even at all.

chess openings are a very difficult matter.

This is pure gold

I'm not sure how to evaluate the 2nd position, but as for the 1st one, I think that Black equalises with a gambit - 1. d4 f5 2. d5 Nf6 3. Nf3 c6 4. c4 e6 5. dxe6 Bb4+! 6. Bd2 Bxd2+ 7. Qxd2 Ne4 8. Qd4?! O-O. (This is an idea by Komodo 9.02.)

[d]rnbq1rk1/pp1p2pp/2p1P3/5p2/2PQn3/5N2/PP2PPPP/RN2KB1R w KQ - 3 9

The key thing is that the knight is tremendously strongly placed on e4, controlling the d2 square, and this provides Black with enough compensation for the pawn. Black can play Qa5+ followed by dxe6 at pretty any moment, so the pawn sacrifice is temporary. White shouldn't play exd7, as that would open the d-file for Black's rooks while White is still far from castling (has yet to develop the f1 bishop).

Feel free to prove White's advantage, but I'd rather play with Black after 1. d4 f5 2. d5.

Sorry for the derailment and the necromancy, just the 2. d5 idea is too funny to be buried in the depths of the forum.

[Leo]

Million dollar question -> how long until engines can play openings by themself above GM level. Or more precisely, when they can construct new openings.

in 100-150 years, certainly top engines will be much better from now, but I am not sure they will be able to then play perfectly more than 50% of openings.(they play perfectly only 1% now)

certainly, your grand-grand-grand-children will see it.

now the million bucks.

And maybe he will be a grand-grand master.

[Dann Corbit]

I have these records in my database for those positions:

[d]rnbq1rk1/pp1p2pp/2p1P3/5p2/2PQn3/5N2/PP2PPPP/RN2KB1R w KQ - acd 37; acs 523; bm Nc3; ce 26; pm Nc3; pv Nc3 Nxc3 exd7 Nxe2 Bxe2 Qa5+ Qd2 Qxd2+ Kxd2 Nxd7 Kc2 Nf6 Rad1 Ne4 Rhf1 g6 Nd4 Bd7 f3 Nc5 Rfe1 Rad8 Nb3 Nxb3 axb3 Kf7 b4 Be6 Kc3 Rfe8 f4 h6 Rd3 a6 Rxd8 Rxd8 Bd3 Kf6 b3 Bf7;

[d]rnbqkbnr/ppp3pp/4p3/3p1p2/2PP4/2N5/PP2PPPP/R1BQKBNR w KQkq - acd 37; acs 2506; bm Bf4; c3 "Nf3"; cce 133; ce 75; pm Nf3 {495} g3 {238} Bf4 {215} cxd5 {98} e3 {70} g4 {23} Qc2 {18} Nh3 {14} c5 {10} Qb3 {2} h3 {2} f4 {1} h4 {1}; pv Bf4 Nf6 e3 Bb4 Bd3 O-O Ne2 c6 a3 Be7 h3 Nbd7 Rc1 dxc4 Bxc4 Nd5 Bb3 N7f6 Be5 b6 O-O Bb7 Nf4 Qd7 Ncxd5 Nxd5 Nd3 Ba6 Re1 Rad8 Qc2 Bd6 Red1 Bxd3 Qxd3 c5 Bxd6 Qxd6 dxc5; white_wins 447; black_wins 193; draws 210; Opening ECO:A84; Dutch: Rubinstein, 3...d5; 1.d4 f5 2.c4 e6 3.Nc3 d5 *;

[d]rnbqkbnr/ppppp1pp/8/3P1p2/8/8/PPP1PPPP/RNBQKBNR b KQkq - acd 37; acs 2213; bm c6; cce -0; ce 0; pm Nf6 {7} c5 {2} c6 {2} d6 {1} g6 {1}; pv c6 c4 Nf6 Nf3 d6 Nc3 e5 dxe6 Bxe6 e3 g6 Ng5 Bg8 b3 h6 Nf3 Bg7 Qd2 Bf7 Bb2 O-O Rd1 Na6 Be2 Re8 O-O Ne4 Qc1 Nac5 Qa1 Ng5 Nxg5 Qxg5 Na4 Bxb2 Qxb2 Ne4 a3; white_wins 3; black_wins 3; draws 3;