Blunders! Blunders!

[Volker Pittlik]

I went through the last issue of The Week in Chess looking for strange move s by high rated players. I hoped to find virtuoso moves like clever sacrifices and so on. I was also interested if there would be one or another blunder. It seems the latter are in a vast majority.

Here are two examples from games with slow time controls (90 minutes initially + 30 seconds per moves). Therefore I guess the moves in question are not a result of time trouble. I can't deny that the PGN is confused, but see yourself:

Code: Select all

[Event "XXXV Open"]
[Site "La Roda ESP"]
[Date "2008.03.22"]
[Round "6"]
[White "Sanchez, Joseph"]
[Black "Corrales Jimenez, F"]
[Result "0-1"]
[WhiteElo "2492"]
[BlackElo "2509"]
[ECO "B80"]
[EventDate "2008.03.19"]
[Annotator " Glaurung 2.0.1"]
[TimeControl "5400+30"]

1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6 6.Be3 e6 7.Qd2 b5 8.O-O-O
    { Move out of book ( f3 100%)}
8...b4 9.Na4 Qa5 10.b3 Nxe4 11.Qd3 Bb7 12.f3 Nc3 13.Nxc3 bxc3 14.Kb1 Nd7
15.Ne2 Rc8 16.Bd4 Ne5 17.Bxe5 $201 dxe5 $4 $18 $201
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     -1.66 / +327.00}
    ( 17...Qxe5 18.f4 Qc5 19.Qd4 d5 20.h4 a5 21.Qxc5 Rxc5 22.Rd3 Bd6 23.
    Rhh3 Ke7 24.Rxc3 Rxc3 25.Rxc3 h5 26.g3 $17 )
18.g3 $4 $17 $201
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     +327.00 / -2.45}
    ( 18.Qd7# $18 )
18...Be7 $4 $18 $201
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     -2.45 / +3.86}
    ( 18...Bd5 19.Qe3 Bb4 20.Rd3 f6 21.h4 O-O 22.Bh3 Rfd8 23.Rhd1 Rc6 24.
    Bg4 Bc5 25.Qc1 Bd6 $17 )
19.Bh3 $4 $17 $201
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     +3.86 / -2.88}
    ( 19.Qd7+ Kf8 20.Qxb7 Re8 21.Rd3 h5 22.h4 g5 23.hxg5 Bxg5 24.a4 Rd8
    25.Rxd8+ Bxd8 26.Qc8 Qc7 27.Qxa6 Bg5 28.Qa8+ Kg7 $18 )
19...Qc5 $4 $18 $201
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     -2.88 / +3.29}
    ( 19...O-O 20.Qe3 Bd8 21.Rhf1 Bb6 22.Qg5 Rfd8 23.Rxd8+ Bxd8 24.Qe3 Bb6
    25.Qg5 Qa3 26.Qc1 Qe7 27.Bg4 $17 )
20.Rhe1 $4 $17
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     +3.29 / -2.49}
    ( 20.Qd7+ Kf8 21.Qxb7 Qc6 22.Qxc6 Rxc6 23.Rd3 f5 24.Nxc3 Kf7 25.Kb2
    Rhc8 26.Re1 Bb4 27.Ree3 Rd6 28.a3 Rxd3 29.Rxd3 $18 )
20...O-O $201 21.Ka1 $2 $19
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     -2.88 / -4.58}
    ( 21.Qxc3 Qf2 22.Qxe5 Bf6 23.Qf4 Qxh2 24.Qb4 Bxf3 25.Bg4 a5 26.Qf4 Bc6
    27.Qe3 Rcd8 28.Bf3 $17 )
21...Qa3 $201 22.Rb1 $2 $19
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     -3.43 / -5.84}
    ( 22.Nxc3 Rfd8 23.Qe3 Qb4 24.Kb2 Rd4 25.Na4 Bxf3 26.Bxe6 Qa3+ 27.Kb1
    Rxa4 28.bxa4 Qxe3 29.Rxe3 Rb8+ 30.Kc1 Bxd1 31.Kxd1 fxe6 32.Rxe5 $19 )
22...Rc5 $201 23.b4 $2 $19
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     -5.07 / -6.96}
    ( 23.Nc1 Bg5 24.f4 exf4 25.b4 Rd5 26.Rb3 Qa4 27.Qf3 Rb5 28.Qg4 fxg3
    29.Qxg3 Bd2 30.Rf1 Rxb4 $19 )
23...Bd5 0-1

[diag]2r1kb1r/1b3ppp/p3p3/q3p3/8/1PpQ1P2/P1P1N1PP/1K1R1B1R w k - 0 18[/diag]

2r1kb1r/1b3ppp/p3p3/q3p3/8/1PpQ1P2/P1P1N1PP/1K1R1B1R w k - 0 18

after 17...dxe5??

Is that usual that two players of that level don't see a mate in one?


Here anthoer games with a a whole series of blunders in a row:

Code: Select all

[Event "31st San Sebastian Open"]
[Site "Donostia ESP"]
[Date "2008.03.22"]
[Round "8"]
[White "Mateo, Ra"]
[Black "San Segundo Carrillo, P"]
[Result "0-1"]
[WhiteElo "2484"]
[BlackElo "2542"]
[ECO "B14"]
[EventDate "2008.03.16"]
[Annotator " Glaurung 2.0.1"]
[TimeControl "5400+30"]

1.e4 c6 2.d4 d5 3.exd5 cxd5 4.c4 Nf6 5.Nc3 e6 6.Bf4
    { Move out of book ( Nf3 97% c5 1% cxd5 1% a3 1% Bg5 0%)}
6...Bb4 7.Bd3 dxc4 8.Bxc4 O-O 9.Ne2 a6 10.a3 Be7 11.Qd3 b5 12.Qf3 Ra7 13.
Bd3 Rd7 14.Rd1 Bb7 15.Qh3 Nc6 16.O-O g6 17.Be5 Nxe5 18.dxe5 Nd5 19.Nxd5
Bxd5 20.Nd4 $201 Bc4 $2 $10 $201
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     -1.60 / +0.19}
    ( 20...Bxg2 21.Qxg2 Rxd4 22.Be2 Bh4 23.b3 Qb6 24.Qf3 Rxd1 25.Rxd1 f6
    26.Qf4 Bg5 27.Qd4 Qxd4 28.Rxd4 fxe5 $17 )
21.Qf3 $4 $19 $201
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     +0.19 / -4.68}
    ( 21.Nc6 Rxd3 22.Rxd3 Qxd3 23.Nxe7+ Kg7 24.Qxd3 Bxd3 25.Rd1 Be2 26.Rd6
    Re8 27.Nc6 Ra8 28.Nd4 Bc4 29.f4 Bd5 30.Kf2 $10 )
21...Bb3 $4 $18 $201
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     -4.68 / +4.09}
    ( 21...Rxd4 22.Bxc4 bxc4 23.Qe2 Bc5 24.Rxd4 Bxd4 25.Qxc4 Bxe5 26.b4
    Qd6 27.g3 Rd8 28.Re1 Bd4 29.Rc1 Qb6 30.Qe2 $19 )
22.Rc1 $4 $19 $201
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     +4.09 / -5.19}
    ( 22.Nxb3 Qc7 23.Qg3 Rfd8 24.Rc1 Qb7 25.Be2 Rc7 26.Rxc7 Qxc7 27.Rc1
    Qb6 28.Bf3 b4 29.axb4 Bxb4 $18 )
22...Bg5 $4 $17 $201
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     -5.19 / -1.86}
    ( 22...Rxd4 23.Be2 Bd5 24.Qc3 Rf4 25.Rfd1 Qb6 26.Qe3 Qxe3 27.fxe3 Re4
    28.Bf3 Rxe5 29.Bxd5 exd5 30.Kf2 Kg7 $19 )
23.Be4 $4 $19
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     -1.86 / -5.68}
    ( 23.Rc3 Bd5 24.Qg4 Bxg2 25.Nxe6 fxe6 26.Kxg2 Bd2 27.Qxe6+ Kh8 28.Be4
    Bxc3 29.bxc3 Re8 30.Qxa6 Qg5+ 31.Kh1 Qxe5 $17 )
23...Rxd4 24.Rfe1 $201 Bd5 $19
    { Glaurung 2.0.1: -8.27 (-5.86)}
    ( 24...Bxc1 25.Rxc1 Qg5 26.Re1 Qxe5 27.Qe3 Bd5 28.Bf3 Qxe3 29.fxe3 Rd2
    30.Re2 Rxe2 31.Bxe2 Kg7 32.Kf2 Kf6 $19 )
25.g3 Bxe4 26.Rxe4 $201 Qa8 $19
    { Glaurung 2.0.1: -13.70 (-6.01)}
    ( 26...Bxc1 27.Re2 Rd2 28.Rxd2 Qxd2 29.b4 Qe1+ 30.Kg2 Qxe5 31.Kg1 Qa1
    32.Qd1 Qxa3 33.Qd4 Bb2 34.Qe4 Qa1+ 35.Kg2 Bd4 $19 )
27.Rce1 Rc8 $201 28.h4 $2 $19
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     -5.05 / -6.66}
    ( 28.Kg2 Qd5 29.R4e2 Qd7 30.h4 Bc1 31.Kh2 a5 32.Kg2 h6 33.Qf6 Rd3 34.
    Kg1 Rd1 35.Rf1 $19 )
28...Rcc4 29.hxg5 Rxe4 30.Qg2 $201 Qc6 $19
    { Glaurung 2.0.1: -22.25 (-9.50)}
    ( 30...Rxe1+ 31.Kh2 Qxg2+ 32.Kxg2 b4 33.axb4 Rxb4 34.b3 Rxb3 35.Kh3 a5
    36.Kg2 a4 37.f4 a3 38.Kf2 Ree3 39.Kg2 Rxg3+ 40.Kf2 $19 )
31.Rd1 Rxe5 0-1


Regards


Volker


BTW: Does someone have an idea what this $* output in the PGN means? In Scid annotation symbole are visible instead.

[Teemu Pudas]

BTW: Does someone have an idea what this $* output in the PGN means? In Scid annotation symbole are visible instead.

They're NAG, i.e. Numeric Annotation Glyphs.

[Onno Garms]

Is that usual that two players of that level don't see a mate in one?

At least we know from Kramnik vs Fritz that 5000 Elo are not enough to see a mate in one. (Fritz's operator said he was also surprised by the mate. I think his rating is 2200 + something.)

However a queen protected by a rook from the own side is a much more common pattern than a queen protected by a knight from f8 which had never been there before in the game.

[Volker Pittlik]

...
At least we know from Kramnik vs Fritz that 5000 Elo are not enough to see a mate in one. ...

Although Elo rating wasn't established at that time overseen mates (and other big variance in score) can be found some decades ago too. Something is sure: Neither Tal nor Spassky cheated by using a computer!

Code: Select all

[Event "URS-chTJ"]
[Site "Leningrad RUS"]
[Date "1954.??.??"]
[Round "?"]
[White "Tal, Mikhail N"]
[Black "Spassky, Boris V"]
[Result "0-1"]
[ECO "A61"]
[EventDate "1954.??.??"]
[Annotator " Glaurung 2.0.1"]

1.c4 Nf6 2.Nc3 e6 3.d4 c5 4.d5 exd5 5.cxd5 g6 6.Nf3 Bg7 7.Bf4 d6 8.h3
    { Move out of book ( Qa4+ 81% e4 15% e3 4%)}
8...O-O 9.e3 Ne8 10.Be2 Nd7 11.O-O Ne5 12.Bxe5 dxe5 13.Nd2 f5 14.Qb3 Nd6
15.Nc4 e4 16.Nb5 Nxb5 17.Qxb5 b6 18.d6 Bd7 19.Qb3 b5 20.Nb6+ c4 21.Bxc4+
bxc4 22.Qxc4+ Rf7 23.Nxa8 Qxa8 24.Qb3 Be5 25.Rac1 Kg7 26.Rfd1 a5 27.Rc7
Qe8 28.Qd5 a4 29.b4 axb3 30.axb3 Bf6 31.Rb7 Qe5 32.Qc4 f4 33.exf4 Qxf4 34.
g3 Qf3 35.Qd5 Bc3 36.Rf1 Kh6 $201 37.Qc4 $4 $19 $201
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     -0.52 / -327.00}
    ( 37.h4 Kg7 38.Rc7 Bb2 39.h5 Qxh5 40.Qxh5 gxh5 41.Kg2 Kf6 42.Rd1 Be5
    43.Rc4 Kf5 44.b4 )
37...Qf6 $4 $10
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     -327.00 / +0.35}
    ( 37...Bxh3 38.Qxe4 Qxe4 39.f3 Rxf3 40.Rxh7+ Kxh7 41.Rxf3 Qxf3 42.Kh2
    Qg2# $19 )
38.Qe2 Bd4 39.Rb4 e3 $201 40.Qd3 $4 $19 $201
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     -1.78 / -21.54}
    ( 40.Rxd4 Qxd4 41.Qxe3+ Qxe3 42.fxe3 Rxf1+ 43.Kxf1 Kg5 44.h4+ Kf5 45.
    Ke2 Bc6 46.Kd3 Ke5 47.b4 Bb5+ 48.Kc3 h5 49.e4 Kxd6 $17 )
40...Qxf2+ $4 $17
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     -21.54 / -2.60}
    ( 40...exf2+ 41.Kh2 Qe6 42.Qe3+ Bxe3 43.Rh4+ Kg7 44.Rg4 Qxd6 45.Re4
    Rf3 46.Rf4 Bxf4 47.gxf4 Qxf4+ 48.Kh1 Bxh3 $19 )
41.Rxf2 exf2+ 42.Kh2 f1=N+ 43.Qxf1 Rxf1 44.Rxd4 Rf2+ 45.Kg1 Rf3 0-1


Volker

I have a lot more of this. It seems there are "blunders" (more of three pawns rating difference) in ~a third of all games rated 2650+. If there is interest I can post some more.

[Onno Garms]

Hi Volker,

how did you find these games? I guess you did not feed all the games into glaurung manually, did you? Do you know a way to let an engine analyse all games in a pgn file?

Onno

[Volker Pittlik]

Hi Volker,

how did you find these games? I guess you did not feed all the games into glaurung manually, did you? Do you know a way to let an engine analyse all games in a pgn file?

Onno

I use Scid for it. It's a very fine chess database application. If you don't know it check the links on the bottom of the main page. I collected a huge number of games (as most of us I assume). I filtered for standard time control games in which both players have a rating of 2650 Elo or more. And then I let Glaurung or another engine annotate it. There is a very nice feature in the enhanced Scid. You can define a margin from where on a move is annotated as blunder. The analysis is still running but I guess it will be finished today or tomorrow. I will offer the games with blunders in it (more than 1000) for download then.


Volker

[Volker Pittlik]

... The analysis is still running but I guess it will be finished today or tomorrow. I will offer the games with blunders in it (more than 1000) for download then...

Sorry there will be a delay. I'm not satisfied with the quality of the analysis yet. I'll start again with other engine settings and publish the results in parts.

Volker

[Onno Garms]

I use Scid for it. It's a very fine chess database application.

Thank you. I still had the outdated 3.6.1 (which is the newest on sourceforge). After upgrading to 3.6.22 (and overcoming installation problems by manual inspection of the makefile) I got this working.

[Volker Pittlik]

...I'll start again with other engine settings and publish the results in parts.

Volker

I went through 216 games now and found strange moves in more than a third of them. I selected only game in which both player had a rating of 2400 Elo or more. I used a rating scale found on the internet somewhere to estimate player who were active before the Elo rating came in use. I also tried to avoid all bltz-, simul- and similar games. I can't guarantee I excluded all of them.

I let the engines calculate for 15 seconds per move and set a margin of 1.5 pawns. It is possible that there are false blunder claims among my result or that blunders have been overseen. But I'm pretty sure more than 75% are correct.

Here is one example from an older game:

Code: Select all

[Event "Buenos Aires ol f-A"]
[Site "Buenos Aires ARG"]
[Date "1939.??.??"]
[Round "13"]
[White "Keres, Paul"]
[Black "Mikenas, Vladas I"]
[Result "1-0"]
[ECO "B15u"]
[EventDate "1939.??.??"]
[Annotator " Glaurung 2.0.1"]

1.e4 c6 2.d4 d5 3.Nc3 dxe4 4.Nxe4 Nf6 5.Nxf6+ exf6 6.Bc4 Bd6 7.Qh5
    { Move out of book ( Qe2+ 83% Ne2 17%)}
7...O-O 8.Ne2 g6 9.Qf3 Re8 10.Bh6 Bf5 11.O-O-O Be4 12.Qb3 Qc7 13.f3 Bd5
14.Bxd5 Rxe2 15.Rhe1 Rxe1 16.Rxe1 Na6 17.Bc4 b5 18.Qe3 Qd7 19.Bb3 Nc7 20.
Qf2 a5 21.a4 bxa4 22.Bxa4 Nd5 23.Qh4 g5 24.Qe4 Bxh2 25.c4 Nb4 26.Qe7 Bf4+
27.Kb1 $201 Qf5+ $4 $18
    { Glaurung 2.0.1: Analysis reports blunder, with score change from/to:
     -0.43 / +6.70}
    ( 27...Qxe7 28.Rxe7 Rd8 29.c5 Nd5 30.Ra7 Re8 31.Bb3 Nc7 32.Rb7 Ne6 33.
    Bxe6 Rxe6 34.Ra7 Re8 35.Rxa5 Be3 36.Ra4 f5 $10 )
28.Ka1 Be5 29.dxe5 Nd3 $201 30.Bc2 $18
    { Glaurung 2.0.1: +327.00 (+9.23)}
    ( 30.e6 Qg6 31.Bxc6 Nb4 32.Bxa8 Nc2+ 33.Kb1 Na3+ 34.Ka2 Qxh6 35.exf7+
    Kg7 36.f8=Q+ Kg6 37.Qfxf6+ Kh5 38.Rh1# $18 )
30...fxe5 $201 31.Rd1 $18
    { Glaurung 2.0.1: +327.00 (+6.70)}
    ( 31.Bxd3 Qf4 32.g3 Qxg3 33.Qf6 Qxe1+ 34.Bb1 Qxb1+ 35.Kxb1 g4 36.Qg7#
    $18 )
31...Qg6 32.Bxg5 h6 33.Bf6 e4 34.fxe4 Nb4 35.Rd8+ 1-0


The games can be downloaded here: http://www.vpittlik.org/chess/blunders.080417.pgn.bz2

Because I have so many of thes games (more than 300.000) to analyze all of them would take years. Maybe I will continue with certain events or players, but everyone who is interested can do it himself.

Volker