Genius move??? **cough**

[Dann Corbit]

Given this position, a facebook post suggests offering the bishop by moving it to d6 and calls it a genius move:
[d]7k/7p/7N/p7/1B6/6q1/2P3P1/6K1 w - -
Here is the resulting position:
[d]7k/7p/3B3N/p7/8/6q1/2P3P1/6K1 b - -
This is, of course, an easy win for black by simply refusing the bishop (leading to a fork of king/queen) and moving the queen to e3 which is a win for black in 18 moves.

You actually lose faster by doing the "genius" move.
I think anyone who actually plays chess will instantly see that taking the bishop is stupid.
Crystal 5 says:
23/37 00:25 1,681,302k 66,824k +M18 Qe3+ Kf1 Qxh6 Be5+ Kg8 Kf2 Qg5 Bb8 Qc5+ Kf3 Qxc2 Bf4 Qd3+ Kf2 Qe4 Be3 a4 Bc1 Qc2+ Kf3 Qxc1 g4 a3 g5 Qxg5 Ke4 a2 Kf3 Qe5 Kg2 a1Q Kh3 Qh1+ Kg4 h5+

[towforce]

Well that ruins my plan to learn chess brilliancy from random Facebook posts.

[Ajedrecista]

Hello Dann:

Swindle matches better the proposed Bd6. This is easily refuted with Qe3+ and Qxh6, as you already noted.

Taking a look into the cramped position of the black king, let us recycle the position with a minor, subtle change that still leaves a winning position for white, but with some details: a bad move can lead to a drawn endgame despite being Q+N vs. P up! Just place the c2-pawn on b2 or d2 for support Bc3+:

[d]7k/7p/7N/p7/1B6/6q1/3P2P1/6K1 w - - 0 1

[pgn][Event ""]
[Site ""]
[Round ""]
[Date "2026.01.06"]
[SetUp "1"]
[FEN "7k/7p/7N/p7/1B6/6q1/3P2P1/6K1 w - - 0 1"]
[Result "1-0"]

1.Bc3+ Qxc3 (1...Qg7 (1...Qe5 2.Bxe5# 1-0) 2.Nf5 {(Winning the queen for free)}) 2.dxc3 a4 3.Nf5 a3 4.Nd4 a2 5.Nb3 Kg7
6.Kf2 Kg6 7.Ke2 Kg5 8.Kd2 Kg4 9.Kc2 Kg3 10.Kb2 (10.c4? Kxg2 (10...h5) {(10.Kb2 is the only winning move)}
11.c5 h5 12.c6 h4 13.c7 h3 14.c8=Q a1=Q! (14...h2? {(14...a1=Q is the only drawing move)} 15.Qg4+ Kf2
16.Nc1 {(16.Nc5 also wins)} h1=Q 17.Nd3+ Kf1 18.Qf4+ Kg2 19.Qe4+ Kg1 20.Qe1+ Kh2
21.Qh4+ Kg2 22.Nf4+ Kg1 23.Qe1+ Kh2 24.Qf2+ Qg2 25.Qxg2# 1-0) 15.Nxa1 h2 1/2-1/2)) 1-0[/pgn]

There are multiple winning paths, like 3.- Ng4 instead of 3.- Nf5; or 4.- Ne3 instead of 4.- Nd4, and many more. I share some interesting, worthy sidelines in the PGN, like how white can miss the win and, if that, how black can miss the draw with clever Q+N manoeuvres by white.

The original position was a wanna-be for something more deep!

Regards from Spain.

Ajedrecista.

[Dann Corbit]

This position is much more interesting than the original proposal.
I would not rate the original as a swindle either.
For me a swindle is reaching for a win when the best you can do is draw, but not with a losing move.
Even going from a lost position to try to get a draw should not result in a worse position.
If the swindle move does not worsen the outcome (e.g. a draw still draws or a bad position does not get worse) that is a swindle that will work on weak humans but not against a chess engine unless it is buggy.
Robert Hyatt had swindle code in Crafty, but the swindles never damaged the position.

[Jouni]

Analysis by Stockfish 17.1:
...
1.Bxa5 Qe3+ 2.Kf1 Qxh6 3.Bc3+ Kg8 4.Be1 Qf4+ 5.Ke2 Qe4+ 6.Kd2 Qxg2+ 7.Kd1 h5 8.c4 Qg4+ 9.Kd2 Qxc4 10.Ke3 h4 11.Bf2 Qg4 12.Bxh4 Qxh4 13.Kd3 Kg7 14.Kc3 Kf6 15.Kb3 Ke6 16.Kb2 Qd4+ 17.Ka3 Kd5 18.Ka2 Kc4 19.Kb1 Kc3 20.Ka2 Qb4 21.Ka1 Qb2#
-+ (-#21) Depth: 55/43

[chrisw]

This position is much more interesting than the original proposal.
I would not rate the original as a swindle either.
For me a swindle is reaching for a win when the best you can do is draw, but not with a losing move.
Even going from a lost position to try to get a draw should not result in a worse position.
If the swindle move does not worsen the outcome (e.g. a draw still draws or a bad position does not get worse) that is a swindle that will work on weak humans but not against a chess engine unless it is buggy.
Robert Hyatt had swindle code in Crafty, but the swindles never damaged the position.

Really, where? Can you list the code snippet? Or don’t believe you.

[mar]

Really, where? Can you list the code snippet? Or don’t believe you.

here, look for "swindle mode" https://github.com/MartinMSPedersen/Cra ... rce/root.c

[Dann Corbit]

K:\chess\crafty-241-ja\src\crafty-24.1>grepcarl swindle *
analyze.c ( 30): int save_swindle_mode = swindle_mode;
analyze.c ( 32): swindle_mode = 0;
analyze.c ( 185): swindle_mode = save_swindle_mode;
annotate.c ( 88): int save_swindle_mode;
annotate.c ( 100): save_swindle_mode = swindle_mode;
annotate.c ( 171): swindle_mode = 0;
annotate.c ( 531): swindle_mode = save_swindle_mode;
crafty.hlp ( 87): swindle on|off................. enables/disables swindle mode.
data.c ( 465): int swindle_mode = 1; /* try to swindle */
data.h ( 8): extern int swindle_mode;
iterate.c ( 130): if (EGTB_draw && !puzzling && swindle_mode)
iterate.c ( 135): Print(128, "Drawn at root, trying for swindle.\n");
main.c ( 2225): * files. EGTB "swindler" added. If the position is drawn at the *
main.c ( 2246): * to EGTB "swindle" code. It now only tries the searches if it is *
main.c ( 2248): * a swindle in KRP vs KR if you don't have the P. Some very *
main.c ( 2326): * command it would crash. Analyze mode now turns off the "swindle" *
main.c ( 2346): * the opponent's king position. New swindle on*off command allows *
main.c ( 2347): * the user to disable "swindle mode" so that Crafty will report *
main.c ( 2348): * tablebase draws as draws. Swindle mode is now disabled auto- *
main.c ( 2708): * rather than even in material, which gives 'swindle mode' a chance *
option.c ( 3556): * "swindle" command turns swindle mode off/on. *
option.c ( 3560): else if (OptionMatch("swindle", *args)) {
option.c ( 3562): swindle_mode = 1;
option.c ( 3564): swindle_mode = 0;
option.c ( 3566): printf("usage: swindle on|off\n");
root.c ( 44): if (swindle_mode && (tb_value == DrawScore(wtm)))
root.c ( 164): * drawing moves when we are in the swindle-mode situation *
search.c ( 129): * can be zero when trying to swindle the opponent, so *
search.c ( 135): * Note that in "swindle mode" this can be turned off by *

K:\chess\crafty-241-ja\src\crafty-24.1>

[chrisw]

This appears to be using the egtbs and only acts at the root. Not entirely sure how on earth this works conceptually if at all?
Description of concept?

[Ajedrecista]

Hello Dann:

This position is much more interesting than the original proposal.
[...]

Thanks for your compliment. I wrongly stated in my PGN that after 1.- Bc3+, Qg7; 2.- Nf5, white would win the black queen for free because I expected something like 2.- ... a3 or 2.- ..., Kg8 and then 3.- Nxg7, but 2.- ..., Qxc3 is still available, so queen for bishop instead of queen for nothing.

Nevertheless, this Nf5 move gaining a tempo (maybe the wrong word) chasing the black passed pawn gave me to think in a more extreme position with minor changes:

[d]7k/7p/6qN/8/8/p1P1B3/6P1/6K1 w - - 0 1

These are the main ideas behind the rearragement of pieces:

• The blacked passed pawn is more advanced and apparently more difficult to stop.

• Be3 is defending Nh6, a more natural setup for me.

• The white pawn on c3 is temporarily blocking the a1-h8 diagonal for the bishop, crucial for stopping the passed pawn.

[pgn][Event ""]
[Site ""]
[Round ""]
[Date "2026.01.07"]
[SetUp "1"]
[FEN "7k/7p/6qN/8/8/p1P1B3/6P1/6K1 w - - 0 1"]
[Result "1-0"]

1.Bd4+ Qg7 (1...Qf6 2.Bxf6# 1-0) 2.Nf5 {(2.Bxg7+ also wins)} 2...a2 (2...Qxd4+ 3.Nxd4 a2 4.Nc2) 3.Bxg7+ Kg8 4.c4 1-0[/pgn]

------------

Then, the 'most difficult yet' exploiting the same concept of the black cornered king: two threatening connected black pawns in the queenside and white pieces far from the action... sure? Depending on where the white pawn on the queenside lays, it is a draw or even a white win according to SF! Exclamation marks are only moves to keep the win or draw:

[pgn][Event ""]
[Site ""]
[Round ""]
[Date "2026.01.07"]
[SetUp "1"]
[FEN "7k/7p/6qN/8/1p6/p1P1B3/6P1/6K1 w - - 0 1"]
[Result "1/2-1/2"]

1.Bd4+! Qg7 (1...Qf6 2.Bxf6# 1-0) 2.Bxg7+! (2.Nf5 b3 {(2...Kg8, unpinning, followed by 3...b3 also wins for black)} 3.Bxg7+ Kg8 4.c4 b2 {(Other tries to stop black queening also fail)}) 2...Kxg7 3.Nf5+! Kf6 4.Nd4! a2 {(4...bxc3 also is a draw)} 5.Nb3! (5.Nc2 b3 0-1) 1/2-1/2[/pgn]

[pgn][Event ""]
[Site ""]
[Round ""]
[Date "2026.01.07"]
[SetUp "1"]
[FEN "7k/7p/6qN/8/1p6/p3B3/3P2P1/6K1 w - - 0 1"]

1.Bd4+! Qg7 (1...Qf6 2.Bxf6# 1-0) 2.Bxg7+! (2.Nf5 b3 {(2...Kg8, unpinning, followed by 3...b3 also wins for black)}) 2...Kxg7 3.Nf5+! Kf6 4.Nd4! a2 5.Nb3! (5.Nc2 b3 0-1) Ke5 6.d3! Kf4 (6...a1=Q+ 7.Nxa1! Kd4 8.Nc2+! Kc3 9.Nxb4+! Kxb4 {(White checkmates in 22 moves)}) 7.Kf2! h5 8.g3+! 1-0[/pgn]

I found a series of eight only moves in the last problem to keep the white win (nine in one sideline). With other black moves like 4.- ..., h5, there are more than one winning moves by white.

------------

This failed genius move of the OP brought some interesting positions after all! More problems with the same idea/s are welcome.

Regards from Spain.

Ajedrecista.