I've never seen or heard of a competitive human master game in which underpromotion to a bishop was the only winning move, has anyone else here? However, while testing dedicated chess computers for ICD thirty years ago, I had one game that halted after a bishop underpromotion. When I checked it out, it was indeed the only winning move, and the other side wasn't programmed to recognize bishop underpromotion!OliverBr wrote: ↑Thu Sep 03, 2020 7:38 am Hello together,
chess puzzle lovers will love this: A position from a real game where an under-promotion to bishop happens. And even better, a pinned pawn under-promoted.
If you ask me, this is a 1 in a million situation:
[d]4b3/1k1P4/1P6/1K3p2/5N2/8/8/8 w - - 1 67 bm dxe8=b
(Mate in 7 with underpromotion to bishop)
And here the complete game between Fruit 2.1 and OliThink 5.7.0:
[pgn]
[Event "?"]
[Site "?"]
[Date "2020.09.03"]
[Round "4256"]
[White "Fruit 2.1"]
[Black "OliThink 5.7.0"]
[Result "1-0"]
1. c3 e5 2. c4 Nc6 3. Qc2 Nf6 4. Nf3 e4 5. Ng5 d5 6. cxd5 Qxd5 7. h4 Nb4 8. Qd1
Qf5 9. d3 Bc5 10. Be3 exd3 11. exd3 Bxe3 12. fxe3 Ng4 13. Qd2 Qc5 14. Nc3 Qxe3+
15. Qxe3+ Nxe3 16. Kd2 f6 17. Nf3 Nxf1+ 18. Rhxf1 O-O 19. a3 Nc6 20. Nb5 Bf5
21. Nfd4 Nxd4 22. Nxd4 Bg4 23. Rae1 Rfe8 24. Nb5 Rxe1 25. Rxe1 c6 26. Nd6 b6
27. Rc1 c5 28. b4 cxb4 29. axb4 Rd8 30. Ne4 Be6 31. Rc7 Bd5 32. Ng3 g6 33. Ne2
Bxg2 34. Rxa7 Be4 35. d4 f5 36. Rc7 Bg2 37. Kc3 b5 38. Nf4 Bf3 39. Ne6 Re8
40. Ng5 Bd5 41. Nxh7 Re3+ 42. Kd2 Re6 43. Ng5 Rd6 44. Ke3 Rd8 45. Rc5 Kg7
46. Rxb5 Kh6 47. Nh3 Bc4 48. Rc5 Re8+ 49. Kf2 Re2+ 50. Kf3 Rc2 51. Nf4 g5
52. hxg5+ Kxg5 53. Nh3+ Kf6 54. Kf4 Bd3 55. Ke3 Bc4 56. Nf4 Bb3 57. Rxc2 Bxc2
58. b5 Ke7 59. d5 Kd6 60. b6 Bb3 61. Kd4 Ba2 62. Kc3 Kd7 63. Kb4 Kc8 64. Kb5 Kb7
65. d6 Bf7 66. d7 Be8 67. dxe8=B 1-0
[/pgn]
For information: OliThink 5.7.0 has a little bug, where it can't do underpromtions of pinned pawns. Promotion of pinned-pawns are already very rare, now who would ever think there was an underpromotion of a pinned pawn. An not artificially constructed, it's real.
It could only be detected because of that little bug in OliThink 5.7.0, after Fruit's move, it stalled. In OliThink 5.7.1 this bug is fixed.
However, your example is much less special since underpromotion to a knight (which is often the only winning move) also wins. I suppose it is quite rare that bishop and knight work but rook and queen don't, but probably much more common than positions where only a bishop works.