18

Wikipedia's article on the fifty-move rule states:

All of the basic checkmates can be accomplished in well under 50 moves. However, in the 20th century it was discovered that some positions of certain endgames can only be won in more than fifty moves (without a capture or a pawn move). The rule was changed to include certain exceptions in which one hundred moves were allowed with particular material combinations. However, more and more exceptions were discovered and in 1992 FIDE abolished all such exceptions and reinstated the strict fifty-move rule.

I have never seen an example of this, though I used to think the K, N, B vs. K ending was one. When is it possible to force a win in 50 moves or more, but not less than 50 moves?

Brian Towers
  • 96,800
  • 11
  • 239
  • 397
Daniel
  • 6,172
  • 5
  • 39
  • 74

3 Answers3

14

Yes, there are quite a few endgames that require more than 50 moves with best play from the defender.

Wikipedia has a list of material imbalances where it could take more than 50 moves depending on the specific position. Here are a few examples:

  • Rook and bishop vs lone rook (often a draw though)
  • Two bishops vs a lone knight
  • Two knights vs a lone pawn (the defender must have a pawn)
  • Four knights vs a lone queen (would never happen over the board…)
  • Queen and rook vs queen and rook
  • Two queens vs two queens
Nikana Reklawyks
  • 2,465
  • 19
  • 34
Andrew
  • 15,144
  • 4
  • 65
  • 105
  • 6
    Wow... I see that 517 moves to force a win has even bee proven! That's amazing! – Daniel May 02 '12 at 20:33
  • I wonder what value of N would best equalize the game if there were a rule that after N turns without a pawn move or piece capture a win by White would no longer be recognized (meaning that for White to win, he must maintain sufficient advantage that black could not temporize for N turns without a piece capture or pawn move). – supercat Oct 28 '15 at 16:19
  • 1
    I should add to the list "blocked pair of border pawns plus R vs B", since this was the practical example that started all the FIDE changes. – Hauke Reddmann Mar 01 '21 at 09:29
  • @supercat I think what you're asking is the idea behind DTR EGTBs (see https://en.wikipedia.org/wiki/Endgame_tablebase#Metrics:_Depth_to_conversion_and_depth_to_mate) but they're hard to generate. – Martin Rattigan May 11 '21 at 15:58
  • @MartinRattigan: What's the difference between DTZ and DTR, and in what circumstances would they not equal DTC? I think all three cases would count the number of moves until a pawn is moved or a piece is captured in such a way that the resulting endgame is still a win for the same side. I guess that DTC might not count situations in which the non-winning side would be forced to move a pawn or make a capture, but in what cases would it be useful to exclude those? – supercat May 11 '21 at 16:03
  • If I understand it correctly, the metric in a DTR database is the smallest n such that the position would be winnable under an n move rule. (But I don't guarantee that I understand it correctly.) – Martin Rattigan May 11 '21 at 16:10
  • The DTZ metric is just the distance to zeroing the ply count or mate with a fixed n move rule. E.g. the Syzygy tablebases are DTZ50. – Martin Rattigan May 11 '21 at 16:14
  • The wins are of different length depending on if there is an n move rule in place or not and what n is. Obviously whether the resulting endgame you mention is a win or not is also dependent on that. This gives a fuller explanation: http://galen.metapath.org/egtb50/ – Martin Rattigan May 11 '21 at 16:24
1

A more comprehensive list of endgames sorted by the number of moves it could take to mate (based on Nalimov tablebases) is at http://kirill-kryukov.com/chess/longest-checkmates/longest-checkmates-sorted-by-length.shtml

These vary in length from 1 move to 262 moves. It does not include the mate in 549 because that endgame has 7 pieces on the board and the tablebases only include positions with up to 6 pieces on the board.

As for KBN vs. K, that can be done in 33 moves.

Jimmy Vermeer
  • 11
  • 1
0

The following KNN v KPP endgame position is one: enter image description here

Ply count=0. Black to play. White win without 50 move rule. Draw under 50 move rule.

This is the third position in the "two knights v pawn" section of Averbakh's "Knight Endings" (Batsford edition) where it gives a White win in 17 moves, but Black's defence can be improved.

Martin Rattigan
  • 482
  • 3
  • 5
  • The game may be winnable without the 50 move rule in effect but not wit the 50 move rule in effect. This would necessarily be the case if there were a forced win without the 50 move rule in effect that's longer than around 6000 moves. Haworth of Reading university is predicting longest mate over 5000 moves already in 10 man endgames. – Martin Rattigan May 11 '21 at 14:49
  • See http://centaur.reading.ac.uk/36276/3/HaworthLaw.pdf – Martin Rattigan May 11 '21 at 15:00
  • @Brian Towers. I thought the section you deleted was useful at any rate. If the 50 move rule applies it applies from the outset. The WFCC places no limit on the number of pieces in an endgame problem and I don't expect anything but computing resources to limit the number of pieces for which people will try to create tablebases. – Martin Rattigan May 11 '21 at 16:42