Can the longest win be even longer with 50 move rule?

[deductor11]

Are there any know examples, where the 50 move rule makes the length required to win a position longer than what would be needed without the 50 move rule?

From a theoretical perspective the 50 move should in general be disregarded, however in endgames with pawns it might (at least for historical reasons) make sense to make table bases with buildin 50 move rule. This could affect the evaluation of certain positions. In general more positions will be drawn (the 50 move rule only benefit the defense), however for some positions the presense of the rule might make the shortest win (much?) longer than it would be without the 50 move rule!!

Such examples, would - I think - be curious, counter intuitive and would be of general interest for the chess community!

[guyhaw]

Yes, there are examples of positions where the 50-move-rule prolongs a win rather than turns it into a draw.
Marc Bourzutschky suggested that DTZ50 Endgame Tables (EGTs) be investigated, prior to DTR and DTZR EGTs being generated:
DTZ50 assumes 50-move draw-claims are made if possible, but otherwise is like the DTZ metric - "depth to zeroing of move-count", i.e. to mate, conversion or pushed Pawn.
See http://www.is.reading.ac.uk/people/G.Ha ... blications, Tamplin/Haworth (2003) or MB/JT/GH (2005): examples have been given, as has the absolute and %-effect in terms of delayed wins and new draws.
QP-Q1 = 8/8/1P5Q/1K6/3q4/8/5k2/8 w is an example, q.v. http://chess.jaet.org/cgi-bin/dtx ... 1.b7 and DTC=51m draw; 1.Qg5''' and now DTZ50 = 50.

The DTR metric is the length of the longest phase of play when DTR-minimaxing is used. With a DTR strategy, one limbos under as many k-move rules as possible. This avoids the need for separate DTZ51, DTZ50, DTZ49 etc EGTs.

g

[deductor11]

The DTR metric is the length of the longest phase of play when DTR-minimaxing is used. With a DTR strategy, one limbos under as many k-move rules as possible. This avoids the need for separate DTZ51, DTZ50, DTZ49 etc EGTs.

Not sure I understand this? Do you mean that the penalty is not the number of moves to mate, but rather the maximal number of moves without captures or pawn moves, needed to win?

In your example

See http://www.is.reading.ac.uk/people/G.Ha ... blications, Tamplin/Haworth (2003) or MB/JT/GH (2005): examples have been given, as has the absolute and %-effect in terms of delayed wins and new draws.
QP-Q1 = 8/8/1P5Q/1K6/3q4/8/5k2/8 w is an example, q.v. http://chess.jaet.org/cgi-bin/dtx ... 1.b7 and DTC=51m draw; 1.Qg5''' and now DTZ50 = 50.

the DTZ50 column just displays question marks. In the position 8/8/1P5Q/1K6/3q4/8/5k2/8 w how many moves are there to the mate under the 50 move rule? Not quite sure how this follows from the table?

[guyhaw]

Metrics like DTC, DTM, DTZ (which ignore the 50-move rule) are bound to give depths no greater than the DTC50, DTM50 and DTZ50 metrics respectively - for the simple reason that they are not constrained as the DTXk metrics are.
Generalising, the smaller the 'k' of the k-move rule, the greater the constraint, so as k decreases, DTZk does not decrease but tends to increase, at least until the 'win' turns into a 'draw'.
The Tamplin/H and Bourzutschky/T/H papers measured the cost of the 50-move rule in terms of both % of wins which became draws and % of wins which are delayed in DTZ terms, not in DTM terms.

Note that if 'White' (with the win) is minimising DTZ (or DTZ50) and Black is maximising DTM, the number of moves to mate will be >= dtm, the Depth to Mate of the position. However, if Black is maximising DTZ50 rather than DTM, the moves to mate may, as a side-effect, be reduced rather than increased. Here is an example:
KBBKNN: 8/8/6n1/8/k3BB2/8/n1K5/8 w: dtm=67m, dtz=1m, dtz50=28m, 1.Bxg6 {optimal under DTC/M/Z but now DTZ=54m}, so rather ...
SZ50- v SZ50+: 1. Bd6" Nh8' 2. Bc6+" Ka5° 3. Kb3" Nc1+' 4. Kc4" Nf7' 5. Bc7+" Ka6° 6. Bd5" Nh8' 7. Bf3' Ng6' 8. Bd6" Nh4' 9. Be4" Ne2' 10. Bh2" Ka5' 11. Bc7+' Ka6' 12. Kc5' Ka7' 13. Bd3' Ng1' 14. Bg3 Ng2' 15. Kc6' Nh3' 16. Bf1' Nhf4' 17. Bf2+" Kb8' 18. Bb6' Ka8' 19. Ba6' Kb8' 20. Bc4' Nh5' 21. Bc7+' Ka7 22. Be5' Nhf4' 23. Bd6' Nh5 24. Kc7' Nf6' 25. Bc5+' Ka8° 26. Bb5 Nd5+' 27. Kc8" Ne1 28. Bc6#' {but note that dtm was 67m originally, so Black has conceded more in DTM terms than White}.

Note also the question "how many moves to mate under the 50-move rule?" requires the DTM50 metric, which minimaxes mate subject to not allowing a 50-move-rule draw-claim (which is assumed to be made). It cannot be inferred exactly by following 'SZ50- v SZ50+' lines.
John Tamplin's DTZ and DTZ50 EGTs seem not to be online at the moment: I've asked about that.
g