The FIDE 50-move rule: precise wording

[guyhaw]

See http://www.fide.com/info/handbook?id=124&view=article ... article 5.2e.

e) The game may be drawn if each player has made at least the last 50 consecutive moves without the movement of any pawn and without any capture. (See Article 9.3)

9.3 The game is drawn, upon a correct claim by the player having the move, if he writes his move on his scoresheet, and declares to the arbiter his intention to make this move which shall result in the last 50 moves having been made by each player without the movement of any pawn and without any capture, or ...
the last 50 consecutive moves have been made by each player without the movement of any pawn and without any capture.


So, say Black is defending and is about to move. If Black made the first move of the current phase and both sides have played 50 moves in this phase, Black can claim the draw, regardless of its next move (even if this is a forced capture or P-push to loss, in which case the phase might have started with DTZ = 50 moves, 101 plies, and therefore is a draw). If White made the first move and Black is forced to capture on the 100th ply, DTZ might have been 50 moves and this is a win.

This is the problem with measuring depth in winner's moves, and I don't think there is a simple way, post-hoc, to convert depth in plies to depth in moves, or vice-versa. One would never know whether a forced phase-change to loss was 'on' or not.

A comparison of Wirth's maxDTC positions and 'Nalimov's' maxDTC positions, courtesy of Marc B's code, showed that the set of maximal positions was different, especially when White was dominant (KQQQK) and tended to force the capture of one of its men.

g

[notnale]

Wasn't this problem already addressed in the dtx thread?

[syzygy]

Yes, this was addressed in the dtx thread.

In border cases a winning position can be drawn by a delay of a single ply, or a 50-move rule draw can be lost by a speed up of a single ply. In these border cases dtz50 must be known in plies, so that optimal moves can be distinguished from moves that are one ply "less good".

If dtz50 is stored in moves, then dtz50 can still be reconstructed in plies provided that both the wtm- and the btm-table are available. However, if WDL50-tables are already available, it seems to make more sense to only store the smallest of each (wtm, btm)-pair of DTZ50-tables (and thus in plies).

For tables with maxDTZ50 < 50 (in moves), there seems to be no problem in storing dtz50 in full moves, even when only storing one table of each (wtm, btm)-pair. (It can lead to slightly suboptimal play after play has already been suboptimal. That seems acceptable.)

WDL50 must be stored for both wtm and btm, unless maxDTZ50 < 50 in which case it might be acceptable to only store the smallest of each pair (wtm, btm). Of course, storing only one side increases access time significantly.

[notnale]

Storing WTM only in plies would take a lot less space then storing both WTM and BTM in moves