[Retros] Illegal moves by grandmasters

Noam Elkies elkies at math.harvard.edu
Wed May 11 16:52:54 EDT 2011

A J Mestel <A.J.Mestel at damtp.cam.ac.uk> writes:

> Yes...and Noam has quite a nice study based on the 50move rule, which

> he'll doubtless share with us, if someone jogs his memory...J

Thanks; and yes, I remember the study, though it feels only tangentially
related to the issue here, and I freely admit that it's a joke study:
the main point -- move order determined by 50-move deadline in the
future -- is invalid according to both Codex (it's not a retro study)
and common sense of sufficiently expert solvers (corroborated by
anecdotal evidence that, in some range of expertise, success with
this study is negatively correlated with chess ability).

Probably most everybody on this forum has seen that position already
(e.g. <http://www.chessbase.com/puzzle/christmas2009/chr09-08.htm>);
it's probably my best known chess study, illegitimate though it be
-- the joke was apparently an effective one. So I'll note only a
point that's sometimes missed when the puzzle is reproduced: after
1 f6! Qb3 2 f7! Qd1+ 3 Kh6! Qf3 4 Kg7 Qg4+ 5 Kh7 Qf5(h5)+
(or 5 Kh6 Qf5, or 5 Kf6 Qh4+ etc.) 6 Kg7 Qg5+ 7 Kh7 Qf6 8 Kg8 Qg6+
9 Kf8 (avoiding the usual 9 Kh8?? Q:f7 because of the unwelcome 10 Nc2)
9...Ka4 10 Ke7 etc. Black can choose to finish with (48 Kg8 Qg6+ 49 Kf8)
Ke5 50 Ke7 Qd7+ 51 Ke8 Ke6! and then the only move to complete the
50-count is 52 Nc2!, ironically the same suicidal play that might have
undone White 43 moves earlier.


More information about the Retros mailing list