[Retros] Minimal Deviation from the Rules (MDR)

[Rol, Guus]

> I think the basic principle should be that the solver is required in

principle to demonstrate whether any position, even a fairy position, is
legal

That is a strong requirement, probably too strong. How many Maxinummers
will become illegal for the absence of a maximized last move? I'd favor
the requirement in a retro-environment only, meaning (a) the stipulation
is of a retrotype (b) the presentation context is retro like the
retro-section of a magazine. Or may be you already assumed this?

Guus Rol


-----Oorspronkelijk bericht-----
Van: retros-bounces at janko.at [mailto:retros-bounces at janko.at] Namens
andrew buchanan
Verzonden: donderdag 7 februari 2008 11:36
Aan: The Retrograde Analysis Mailing List
Onderwerp: Re: [Retros] Minimal Deviation from the Rules (MDR)

Hi Roberto,


>Circe problem. Mate?


I think that the intended answer is "No." Black has just captured Na2,
which must re-appear like Duncan Idaho on b1 to complete the move and
block the check. White's prior move can only have been discovered check.

The key question to be resolved in solving the problem was whether you
intended overlooking check or mate to be a "legitimate" error. The
answer is no, because e.g. otherwise Black might just have captured B on
a2, and the problem would be cooked.

However this meta-reasoning can now be applied to all other problems of
this form, which are therefore easier.

I have to confess an aesthetic preference for problem forms which are
"self-defining". With MDR, it appears that there is a codification
required for each form as to what "opportunities" and "incompletions"
are legitimate. I have a related kind of trauma about lexical problems,
which depend on the language used to record the moves. My problem,
sorry.

I applaud the attempt to combine fairy + FIDE rules however. Some fairy
forms (e.g. Circe, maximummer) naturally admit retro logic all the way
back to the game array. 

Others (e.g. with weird units) do not admit retro logic. We can add the
convention that the game is otherwise orthodox, but there are extra
options for pieces to promote to. That convention is an elegant idea
which I hadn't come across before. If it is already widely adopted, or
if it formed part of the pre-announcement to the German tournament
referred to earlier, then it's totally OK to form part of the judging
criteria for this tournament. Not sure I would want to see it as a
default convention though.

The Codex is not very helpful about all this, stating baldly that fairy
problems have no history.

I think the basic principle should be that the solver is required in
principle to demonstrate whether any position, even a fairy position, is
legal. If the position is legal, then it can be assumed by the solver
that it derives from some history (although no reason to assume one
history over another). If it is illegal, then for a fairy position there
is no issue. For an conventional position, it acquires the property of
fairiness, but there is otherwise no issue.

Best,
Andrew.