[Retros] A question on stipulations
andrew at anselan.com
Sat Nov 21 08:57:43 EST 2009
A few more thoughts on this...
(1) Most commonly when the stipulation is "Mate?" we are talking about a conventional RA problem where the interest is in figuring out the history of the game. It is an additional stylistic flourish if the game ends dramatically with a mate, and the composer calls attention to this in the stipulation. (It seems to me that a mere check in the final position is regarded as a slight defect, but a mate paradoxically is not, because of its finality.)
(2) Unless there is an hint of "funny business" (e.g. e.p.) it does make sense for the solver to explore first the more promising avenue that the position is legal, and therefore hunt for a proof game or perhaps just a way to untangle the position. I am not sure I like the word "assumption", because it's a bit ambiguous - if I *assume* the position is legal, why would I then have to *prove* it with a proof game? If I want to prove that the position is legal, I just resort to my assumption!
(3) Renny wrote:
> In the first place a non-existence proof
> is much more difficult than an existence demonstration (typically
> requiring computer assistance to be certain), and in the second place
> such a proof is much less satisfying than a game with some ingenious
> twists and turns.
A non-existence proof could be very short. E.g. 2 White kings. Or a parity argument. Or an unpromoted piece having escaped from a pawn cage etc etc. And such an argument could be quite satisfying. However the thing is that (except for the specialist topic of Illegal clusters) retro enthusiasts don't really like *totally* illegal diagrams. We would like there to be some escape clause, some unlikely scenario which we are forced to conclude is actually what happened.
The famous quote from Sherlock Holmes applies: "How often have I said to you that when you have eliminated the
impossible, whatever remains, however improbable, must be the truth?"
----- Original Message ----
From: Mario Richter <mri_two at t-online.de>
To: The Retrograde Analysis Mailing List <retros at janko.at>
Sent: Sat, November 21, 2009 8:35:03 AM
Subject: Re: [Retros] A question on stipulations
> the answer might be "No", because the side which seems to be mated could
> have a right to capture en passant.
Other possibilities are: 50-moves-rule, parity, retro-opposition, ...
Reduced to their logical kernel, those problems simply ask
the question: "Is this position legal with a specified side to move?"
so they are in good company with similiar types like "Mate in one?",
"Who can mate in n?" a.s.o.
Renny> if ... the proof is an elaborate explanation of why every possible
Renny> attempt at resolving the position leads to a dead end,
Renny> would they publish such a problem?
(Just a thought experiment: Replace "Is Black mated?" by "Can White castle?".
Would you still argue, that if an elaborate proof shows, that all attempts
to resolve the position with preservation of White's castling right leads
to a dead end, then the problem is not worth publishing?)
Retros mailing list
Retros at janko.at
More information about the Retros