Another checkmate in 1

It's a mate in 3, en passant can't happen

Continued from post #20:

I've seen the problems from your post (the first one is wrong but that is not the point). Yes, these are pretty simple retrograde problems with e.p. - based on retro-analysis but not on a convention. Like having to release a complex position in order to justify a "mate-in-1" solution. The mating move requires no convention but it's nevertheless a retro-problem due to the expected analysis

Interestingly, Andrew Buchanan (who I commonly disagree with) made the precise same analysis a few years ago (I did it 16 years ago). He said that the retro-conventions overlap the legality law/convention since it is unnecessary to have a convention for things that are "certain" and leave no choice due to inconsistency (example first paragraph post #20). So he wrote to me that he would redesign the conventions to eliminate the "certain" states. I don't know if he actually did the rewriting but he announced it. That would remove discussions like ours about whether an e.p. move is permitted by convention or by the consequences of the legality law!

In my view the overlap is a futile issue as the rewrite effectively changes nothing but theoretically Andrew's view coincides with what I wrote about mathematical Model Theory!

Btw, do you recall my "#1 or #2" retro problem? That one only worked by the power of the e.p. convention. It's not that common.

Your distinction between applying retro-analysis and applying a convention seems odd in this case, because the e.p. convention is exactly about applying retro-analysis. Here's the e.p. convention we've been discussing, directly quoted from the WFCC. https://www.wfcc.ch/1999-2012/codex/#b20

En-passant convention. An en-passant capture on the first move is permitted only if it can be proved that the last move was the double step of the pawn which is to be captured.

"Proving" the pawn double-step is obviously done by retro-analysis. So when you apply retro-analysis to the Loyd and the 5 linked problems, you are doing exactly what the convention is telling you to do. You can't help but follow the convention, which is obviously relevant to all of these e.p. key compositions.

Your argument seems to be that we don't need the e.p. convention for this whole branch of e.p. key problems, as long as we use the more general conventions about legality and retro-analysis. But "not needing to apply the convention" is different from "not applying the convention at all, as if it doesn't exist". The reality is that such an e.p. convention exists, and it was created to apply to this type of e.p. key problems.

So your answer to my question is yes, you don't have an issue with this particular Loyd problem, but with the e.p. convention itself – you believe it's redundant for a whole branch of e.p. key problems. Back in post #9 you claimed that there are "countless" puzzles where the e.p. convention is more relevant, compared with the Loyd. You haven't provided any example, though now you referred to your own composition as such but added "it's not that common". I don't recall it but I'm sure it's fine composition that uses the e.p. convention in an original way. Unlike the Loyd and the whole branch of e.p. key problems that use the same convention in the ordinary way.

Previous reply removed for reformatting!

Ah, I have no problem with either Sam's problem or with the e.p. convention - Andrew had some objections - but I noted there is a senior principle that solves the puzzle - the legality principle FIDE law 3.10.3 and Codex article 14 - without referring to a specific retro-property convention or even knowing that there is such a convention.

The legality law+convention is the basis of all purely retro-analytical constructions - these don't permit e.p. and castling and other "missing information" conventions. Position releases are fully legitimated by finding just one proof game! But it is more powerful than that. Not tied to a particular composition type, it insists on finding a proof game for every orthodox composition - retro or otherwise. And it is even more powerful than that. Because it applies to all positions it also applies to all solution moves for all orthodox constructions. So it dismisses all solution moves for which there is no proof game. And the last consequence of the legality convention: you must always allow any move that is 100% certainly legal or a contradiction can be proved in the manner I described in post #20. Which in itself does not lead to a contradiction as long as the "retro-property conventions" are well designed. So far they are with the annotation that the retro-property conventions overlap with the legality law+convention. That is what Andrew disliked if I understood him well. And which implies that you can solve Sam's problem with either one of hem!

The e.p. convention is defensive (my preferred term is recessive, Andrew's: pessimistic). It is not necessary for Sam's problem and all those you found in the pdb but it is necessary to shield all the twomovers and threemovers and endgame studies with pawn formations and histories with and without e.p. right. They are the countless number of puzzles that hang by a thread from the e.p. convention. There are also problems where the e.p. moves become visible (even when "no-e.p.-right" is possible) when they are tied to dependent retro-properties like in PRA and AP logic. In the "#1 or #2" composition I mentioned, the e.p.move is the thematic try. It is really hard to disprove the e.p. right in that one! I'll look up if I showed it to you or somebody else. Like many of my compositions it was never published in the mainstream composition media!

Good to know what you meant by the "countless problems" where the e.p. convention is more relevant than the Loyd. In most of these problems, the e.p. moves are purely incidental, potential cooks ruled out by the convention, and not relevant to the actual problem content. So it's ironic that in such problems where the e.p. moves are disallowed, the e.p. convention is relevant, according to your system. Also, this means that if the Loyd problem was badly constructed and the pawn double-step isn't proven, then for you the convention suddenly becomes relevant, to indicate the problem is unsound.

Oh yes, I remember Andrew calling the e.p. and castling conventions "pessimistic" and "optimistic" respectively.

Publishing originals in forums can be messy indeed. A while back I decided to find all my online originals and bookmark them. Less of a headache if I decided to make another collection of my compositions!

Yes, I think that reflects my message!

Note that the conventions on 50M-draw and 3rep-draw and "on-move-side-change" are also "pessimistic". There are almost always proof games for common (com)positions where the "optimistic" state is possible and not disprovable so you need a pessimistic convention to keep them out. You can also make joke problems which ask for the "safest way to win from a diagram". The solution is a move which e.g. gets rid of a possible 3-rep draw as soon as possible without throwing away the win! Some posters on chess.com make those jokes unintentionally

When Andrew is done the e.p. convention might look like this: when histories for a diagram exist with both e.p. right and no e.p. right, then you are not permitted to play e.p. This definition isolates the "uncertain" retro-property states and leaves all the provable states to the legality convention.

Ha, Ha I noticed I am confusing "optimistic" and "pessimistic" myself since they have an emotional connection. In the case of e.p. it is clear that there is an action versus no action. But for instance "black on move" and "white on move" have the same logical relationship and the same mutual exclusivity but are hard to classify in pessimistic and optimistic by action. It all depends on how exactly you define the retro-property involved and its default value. Another time I'll rewrite it.

Hey Arisktotle, I have an honest question: this problem got linked to on CG today, I tried solving it and consecutively found this topic and your post. I have no doubt it's correct, but I also having an issue with understanding the flaw of the rook having no retraction move. I'm by no means a retrograde expert (I just do them for fun) but it does intrigue me.

From what I understand we try to prove Kc8-c7 was not the last move, and the rook on h8 has no retraction other than Black being in check. However, does that matter, given we try to prove the move was not played (a former position is illegal so it can't be the last move)? I tried searching up the relevant terms (retraction, retro counters) quite a bit, but somehow can't seem to find the convention where this specific example applies. Would appreciate it you can enlighten me, because this really fascinates me!

Not sure why Arisktotle hasn't answered this, but I'll try to explain. The "flaw" of the rook's placement is not so much a defect in the problem, but rather a missed opportunity to give the composition a little bit more content. Whether it's h8-R or g8-R, Black's last move couldn't have been Kd8-c7, but with g8-R, the retro-analysis is less obvious because of the potential previous move, Rh8xg8+, which is ruled out for additional reasons. A general principle in composing problems is "economy of force", which basically means you want to present an idea as fully as possible while minimising the pieces used. So here having the R on g8 instead is like adding an extra variation to the composition for free, without needing to add extra material.

Rocky explained it correctly. It is an aesthetic flaw, not a technical one! You might say: "a missed freebie"! I don't blame Sam Loyd for any shortcomings in the puzzle. Different era, different standards and objectives. But modern readers might be interested to see where we are now.

So this issue is not about conventions though we discussed the question on whether the e.p. capture is "conventional" as well!

What if black promoted to that dark squared bishop? So the f8 bishop never needed to get out?

Yes that's possible. It however also loses 1 black unit in the overall piece count and makes it impossible to retract everything to the Home position - if I counted well!

@Rocky64 & @Arisktotle

Thanks, that helps. Learned so much about these type of studies from this thread alone, so appreciated!