Win, Draw, or Loss? You Decide :)

So the truth is, this problem was "retro" in more ways than one... it was composed without the aid of computer analysis!  Fortunately, I don't think there is nearly as much to look at in Position 4 as there would be in an ordinary chess position. The [very short] line I saw that convinced me White wins goes like: 1.Rxe8 (it's provably White's move!) Nf6+  2.Kxe7 Nxe8/Nxh7  3.Nxa7  and by my count, Black is coming up [at least] a full piece short of legitimate drawing chances.

Thank you very much for the feedback regarding the problem. And I agree with you - the difference made by having multiple capture and promotion possibilities (which this one didn't have) is, in terms of difficulty, a very real thing indeed in my experience.

FIFTH DIMENSION

Cobra91, as you may recall, I mentioned that 2 out of 5 motivations were essentially the same which is a bit of a pity as the others were all different. I found a way to separate the Bishop and Rook substitution as well, though there remains a similarity in the end.

In this version there is still "placement room" as the white knight is in a somewhat arbitrary spot and may possibly be removed by a capture. Some of the pawn positions aren't fixed either.

Note on the conventions: By default "white starts" unless can be proved otherwise!

^ Clearly, your standards are far higher than mine. I'm ashamed to admit it, but I was satisfied once I'd put together a position that was sound, and met the requirements of the construction challenge. Trying to make changes and fulfill even loftier criteria was something I'd never have had the patience for.

With that being said, I view the similarities your new position has with the one from post #28/41 as [by far] the greatest compliment I've ever received regarding any of my compositions, so thank you very much for that! And the deeper analysis needed for queen's rook + light bishop replacements makes this worthy of publishing considerations, I'd say. 

EDIT: Just realized the situation here has become analogous to what happened with "The High Road". I'm not even willing to show what my own solution for that one was, after seeing the wizardry made possible by superior motivation and ingenuity.

Thanks once more for your appreciation! And indeed it is also a compliment to #28/41 where you drew up a very promising scheme which I was glad to copy. It is sort of a joint composition which also satisfies my criteria for being "publishable". Personally I would leave it in the drawer for a while until I was completely happy with the efficiency and effectiveness of all the details but I wouldn't venture to add new ideas any more unless I were to stumble on them. I leave it all up to you though. It was your challenge and it is appropriate you decide its fate!

The challenge itself was not my idea, actually - it was presented somewhere in these forums a number of years ago, by a member whose handle I don't recall. My initial attempt was dreadful; it was poorly conceived, poorly built, and e.p. status couldn't be truly verified in all positions. A little more recently, I became fascinated with the idea of a proper and absolute solution. Thanks to a sounder concept, I soon wound up with the position from the 1st post (which also made use of e.p, but was elegantly bulletproof).

Getting a completely different idea to work later on in the thread was just a bonus for me. I'd be perfectly happy to see your adaptation published somewhere, whether I received any "credit" or not.

I might feel differently about something similar happening with the other position (post #1), though. If that one was ever adapted into something "publishable", I'd really want the original version to be cited at some point, along with the web page where it can be found (1st page of the thread). 

The cheapest constructions:

^ Unimpressive. Since White is not provably winning in either position (without the help of a convenient convention for chess problems), they merely illustrate why the position from the 1st post was superior to all of those that followed. The task is easy when relying on unnecessary extra rules to fill in the retro-analytical gaps.

So to "top" my 1st construction, you'd need (at the very least) to ensure that White has no legal retro-moves in the diagrammed position (the one with 2 rooks and 2 bishops for White).

I wasn't trying to top anything! I merely showed a new, rather easy way of fulfilling the requirements in a very efficient manner. Most problems do not top other problems but they show different dimensions of implementing a theme. That's why judging them in contests is as difficult as comparing van Gogh to Picasso.

The requirement that the position must show all the "state" information was never in your original challenge (while it was in your highroad challenge!). Apparently you felt the need to add it later, because - not knowing the conventions - you believed it to be part of the challenge given to you. Which I strongly doubt, because it is an uncommon assumption for problems with forward stipulations. Just as the opposite is the case for purely retro-analytical problems, which your challenge clearly is not.

There are very good reasons why problems with forward stipulations always rely on the retro-conventions. One of the most important retro-active properties - castling right - can almost never be proved by retro-analysis. You can delete 99% (no exaggeration) of all hybrids when you don't accept the active presence of the conventions. Of course, the fact that your problem determines all state information is a compositional plus but in no way invalidates other problems which have their own plusses.

I can buy that you are not impressed with the last 2 problems - they were composed within an hour to show an idea - but I don't buy that your challenge is realistic without the retro-conventions. That would probably restrict the final entry list to your product plus its clones. Not very exciting.

I find that very surprising, since your need [and ability] to top everything I've ever done is more or less the sole reason why I've permanently ceased all activity on all of my current or future composition-related ideas and projects. I can't really afford to spend 1-2 hours a day on stuff that will just get usurped within ~12 hours by some pro with more experience and natural talent.

I suppose some leniency would be needed for the forward-only problems making good use of castling (they should only be considered "busted" if one can prove that castling rights do not exist), but in all honesty, I despise most pure retros that make use of the convention. It's too convenient, from a retro-analytical perspective, because there are so many amazing things that can be proven based on castling rights alone. Any retro-related content made possible by "automatic" castling rights had better be earth-shattering, as otherwise the composer is just taking the easiest/cheapest route to a high-quality product (in other words, my first thought is usually, "I could do that... if I was willing to stoop to that level  ).

Yes, and you can delete 99% (no exaggeration) of all non-miniature SPGs when you don't accept the necessary presence of minor move-order transpositions which have no effect whatsoever on the [usually sophisticated] idea being shown. I just posted the solutions (1 or 2 days ago) to my two SPG puzzles from the "proof game" thread, and I suspect their ideas cannot be implemented in an "acceptable" way under current regulations... which is a shame, because the concepts there have some serious promise, in my opinion.

In any case, it seems there are inconsistencies in the rules, and it's probably because those rules are dictated by the random whims of problemists. 

 Well, this is what I don't buy (you've already said what you can and can't buy  ). For one thing, I'm guessing it's a fairly trivial matter for you to rearrange a few pieces in the above positions, and then start adding pieces (meaning pawns ) until White runs out of retro-moves.

Also, how do you define "clones"? If any attempt at all relying on e.p. is automatically a "clone", then you're just being silly. Otherwise, it ought to be mere child's play for any seasoned pro to devise a brand new scheme (with e.p.) which allows the challenge (along with what you call "extra" criteria) to be met by a queenless and checkless position. I've seen enough professionally composed problems to know that real pros can do anything they want, and can almost always do so easily, simply because "they're just that good". Being a pro yourself, you're even more aware of this fact than I am.

@A: Note that in most cases I topped your problem with your own problem and not with another problem. You did not exhaust the technical possibilities of a particular approach enabling me to expand on it. Professionals would kill themselves for overlooking simple ways to improve their products. Your (and mine) castling problem does however not "top" your e.p. problem or vice versa as they are incomparable. It's a beauty contest!

@B: I gave "castling" as an example because it is an extreme, but the same applies to "who is on move", "repetition", "50M-rule" and often "e.p.". Did you know that almost any position could be a draw in the diagram because you cannot prove retro-analytically that repetitions did not take place and no 50 moves had passed without capture or pawn move? Hybrids simply could not exist without the conventions. Hybrids need to balance forward and backward considerations and therefore cannot meet the same strict requirements as pure retro-problems.

Pure retro-problems that make use of a convention are a contradictio in terminis. Because they have no forward stip they never rely on conventions. They simply are what they are. It's OK to despise things that don't exist. Personally I despise most the barber who shaved (only) all men in the village who didn't shave themselves. 

@C: Who is inconsistent? You demand the elimination of the complete field of hybrid retro's because you don't like the conventions, and at the same time you complain about the restrictions on the rich field of SPG's because it isn't a 100 times richer. Do I see some bias here?

@D: To settle this, you could look up the original challenge and find whatever constructions it yielded. I would be flabbergasted if they met your standards and be essentially different from yours. Whether or not they are "clones" is not a matter of mathematical definition but a "feel" that most pros would agree about. Composing is science and art in equal proportions.

is this a 960 game?

It's a standard game. Nobody makes puzzles for chess 960 (at this time).

@cobra91:

Addendum: ON RETRO CHESS CONVENTIONS

A. The "retro" chess conventions were designed to resolve retro-issues in common checkmate-problems and endgame studies. They were not designed to set up special policies for handling retrograde problems.

B. All conventions yield to proof. There can never be a conflict between what is proofed and what is conventional.

C. With regard to conventions, hybrid retro-problems (having forward stipulations) are indistinguishable from standard checkmate problems and studies. It doesn't matter whether a problem is 90% retro and 10% forward or the other way round which is kind of neat.

D. The application of the conventions is even the same for purely retro-analytical problems as will be explained in the next point. The difference with the hybrids is only that the conventions have zero effect on the analytical type.

E. The number one rule for the conventions is that they only operate when called upon. Just having a diagram with a king on e1 and a rook on h1 does not imply castling right. For instance when white plays Ke1-e2, the castling right will never be determined. The convention on which color starts the solution will never be set in motion without forward stipulation. Therefore all conventions are ineffective on purely retro-analytical problems.

F. Conventional operations start with the "Can I" question: "can I castle?", "can I continue play without draw?", "can I start the solution by playing a black move?"

G. When the answer by the "convention evaluator" is "YES" then the decision is made (the move played) and the property state is manifested. The action determines in retrospect that "white was on move", "black had castling right" or "the position was not drawn by repetition".

H. Deciding states leads to a reduction in proof games. Obviously at least one proof game must always remain for the problem + the solution and such is guaranteed by the "convention evaluator".

I. All the above is not very exciting until conventions start colliding. For that situation some extra retro-conventions were created which will not be discussed here. Suffices to mention that this hardly ever happens in standard mate problems and studies but there are hybrid composers who explore that domain (like me). The basic understanding of convention handling is critical in resolving collisions.

Okay, fair enough. If I'm being honest, stuff like 3-fold repetition and the 50-move rule is the sort of thing that I generally just dismiss in the context of hybrids. And for good reason, which is exactly your point, I suppose.

Sweet! Then the next time I see a pure retro that begins with "Both sides can still castle...", I'll know where I can tell the composer to go shove his nonexistent composition. 

I wasn't demanding the elimination of any field. On the contrary, I was suggesting the expansion of a particular field. And to argue my case, I was also pointing out that problemists like to be ludicrously strict when it suits their personal tastes and preferences, but not when it doesn't.

I can't really do that, as the original challenge (which I did not come up with) is now buried somewhere deep within chess.com's "forums graveyard", and the thread title was a bizarre misnomer that I seemingly can't remember (it was a number of years ago, in my defense). So I have no way to track it down, anymore. 

That shouldn't matter, however. No professionals ever bothered to get involved in the challenge, so you ought to be able to blow all previous submissions out of the water...without even needing to be awake, I'll bet. 

@A: Repetition and 50M are of particular interest in hybrids. The repetition type is one of my specialties. It is fruitful by its relation to "premature repetition" and "DR". 50M problems are made by a number of composers, notably N. Plaksin.

@B: Professionals share your dislike of presetting states (like: all castlings are possible) beyond what is offered by either retro-analysis or the retro-conventions, and they only use it scarcely. This extra stipulation stuff is considered cultural and not conventional and can be used in combination with both the analytical type and the hybrid type. The formal conventions are only in the Codex for Chess Compositions and several of its articles concern the automatic determination of unknown retro-active states.

Much of what we discussed earlier in terms of stipulations is not prescribed by formal conventions but is fuzzy as in habit or culture. Editors will not in general refuse publication of a problem with an unusual stipulation unless it is entered in the wrong category or is of evidently low quality. But there are popular and unpopular genres at any time and there is a (subjective) perception of when a stip is inappropriate in a particular presentation.

I don't know what "premature repetition" and "DR" mean, but if I had to guess, I'd say "DR" probably refers to some sort of alternative/fairy rule involving dual repetition instead of 3-fold repetition. I once wrote a simple proof that 3-fold reps could never be proven by retro-analysis, because no position can be forced to occur more than once in either the retro-play or the forward play. So to make use of the "draw by repetition" rule (in the hybrid context), the relevant condition would need to be a dual rep, rather than a 3-fold rep.

I've seen Plaksin's famous mate in 3, but had no idea there was an entire genre devoted to the theme. I've always considered that one problem alone to be the Holy Grail of ALL forms of chess-related composition, literally of a class and caliber all its own. The thought of many people composing at that level on a regular basis is utterly mind-blowing to me! 

Aha!   So in other words, my two problems from the proof game thread (I know there were actually more than two, but I'm referring to the ones that nobody solved, and that I recently posted the solutions for.) are NOT automatically phased out of existence by formal rules and regulations. Thank goodness you finally clarified this, or I'd have continued to believe they were officially null and void, all based on your personal opinions regarding the PG genre! 

Since you didn't give an example of colliding conventions, I had to quickly assemble one myself.   You probably just assumed the topic would be over my head (or something like that ), but can you at least confirm whether or not the "problem" (I know it doesn't count as a "real" problem ) below constitutes a true "collision" of the standard conventions? And if it does, how would such an issue be resolved? I want to learn about those "extra" conventions! 

@A: "premature repetition" occurs when a "proof game" has a 3-fold repetition before the 3-fold repetition that ends the game in the diagram. That proof game is illegal as the first repetition should have ended the game automatically. The most common example is an SPG  with the stipulation "drawn after n moves" which gives the additional information that the game was drawn by a convention dictating automatic draws. Because, as you noted, the 3-rep cannot be proved, it may be justifiably provided as additional state information in the stip in the way indicated. Btw, 50M and DR are alternative sources of draws!

DR = dead reckoning, based on the game rule that dead positions (unavoidable draws) immediately terminate the game. Search "Andrew Buchanan" if you want examples.

@B: There are not that many people composing hybrid problems as the retro-world does not understand the logical principles underlying them. They get confused when things get interesting (see the next post). There are at least 10 times more SPG's than hybrids.

@C: If you read my comments carefully, you can conclude that I don't like the free order type, that nobody makes it and that it doesn't fit the standard SPG definition, but not that they are somehow outside the legitimate retro-domain. We even tolerate Smullyan!

This also replies to the issue of "duals". In a "proof game sense" your solutions have loads of duals but in the "set of moves sense" they are unique. The duals are sort of a categorical imperative in your type and as such justifiable. A disadvantage is that you could never have people solve them in a puzzle interface since no current puzzle interface is capable of supporting all the move permutations (though it's easy). Back to pen and paper!

Your example is sort of a convention collision (well done!) but is generally not considered as such except in one logic. The assumption that "white is on move" is considered stronger than the assumption that "black can castle" and this decides the issue. When colliding assumptions are of equal preference - like the right to castle vs the absence of e.p. right - then the collision logics are set in motion.

I am writing a book on this subject and I know for certain that I am the only person in the world understanding it at this time. I have proved it by composing and publishing a number of problems nobody understands though a few people can solve them intuitively. I will not even begin to explain it though you can find a few very long threads about the subject on chess.com plus a critical example. Discussing basic things with you requires numerous posts (no criticism; it's your lack in experience and intuition, not intelligence) and this one is 10 times more complicated. You must forgive me for not starting on this trail. Besides, understanding it would only get you in trouble with the rest of the retro-community which is still fooling around in the kindergarten.

Btw, you can find the extra conventions in the forementioned Codex, articles 16 -18, even though these articles are incomplete and confusing in their current form as you might expect from kid stuff.