Unbelievable to me

You'd be surprised how difficult those "little glitches" can be to correct!  I appreciate the encouragement, though.

Whoever comes up with a problem in its final form is, in my view, its sole author. It's obviously a different story when the joint effort is arranged before commencing work on the problem, but such is not the case with any of my compositions. Anyone finding a significant improvement becomes the sole owner of whatever may result.

However, as I've said in the past, I am not professional composer, never have been, and never will be. So if you can turn any scrap metal of mine into pure gold, then I certainly won't complain.   If a masterpiece based on one of my old ideas actually gets published, all the better. I'd never expect to receive credit for somebody else's brilliant adaptation of any idea, whether that idea was "mine" to begin with or not. 

Generally, composers are courteous folk to one another and acknowledge the value of both early and late contributions. To recognize that B took a problem by A and went further with it, the problem may be marked: A+B, or A version by B, or B after A or B dedicated to A. If there is bias, it's in favour of the original composer, who gets to say what level of recognition he would prefer to retain. You are free to let go of early authorship privilege, but other composers might not. Having said that there is a lot of replication, and what constitutes original ownership is moot. For example, the motif of a pawn capturing 5 times in a line is a common one to soak up a lot of pieces before the serious retro logic begins. It is these days known as a Volet pawn. Where the pawn makes a sixth capture to promote, it is known as an extended Volet pawn. That's just a name that's been allocated as a kind of honorific: there is no ownership by him of this idea. Basically it's all a bit like musicians taking techniques and motifs from one another: there's a lot of borrowing happens, as we try to build further.

The other thing is that a lot of copying is entirely accidental. If you are trying to make an Black Allumwandlung Circe h#2, for example there's a limited number of ways that you can design that efficiently, and checking out prior art would be prudent

Exactly - which is why there's no point in trying to "draw the line" in an arbitrary place. How can this be done without opening the floodgates of subjectivity and unleashing the inanity of frivolous rule-conjuring debates upon the world? It can't... and in fact, that's one of several reasons for my aforementioned opinion on the matter.

You'll need to give an example I'm familiar with. Since I don't know about any of the more obscure problem types, I can't give a proper reply to comments associated with them.

I'll now toss out something from 2014. Alas, there were none from that year which I particularly liked, but at least the choice was relatively easy.  Only 5 problems to select from (fewest of any year), and 3 of the 5 are... well, let's just say I hate them (for a number of reasons), and they would be thoroughly unsuitable for this kind of setting (due to a whole host of other reasons). Furthermore, I recently busted one of the remaining 2014 problems, leaving just the one below. So (without "cheating" by pulling an extra puzzle from the 2015 archive), I've got nothing better to offer than this [fairly trivial] pure retro, together with its terribly clumsy stipulation:

[Note: The following problem has recently been modified, and now differs (slightly) from the original version.]

Hopefully it is sound, if nothing else.

Also, there's a question I'm sure will be asked, so I'd better just answer it in advance: no, I was definitely not aware of the "release the position" stipulation at the time. Even at present, I can't be 100% confident that "release the position" would have been appropriate for the above composition, although I do suspect as much.

Nothing keeps you from adapting the stipulation now unless yours offers something extra. The formuation in the next post is a common alternative for yours if you don't want to stay with "release the position". An advantage of certain stipulations is that replies from solvers can be computer-checked in a contest. Your stipulation would score very well in that department!

As a note, the stipulations in my early retro's were not always appreciated either and with valid reasons. I felt I had to add some significance to a diagram even when there was no significance to my significance  I'm over it now!

Well Done! The 7 last single moves are fixed. If you wish to vary your stipulation, you might ask for all unambiguous retractions from the diagram. When somebody's got those right, then he did get the solution!

Oh, and there is a law about retro-analytical compositions. When you're finished composing, try playing forward. Quite often you'll find a move or a series with defined retractions which you can add for free! In your problem, two forward moves are retractable and one is quite surprising!

I see only one retractable forward move (Ke2+ works, but it doesn't look like there are any others). However, I see another possible forward move which would fit nicely with a completely different change I had in mind. If I add a Black pawn to a5, is it possible for Black to play gxf5+ in the final position, without busting the intended solution wide open? 

Indeed, '7' was the intended answer (once a certain level of freedom in the play is achieved, it becomes easy to see why no other precise board configurations can be proven). Your suggested stipulation has the advantage of being a lot less clumsy than my original one, though it does give slightly more away about the nature of the solution. In post #87 (after the edits ), it seems you were implying that "release the position" (which seeks to give nothing away, if my understanding of it is correct) would also work, but would not necessarily be the most desirable alternative.

I had indeed gxf5+ in mind and I didn't think the black a-pawn was "dangerous" But I see now it permits the uncapture of the dark squared bishop on the b-file. After adding the a-pawn I can't see how that bishop can appear on another spot than f4, implying the desired last move. But please check it out.

There is a slight disadvantage to putting the wK in check as it clearly gives away which side is on move. But in this position there is no real option for a last black move so the loss is marginal.

"Release the position" is not so much about giving nothing away as it is about encouraging the solver to find out whatever he can about the proof game. Often there are interesting tidbits left after the initial forced retraction series which may be overlooked. Btw, when I made the suggestion, I hadn't yet solved the problem. It is the most generalized RA-stip.

When choosing for a smaller target - e.g. "last move" - chances are that the solver thinks he found the solution while he actually missed the best parts. In this particular problem the unambiguous last moves will do fine. Uncapturing bBf4 practicallly guarantees that the solver saw through the complete unwind.

For most problem types there are running solving competitions in problemist magazines for first publications. Complete RA-solutions - for instance with proof games - are time-consuming to check. Commonly, only the more or less forced move series near the diagrams are accepted and given as solutions to the RA-problem with the stip "release the position". Which means that for your problem precisely the 7 moves will be specified as "the solution", possibly with a rough outline for the "pain points". But one largely relies on the analytical expertise of the participants when it comes to the unravelling of standard modules.

I was already considering the addition of bP on a5, because it adds another conceptual element to the "background" of the retraction sequence (there's a "rude try" that exploits the extra Black tempi in order to free the kings sooner than intended, and it's a try which can't be thwarted by just the capture analysis alone).

The only potential "hole" in the version with gxf5+ involves a promotion by White. The capture count ensures: (1) d6 comes from d7 and f6 comes from f7;  (2) White could only promote his c-pawn or e-pawn, capturing only once with either of them; and (3) Black's c-pawn captured on b1 and promoted. Therefore, it's impossible for White to promote without capturing on a light square... but of course this light-square capture is also impossible, which closes the last hole, I believe.

Yes, that's why I wouldn't think of Ke2+ as a move that is worth adding to the problem. Adding the Black pawn on a5 does create an option for a recently played Black move, but I still think the extra layer of analysis needed to determine what was captured on f5 (when gxf5+ occurred) is worth the cost of "telegraphing" whose move it is to the solver.

Oh - so the question of whether "release the position" is the "right" stipulation to use is much more subjective than it is objective. Then I guess your 2nd suggestion (post #88) is a bit more appropriate for this specific problem... probably. 

When I'm a little more confident in the soundeness of all the changes I want to make (not just those to the caption), I'll edit the diagram in post #86 to include the better stipulation in place of the old one.

My 2015 problem won't be deemed "acceptable", in that case, but oh well. I'll cross that bridge when I come to it!

Let's simplify this. There is only one assignment in pure Retro-Analysis (RA) and that is to find out all you can about the history. it applies to all RA-problems, it is always expected from the solver and it is sufficient.

The variations in the stipulations are of a lower order and serve different purposes:

A. Find a proof game. Seems completely satisfactory but has the disadvantages that it is (a) long and hard to check (b) it doesn't show that the solver has understood which parts of the solution are necessary and which parts are free choices. There is no structure.

B. Release the position. Often very effective, but there is no precise definition of what analysis is required. Usually consists of mainly retraction moves and some notes on the overall release strategy.

C. Efficiently reduced stipulations, directed to particular positions. When these stuipulation are well-defined one may rest assured that the solver is capable of "releasing the position" and "providing a proof game" as well. The choice of the stipulation is usually made by the composer to direct the attention to the core ideas behind his composition. Example: How many promotions occurred on adjacent squares? Without this stip, the solver might not notice that "adjacent promotion squares" was a theme of the RA-problem.

Efficiently reduced stipulations can also be used for effective communication, e.g. "minimum number of moves to retract until last promotion?" works well for processing replies in a contest. Indicates that the composer feels that answering this question shows his composition was thoroughly uderstood.

D. Crippled reduced stipulations. Example: "who made the last move?". Quite often this question can be answered without uderstanding how the position can be released (for instance in your diagram). It means the stipulation is artificial and does not really cover the essence of the content. Though these stipulations are somewhat inferior, they do exist, and they should be treated in the light of the first paragraph. As a solver you should still find out all of the history you can and not abort when you answered the question! But, imo, these stips are better converted to one of the preceding types.

E. Forward stipulations and other stuff. By other stuff I mean extra complications like "place a white king". Because these things do more than require the solver to simply "find out about the history" all kinds of new considerations may arise which make a simple subject complicated. For instance, "placing a white king" may not fullfill the requirement of an "efficiently reduced stipulation" because it need not guarantee that the solver has seen the full history. Still, the king placement has to be included and the generalized understanding of the RA-type tells the solver he shouldn't stop there. The alternative "place the white king and release the position" is correct but a bit over the top. Experienced solvers are curious enough to visit Rome when near it without marching orders.

Hi I agree with and very much like Ari's taxonomy of RA stipulation types, except I think that E smooshes together two unrelated types, and then focuses on the less interesting one, ignoring what is (in my opinion) the very best type of RA stipulation: the forward one.

If at all possible in an RA problem, I would want to include some kind of forward stipulation, even something as crass as #1, As long as it's sound I don't even care if it links in with the retro content. It at least means that there is some second way to read the diagram which is not retro in nature, and that is a plus. If there is no connection between the forward stip and the retro content, then while not ideal this is a matter of satisfying irony. The eager solver should still feel impelled to perform the full retro analysis. I would interpret #1 in such a situation as effectively "release the position", together with trivial ironic forward play.

This implies that "release the position" is already a slightly tarnishing stipulation because it means the composer was not able to find any forward play at all. He would have looked, for sure.

Of course, if the composer can find forward play and then link it with the retro play: e.g. there are two apparent forward solutions, but only one is legal for retro reasons, then that is better. And indeed best of all is where this forward/backward synergy is designed in from the start.

Forward stipulations can be very simple, and that's attractive too in the laconic world of composition where we want the diagrams to do the talking, not the stipulations. If I see an "efficiently reduced stipulation" with twelve words then I feel less enthused to solve it. What things was the composer unable to force uniquely that he tries to make me look at other things? If he is worried that I will miss the theme then that belongs in the solution, where the composer is allowed to waffle a bit. I don't think the problem and its stipulation should bear the full burden of supplying context to the problem. Non-canonical stipulations are also language dependent which can be an issue in the very international world of chess problem composition.

Yes, "E" smooshes together all unrelated types and I only added it as a garbage collection for complications, not to discard or condemn it. anselan proves my judgement right  by arguing in favor of artificial forward stipulations, something composers do not universally agree on. The A-D categories are pure as they relate to the past which is the real target of RA. The archeologist who tries to find out who made an egyptian death mask in order to sell it at a higher price is "tarnished", not the one who only tries to find it out in the name of science. A long time ago I used to obsessively stipulate my retro's with "mate?", "who mates in 1?" or even "who wins?" and was critcized for it by experts. Today I don't believe the one is better than the other. It all depends on context, available alternatives, personal taste and evolving intuition. Precisely the sort of fuzziness I tried to avoid in my previous post!

There is a terminology thing as well. I only consider forward stips to belong to the RA-type in its "artificial" forms where there is no interesting solution or history connection. The one thing such a stip conveys, is that you must prove a legal history besides giving a forward "solution". As soon as the forward play becomes real - actual challenges with history justification and conventions - then the problem type changes into RAC (my term) - the hybrid retro-active category - in which the retro-analysis serves to enable the forward solution. RAC is the target of my book, RA is not.

By providing such an illuminating & necessary discussion on the question of artificial forward stipulations, *you*, Ari, have proved *I* was right in suggesting that the category be split in two.  :-) I have made relatively few conventional retros, and I bow to your greater experience in these matters. Clearly there is an aesthetic dimension herein an RA: I would still tend to include #1 as a cherry on the cake. But as the focus shifts to "RAC" where forward stipulations are the general rule, these tired old unthematic stipulation elements do become less attractive.

Nit: I am a bit baffled by your notation. You distinguish between Retro-Active Category (RAC) and Retro-Analysis (RA). But why does putting something in a category make it different? Surely the notation should be HRA for Hybrid Retro-Active. It's unfortunate that Analysis & Active begin with the same letter. Maybe ROC for Retro-Orthodox Category?

While consumed in linguistic pedantry, "retroactive" doesn't mean the past affects the future. And "active" isn't the opposite of "retro". In astronomical terms, the opposite of "retrograde" is "prograde", and I have to say this resonates with me since prograde motion is the normal kind, just like forwards movement is the default in chess. However PRA is one of the few TLAs that is already taken in the chess problem world, which is heroically unlucky!

I agree, category E can be split into many parts! But my purpose was not to dintinguish those parts but to separate the categories with stipulations purely concerned with "finding out about the past of a diagram" from categories with different and possbly controversial elements. I did this in response to several previous posts by the Cobra where I thought he was confused about some of this. I never intended a full breakdown of the RA-subject but attempted to dispell some of the mist surrounding it.

My term "retro-active' will be explained in my book by its resemblance to "radio-active". Both refer to effects in present and future resulting from events in the past. Both tend to diminish in time (retro-activity when states become known) and both throw uncertainty in the functioning of normally predictable systems. A far as I am aware there isn't any radio active in radio-activity. If all language breaks down in the way you describe, I am afraid quite a lot of it is in need of repair. Anyway, the analogy to radio-activity is strong enough to justify the term retro-activity - though I can't guarantee it will stick! Note that the term "retro-active" seems meaningless inside the retro-universe but is extermely significant in a place involved in solving forward chess problems - which RAC is all about.

You asked why putting something in a different category makes it different. I fid that a strange question since the answer is "in about a thousand different ways". For one, indeed I will not write a single line about the RA-type in my book. The circumstance that borderlines tend to be vague does not imply that the same forces are in control on either side of it. The gap between disovering certainties (RA) and managing uncertainties (RAC) couldn't be greater.

The point I was evidently failing to make is that if I want to distinguish between two concepts Rock and Roll say, I would not invent a term Rock Category and call one RC while the other is R. The word "Category" is not pointed enough to indicate the distinction. The alternative naming Roll Category and Rock, might equally well have served, so the acronyms are no clues to the contents.

The ambiguous A in RAC remains a banana-skin for readers to slip on. If you must keep C for category then O for orthodox is better, giving you ROC which will never be confused with RA. But up to you.

Do you mean I shouldn't name it "retro-activity" because it leaves me with a deplorably ambiguous A? Isn't that the world on its head? I definitely think that "retro-active" is a witty name that reflects the underlying structure very well (by analogy). Conway would be proud of me! There is something else Conway is particularly famous for, but I can't seem to remember ...

Anyway, my terminology and acronyms are not definite and debatable. I am liable to make mistakes in those areas by my lack of familiarity with the english language. I believe however that all acronyms need to be accustomed to. Not many are self-explanatory, and when they are, then their source description often looks 'forced'. How many problemists would be capable of guessing what "AP" means unless they know the type?

Btw, I do'n't really think it is my call to name the field of retro-activity. I never stumbled on a term describing everything covered by the retro logics and the conventions and therefore made something up myself. Did you ever encounter an umbrella term for it?

Late edit: By whatever name is comes, RAC is a designation of problem types with a certain character and not about "special rules". RAC does not appear in stipulations but is presented through the retro-logics. It is however a useful concept in discussions, articles or contests. A composition tournament might for instance state that only RAC problems are accepted. All RAC-problems share three components: (a) a standard problemformat like directmate, selfmate and so on, plus substantial solution content (b) a retro-logic type like RS, AP or PRA (c) the conventions for retro-problems in general. Whether or not the retro-logic must be explicitly defined is another discussion.