When I first stumbled upon the tag a year ago, it did not have a definition, and there were only a very few puzzles in this category. One of them was this one, which I liked a lot. It seemed to me that "retrograde analysis" - interpreted in general as "backward reasoning" of finding antecedents from consequents - was a nice stop on the spectrum between and . I happened to have an open puzzle at that time which immediately seemed to fit well into this category, and so I added this tag to it. I later added the tag to this question as well.

Naturally, I was happy to see this tag become the subject of the 29th Fortnightly Topic Challenge (and even more so because I wanted to post a real-life puzzle involving a computer program whose output was unexpected). Only to be disappointed to find that 2 puzzles above had the tag removed (mine seems to have been overlooked), and that a tag definition has been (only recently) added to restrict its meaning specifically to board game positions / the history of of gameplay.

The question:

Nothing in the words "retrograde analysis" implies that it should relate to board game positions. If its meaning can be expanded from its origin in chess to other board games, then why not expand it in general to the kind of puzzles mentioned above, that were originally tagged as such without much disapproval earlier? Alternatively, if we think that should indeed be restricted as per its current definition, then what tag should be applied in its place to the puzzles above?

Some options that I can think of are:-

1. : Generalize its meaning.
2. : This comes close, but questions in this category tend to, IMO, involve more guesswork / "creative" thinking, contrived situations, deliberate omissions or cunning in the description, and often invite "too broad" answers, whereas the examples above are often real, of an investigative nature, and have only one correct answer.
3. : This one's definition actually comes very close, but carries a connotation of "crime" or work; which means that someone searching for puzzles with this tag might not expect the kind of puzzles mentioned above.
4. : The current definition is restricted to the incident itself being puzzle solving, and hence can't be applied.
5. A new tag? (probably not...)

TL;DR: Should be used not just for board game positions, but for "backward reasoning" puzzles in general (see examples linked above)? If not, which tag should be used for such puzzles?

• I'm not sure what to do next... Do I accept the only answer posted, and should we then re-tag the puzzles linked to above? – KeyboardWielder Apr 3 '17 at 19:44
• @Rubio: Since you re-removed the tag from this question, could you please post an answer or comment here describing why such puzzles do not fall into the definition of the tag, and suggest a better tag to apply? – KeyboardWielder Apr 10 '17 at 17:33
• @Deusovi: Since you re-removed the tag from this question, could you please post an answer or comment here describing why such puzzles do not fall into the definition of the tag, and suggest a better tag to apply? – KeyboardWielder Apr 10 '17 at 17:34

I realise this response is a little late for the fortnightly challenge, but I think it's still worth sorting this out generally anyway.

In my opinion, we should generalise , to include non-chess/board game specific usages. Whilst chess appears to be the genesis of the term and the most common usage, I don't think there is any fundamental value in fencing this specific usages off from other areas of "backward reasoning".

The upside of generalising the tag is that it gives a home to the types of puzzles you mention, with only the potential, minor downside of diluting the chess specific examples. However, the beauty of a tagging system is that any given puzzle can have many tags, so you can use + to indicate (or search for) domain specific examples.

To take it a step further, if people are uncomfortable with bastardising the "strict" definition of the term (debatable in this instance anyway), we could also rename/alias the tag to something more general... Wikipedia seems to consider "backward induction" to be the more general term covering "the process of reasoning backwards in time, from the end of a problem or situation, to determine a sequence of optimal actions" and describes as the chess specific case.

I don't think any of the other tags you've listed are suitable candidates as they don't capture the specifics of "working backwards" ( is the closest, but it's explicitly different in the sense that retrograde analysis is determining steps taken from a given endpoint in a specific domain, whereas reverse puzzling is about determining an endpoint and a domain from specific steps taken).

• Thanks, Alconja. I like the last sentence that clearly distinguishes reverse-puzzling. I did see those Wikipedia articles too, but prefer the words "backward reasoning" as being more accessible / less technical than "backward induction". A tag isn't as helpful if it is harder to find or understand. – KeyboardWielder Apr 3 '17 at 19:42

In retrospect, the purpose of this question was not so much to define the definition of "retrograde analysis" as to figure out how to tag puzzles that involve "backward reasoning".

The decision (whether tag or some other tag) doesn't matter so long as there is a definition made and applied consistently by the community.

But in either case, I think there should be some relevant tag applied: Because the purpose of a tag is to be able to find puzzles that belong to a certain category. In that light, personally, I would rather prefer that puzzles were tagged with a 'somewhat' related tag, than not have any tag at all that captures its theme. Then at least there's half a chance that someone can find puzzles with that theme. And if a more suitable tag should become available, then the "wrong" tag still serves to help find those puzzles that need re-tagging.

In the untagging of from some of the puzzles mentioned in the question, it seems to me that there is more concern about the application of a wrong tag, without a corresponding effort to substitute a correct tag in its place, and I think that is not as helpful.

• Really. The only reason I imagine to have such a restrictive definition for so few puzzles would be to warn against those puzzles, like mathjax does as double-duty, when a vast portion of solvers couldn't even get started on them. – humn Apr 10 '17 at 21:57