Preempting inadequate answers and the ‘XY problem problem’

Question

A number of times I have asked a certain kind of ‘attractive nuisance’ question – that is, one in which I anticipate a certain kind of answer would be given that I already know I am going to find insufficient or entirely inadequate. Some examples:

• fsck exit status "System should be rebooted"
• Assigning to a variable in a parent context in Bash
• Network interface of type ‘phonet’ – what is it good for?

A more abstract example would be asking ‘is it possible to do this 100% correctly’, where people itch to come up with hacks, workarounds and half-baked solutions of various degrees of robustness – while the point of the question is obtaining a fully correct and reliable solution.⁰

I try to be very explicit in the question body about what kinds of answers I am not interested in. And yet, every time I ask such a question, someone gives such an answer anyway, or an outright XY-type, i.e. frame-challenge answer – where in fact the framing of the question is exactly how I want it, and I don’t want it challenged.

Since I have already said in the question body that such an answer would be inaqeduate, when I get one, I downvote it. I believe it’s only appropriate, but it tends not to be received very well – I get insulting comments and downvotes in retaliation.

I realise that a proper answer would be much harder to come up with than an ‘obvious’ inadequate answer. I suspect that often a real answer should be negative: ‘there is no way to do that’ or ‘there is no way to answer that for sure’. Nevertheless, if that is the right answer, this is the one I want posted – I don’t want the Internet polluted with more tumbleweed of what is already well-known (and may even be subtly wrong), and I don’t want to give ‘A for effort’ upvotes.

What is the most appropriate way to act in such a situation?

---

⁰ I am also reminded of when Gary Bernhardt asked Twitter whether it’s possible to control the lexical environment in which evaled code executes while trying to avoid reflexive ‘don’t use eval’ replies. Of course, expecting nuance from Twitter of all places is silly, but that’s beside the point.

Answer 1

Unfortunately, not much.

The fact of the matter is that the vast majority of questions that look like X-Y problems, are. And for many questions where people claim to be asking for X explicitly because they don't want/need Y, it turns out they're convinced Y doesn't actually solve the problem because they don't understand it enough, or are trying to Rube Goldberg their way to a poor approximation of Y, or some other convoluted reason, with the end result that the most correct answer ends up boiling down to "just use Y and stop wasting our time". The end result is that everyone who spends a decent amount of time looking at questions will tend to classify anything that looks like an X-Y problem as such at first glance - even if it actually isn't.

The other problem is people who just want that tasty reputation from an accepted answer, and so post something that is vaguely related to what was asked, because they didn't bother to read the entirety of what was asked. AKA the Fastest Gun in the West (FGITW) problem.

The only thing you can do is be as explicit as possible in your questions. My suggestion would be some sort of prefix along the lines of the following:

Please note that I am NOT looking for answers containing the following:

• foo, because <reason>
• bar, because <reason>
• quux, because <reason>

---

[Actual question body]

That immediately and neatly solves the problem of people not reading, because you're leading with what you don't want. It should also prevent people from interpreting your question as an X-Y problem, because you've stated the reasons up front why the Y they might assume you need, is not the X that you want. And if your question is an X-Y problem (and you don't know it), the fact that you haven't enumerated the Y in the list of things you've discarded (indicating you aren't aware of it), leaves the door open for that Y to be posted as a valid answer.