I get why people ask questions like that. Many questions on SO are often weird because of the XY problem [0]. Someone asks for X, people answering think doing X is really weird and difficult, and thus dig a bit more. They then find that the asker really wants to achieve Y which is simple to do.

[0]: https://meta.stackexchange.com/questions/66377/what-is-the-x...

That may be true but it often ruins StackOverflow as a repository of answers. There’s nothing worse than actually needing to do X only to find a StackOverflow page where 10 different answers give 10 different Ys.

Or when X needs to be done but instead all the comments hector the question with useless criticism.

If a subject matter expert thinks that people asking for X are falling victim to the XY problem, then I want to hear about it before I go down the mistaken road of implementing X.

If there truly is a circumstance which makes X the right solution, then that might warrant a new question being posted with the additional context to make that so. But I suspect it's much more often the case that the asker is simply being stubborn and doesn't want to change their plans.

You seem to be under the impression that people answering questions on SO are subject matter experts.

You’re correct that they often aren’t, but all the same I find it amusing to read this comment attached to a StackOverflow link where the exact person who suggested “gimme gimme gimme” and the implementer both answered.

Obviously not the ones making the “why are you $X” comments in this case though.

Remember the name of the site that SO replaced, "Experts Exchange".

Obviously it's possible for non-experts to post authoritative-sounding answers on questions that they don't really understand, but I think that is a different category of problem from the "XY-problem" style answers.

I always read their domain name as expert sex change.

But the new question almost always gets closed as a dupe in spite of clarification.

Yes, I could agree with you there. I'm not going to defend the stickler moderators on there (just look how they shut down the interesting comments on the top answer).

There are places to ask the question "I think I need X. Is there a better solution?"

But the answer to the question "How to do X?" really shouldn't include more than a passing reference to "Here's how to do Y." If X is impossible, then that is of course the answer.

And as others have indicated, the most frustrating scenario that happens over and over is that once the "How to do X" question is asked, you can never get an answer to that question if the original poster relents and agrees that what they really need is an answer to "how to do Y?".

Imagine the question is something like "How do I concatenate these strings into an SQL query?"

Would you be willing to be responsible for the fallout of answering that question any other way besides "use parameterized queries instead"? Do you really think that should only be mentioned as a passing reference?

Absolutely, yes. Someone has to write the parameterization libraries. That is, whatever abstraction layer you want to employ, in the end SQL queries are strings.

And the passing reference to the potential misuse here can be a big bold statement at the top of your answer.

Except that you sometimes have to anyway (not for parameters, but to have dynamic conditions, filters etc.)

And since every question about such things get auto corrected to use parameters (which doesn't help, since it's something else entirely), You have millions of buggy self glued sql queries allover code bases. (but hey most of them are using parameters, so yay - progress)

In those cases the best solution would be to use a query builder library like JOOQ, I still wouldn't be willing to recommend someone in that position builds the query by concatenating strings together

Yeah, unfortunately, these days SO is relatively good for shallow questions but mostly terrible for niche questions when you're already at an advanced level on the topic. Often enough you get several inquiries about your use case from people at or maybe even below your level who can't really answer the niche question, you waste time explaining irrelevant details and fending off skepticism, and end up leaving without an answer.

I still occasionally post these _when they aren't urgent_ because there's an off chance someone will know, and a greater chance some future traveller will find it by search and turn up with a good thought or answer.

But yes, it can be a lonely game :)

It's also encouraged to answer your own questions. So if you struggle with something, and find a solution, you can post the Q and answer at the same time even. To make it easier for others in the future.

You say it ruins it, but my in my opinion, that's exactly what makes SO valuable. Often, it's not just the asker of the question that's trying to solve X but asks how to solve Y. So when you search on how to solve Y, you might also find out a better way to solve X that doesn't involve solving Y.

And if you are really trying to solve Y, it's not hard to specify that in your question (I know I can solve X without solving Y, but I don't want to solve X, I want is to solve Y in this case). I wouldn't say "there's nothing worse" than that; it's a minor inconvenience.

I agree. If I'm sure my problem is not an XY problem, but I think it may well not look like that because it's such a weird situation, I will add all the context that explains how I got there.

Then either: I do know what I'm doing, and my explanation satisfies people, and they try to work on Y, or: it turns out I was indeed wrong, and I should have solved X some other way.

Either way it's a win for me (if anyone bothers to answer the question).

But it can be hard for either new people, or people who don't know the domain well enough to even know that it could be an XY problem.

Why don't they just filter out the useless Y's, but keep the good ones?

The XY problem applies when someone asks how to accomplish something. In this case they want to know why something happens; what they’re doing around it is irrelevant.

It’s analogous to asking “why does pizza taste metallic to me?” and getting “why are you eating pizza?” as a reply.

Isn't the better analogy that, "If I take aspirin with water for a hangover it works but if I take it with whiskey it makes the hangover worse. Why doesn't aspirin work with whiskey?"

I would definitely ask them why they think they need to take aspirin with whiskey for their hangover.

No, it's more like, "Why do people literally float a few inches off the ground when they lick an aspirin then take a sip of whiskey?"

If that input indeed did cause that result, you'd expect a forum of human beings to have responses like this:

"Holy shit that result was most surprising! I have to re-evaluate my entire life now. Just curious, though-- what made you decide to lick aspirin and sip whiskey?"

Any response that didn't explicitly confirm the result would be signalling to the world that they are either a troll or a boring contrarian.

Similarly, `man` isn't ever supposed to output `gimme gimme gimme` as reported. That is surprising behavior. And that surprising behavior is trivially reproducible by the likely harmless act of setting your machine to the time specified and running it.

Thus, the fact that this reported case was actually a feature of `man` outs all the people who posted non-sequiturs as low-effort cranks. (Low-effort because they didn't even take the time to run a trivial test.)

Edit: just to be completist about it-- I suppose a respondent in either case could simply challenge the veracity of the report. But that's not what we're talking about here AFAICT.

Edit: to make a steelman argument here-- maybe not cranks, but people who have, for whatever reason, fallen prey to the act of low-effort crankiness.

And yet I think the probing XY questions are still important.

The answer could have been simply: "It's an easter egg. It's from an Abba song." Done. End of discussion, it answered the question.

But by probing how the user wanted the thing to work, and why it was breaking his tests, and that it was actually happening when he used `man -w`, the maintainer actually understood that it was not a harmless easter egg, and fixed it so it didn't happen when called with -w.

Despite the wording of the question I think it's pretty clear the real intention is to get their tests working. And even then, the top answers do in fact directly answer the question as asked.

No, the pizza analogy doesn't work. It's the specific use of man here being questioned ("why is it used in a test?"), not man usage in general.

Oh it actually does. Making it specific doesn't change anything:

When I am asking: "Whenever I eat pizza in the morning, it tastes metallic to me. Has anyone an idea why that could be?"

I dont want someone to answer: "Why are you eating pizza for breakfast?"

This answer is worth nothing. Actually its even worse, cause answering it costs time.

Worth nothing to the asker. But stack exchange is a two-way communications platform, and responders (and other readers) may want to know more context.

I, for one, am still curious why OP is running a test where they call `man` with no input and fail on different stderr.

> Making it specific doesn't change anything

Of course it doesn't, when your "specific" analogy isn't nearly analogous to the situation.

I actually agree with him, it's a perfect analogy.

Calling man in a test and expecting the output to match some particular value is a much weirder thing to do with man than eating pizza in the morning.

It is honestly a lot more likely that the reason man is outputting ‘gimme gimme gimme’ at a particular time of day is something to do with your crazy test setup than that it is a feature of man.

So for the pizza analogy, think of it more as ‘pizza sometimes tastes metallic to me when I cook it under the hood of my car’. The car situation seems likely to be relevant. It would be surprising, when someone asks a question like that, to discover that it’s actually because the frozen pizzas they buy when they drive to the parking lot where they then cook them under their car hood actually contain iron filings that were put there deliberately by the manufacturer.

Why does pizza taste metallic to me when I eat it on a boat?

The analogy is completely fine. "Why does the pineapple on my pizza taste metallic"? "You shouldn't put pineapple on your pizza" if the original bothers you.

Analogies don’t have to be one-to-one. It’s close—and simple—enough to get the larger point across.

It doesn't, though. As I see why asking "why are you eating pizza is wrong", but not why asking "why are you using man in your tests" is wrong. So your analogy is more of a misrepresentation of the situation than anything else.

If you don't want focus on the details, then don't provide them. If you provide them, it has to be because you think they might be relevant. And then it's relevant for people to ask for further details.

> As I see why asking "why are you eating pizza is wrong", but not why asking "why are you using man in your tests" is wrong. So your analogy is more of a misrepresentation of the situation than anything else.

So far you seem to be the only one failing to grasp the original point. Your misunderstanding of the analogy does not transform it into a misrepresentation.

It's analogous because the existence of tests have absolutely nothing to do with the question at all.

This due to the use of "why" and "because" in English to introduce explanations that are fundamentally different and grammatically differentiated in other languages. One is a cause, as in the antecedent cause of a consequent effect, while the other is a purpose, the effect desired prior to undertaking an action. This is the difference between etiology and teleology, which many students fail to grasp. "Why does man print this?" asks for a cause, while "Why are you using man in a test?" asks for a purpose.

The XY problem in theory: An expert with broader knowledge than you realizes what you really should be asking.

The XY problem in practice: Someone doesn't know the answer to your question so they randomly choose a response from the ten things they do know.

Also: If we're calling it the "XY problem," does that make it mansplaining? (XY? Get it? Sorry.)

If someone only has a tangential knowledge of the problem, would it be better if they just don't give their input at all and simply leave the thread empty? Non-experts contributing their opinion doesn't take away space from the experts to answer it properly.

If the advice is really that bad or uninformed, then it should just be downvoted. That's not really related to the "XY problem".

On stack overflow, having several tangential non-answers will make the question look well-answered, potentially misdirecting the real expert away from it by creating a false impression.

> Non-experts contributing their opinion doesn't take away space from the experts to answer it properly.

Ever heard of a thing called signal to noise ratio?

I personally hate when that happens because usually I really want to do X for good reasons. In the end the people asking got help on troubleshooting their specific problem, but the question is still unanswered (or answered by “You don’t want to do that because you’re a beginner”).

Reminds me of the time I asked "How can I do this Python feature in one-line because I'm writing a to-one-line Python transformer".

Still got idiots replying that I shouldn't be doing what I want to do in one line.

If you are so unfamiliar with X that you require the advice of experts to implement it, then how can you say so confidently that the reasons to do X instead of Y are justified, even though those experts disagree?

If it's really true that the conventional wisdom about Y is just wrong, and X really is better in some cases, then it will be a long process to get all the experts to reconsider their assumptions, requiring battle-tested proof, and that probably won't happen in a single SO thread.

I'm not unfamiliar with the problem space and why I want to do X, I'm just looking on implementation details. If the question was "How should I solve that problem? Should I use X?", then sure "use Y instead because it's a better practice" would be a valid answer.

I'm the opposite. I often catch myself asking about Y and then needing to back up and ask about X. I even know the problem by name and can't help (occasionally) asking the wrong question.

I don't disagree, but then the question should be phrased that way, or it should be mentioned than there may be better way for generic use cases but the original question still answered.

The people who answer Y without addressing X are far more annoying and detrimental than the people asking X.

If you see a likely XY problem, answer X, then explain why they might want Y, and then answer Y.

I've been learning to recognize when I've been presented with a solution, rather than a need. Often times the solution answers the need, but lacks knowledge of either side effects, implementation time needed, holes in leaky abstractions, or something else that makes it less than ideal. Walking back to the need helps in proposing a better solution.

I find this frustrating because I often want exactly

Yeah, I get that as well and sometimes feel the same. The difference is that a beginner tries to do Y, fails, think they can solve it by doing subtask X, so they ask X. But us with more experience know why Y didn't work and probably have a good reason for asking X.

But the answerers cannot know that. So adding a small snippet of what have been tried, or why X specific is what's needed is a good idea to steer the conversation in the correct direction.

In my experience, if you tell people your use case for wanting to do X instead of Y, they'll just attack your use case instead.

I, for one, never use places like SO anymore because everyone just wants to pretend you really want Y, because that's an answer they know.

I use SO all the time for gleaning information from the answers, but never for asking questions, for this very reason. The questions that show up in search results are full of great answers and information, but mine ended up full of people who didn't know the answer to my question but were eager to tell me about something they did know. Then I'd put more and more effort into providing context and justifying the question, until eventually the answers dried up. I decided I was better off seeking help on mailing lists, issue trackers, chat rooms, etc.

Yeah honestly I feel like it has gotten a lot worse since I was using it actively maybe 8-10 years ago. It seems to have moved towards a lot of people playing the karma game, and trying to focus on every flaw rather than trying to answer the question and give help. Nowadays, I usually try to find a community or forum directly related to the specific subject I'm working on.

However, it's hard to tell if the quality has actually changed, or whether it's just more geared towards novice programmers and I'm just not asking the type of questions that get help on the platform anymore.

Then specify the context as part of your question: "I want X. I think alternative solutions like Y won't work because...". As it is with many things in life the more effort you invest in the question, the greater the returns :)

Although tbh I quite often want the opposite, so I add "Alternative solutions/workarounds welcome" or "I don't see how I can avoid doing X but maybe..."

> "I want X. I think alternative solutions like Y won't work because..."

Listing all of the things you don't want to do would not only take up more words and effort than the question about what you do want to do, but is hopeless. Somebody who doesn't know the answer to your question but still wants to comment will find something that you were unaware you didn't want to do.

It's better just to answer questions directly, and then when somebody uses that solution and it doesn't get them closer to the solution to their real problem, they'll ask a better question.

Sorry, but that doesn't make sense and sounds like a rationalization for being lazy with the question. You're worrying about a straw man instead of trying to help out people who do know how to deal with the actual problem you have.

And bad solutions can still "solve" problems with horrible trade-offs, so instead of searching again for a better solution people may just employ the worse one unaware of the better possibilities, consequences be damned.

I think it depends on the quality of the question.

For instance, if it's a semi-vague question just pointing in the direction of a solution, then I think it's fair to question the motivation of the poster.

But if it's a very precise question about how to accomplish something specific, I think the burden should be on the person answering to add the clause: "you can do X, but its generally not considered a good idea because Y".

Honestly if you can't answer the original question, I get the impression you're just trying to be a smarty-pants and show off some domain knowledge you have which is related to the task at hand.

The point is that SO is actually less about the person asking a question. That is only one person.

But people look for already asked questions by searching for them. And I don't think that there are many things that are more frustrating, then finding a questions that describes your problem X perfectly and it has only answers to Y.

Since SO exists for the readers and not just the asker, I think that makes it even more important to highlight all the reasons not to do X if it really is unlikely to be the best solution.

Maybe the asker really would be best suited with X due to some contrived circumstance, but is that true of everyone else who finds the thread too? Probably not.

I just don't see why it's a problem to add a caveat in the answer in that case.

I.e. "you can do X to achieve your goal, but you probably shouldn't because Y".

I just don't see how it's helpful or contributive to participate in the discussion if you don't actually know the answer.

This is also why the canonical answer of turning it off and on again is a meme.

I think it’s similar to how experienced users access consumer support processes. When you call a support line, you might know exactly what you want, and just want to be routed to the process you know you need, but 95% of customers are better served by being put through the “have you tried turning it off and on again” process.

Similarly on Stackoverflow, a majority of users probably are doing it wrong and are helped by the XY-troubleshooting approach.

Experienced users who really do want X end up being collateral damage.

Before applying an XY troubleshooting approach, one should learn to recognise XY problems. While some cases may be ambiguous, I’d argue this one isn’t.

Sometimes experienced users are also served well by walking through the steps. At least at our company, you see field technicians who think they know what is happening and try to short-circuit the troubleshooting process. Sometimes they are correct, but oftentimes they miss a significant detail that they would have caught if they had just followed the normal procedure.

Well, in this case I would also like to know, out of pure curiosity. I cannot come up with a viable reason on my own why you would run 'man -w' as part of a test.

Verifying you have your environment and search paths setup correctly.

From the man page:

> -w, --where, --path, --location

Don't actually display the manual page, but do print the location of the source nroff file that would be formatted.

Why would you even ask such a question to begin with? You say curiosity, but there seems to be more to it?

Asking why to a part of code or workflow is a good thing, even only because of curiosity. After knowing it, you may find the reasoning brilliant, or ridiculous and dismiss it and that's it.

However if any chance you have the same situation as OP, it may come out valuable, either currently or in the future.

I would have asked the same question for the same reason. A sibling comment introduced me to a feature that I wasn't familiar with. And that's why I'll continue asking curious questions until the day I die.

Why would there be more to it? I wanted to ask the same question, just because I simply did not know the answer. In fact, I am curious how you became so wary about such a simple, uncomplicated and direct question.

It was meant as joke, question the questioneer, for an honorary -4.

It's a bit of a Stackoverflow trope, so much so that it shows up in these "parody questions" that surface from time to time.

In some cases the people answering have the experience and breadth of knowledge to understand the "real" question being asked, even if the OP doesn't quite have it yet. Other cases, it seems like arrogance and showboating, and doesn't contribute much where a simple direct answer would be best.

The one that gets me is seeing an answered question, with a bajillion upvotes, locked by mods as "off topic" or "not useful" or whatever. Clearly a relevant question with a useful answer, as evidenced by the many comments and kudos. Ugh.

People prefer problems they know the answer to. Sometimes, when they don't know the answer to a problem, they'll try to cram it into a shape they recognize so the answer they already have works.

I think that's much more the case than the whole "XY" issue. It's clear the people asking "clarifying" questions about the workflow didn't have the slightest idea that man did that, much less why. But they can't get points or a feeling of accomplishment from recognizing that they're out of their depth and staying out of the discussion.

See also the questions that amount to "Why aren't you using the operating system/language/framework/library I'm actually familiar with, so I can answer you and get points?"

That is one of the things that turned me off of Linux in the very early days ... every time I tried to find out how to do something, I was a big moron for not using someone else's distro.

Thanks for articulating this. I call them XY inquisitors. We had a couple at work a few years ago and any question directed at them was met with a somewhat derisive "What are you trying to DO?".

Of course there are cases where the question poser of Y benefits from having their X identified, an archetypal case here:

https://stackoverflow.com/questions/2691018/decimalformat-an...

And that's not even an XY example ...

You guys are so confused you don't even get yourselves.

Sure it was. But it was a good example, because the person helpfully starts with "I'm trying to do X."

But they still say. "I'm trying to do X. To do this, I'm doing this complicated Y. How come Y doesn't work?"

This is still XY because they're asking about Y. But in the bad cases, they'd only mention the Y part, and omit that they're trying to do X.

The XY problem is "I'm trying to do Y" -> "Are you sure you don't want X?"

This is "I'm trying to do X" -> "Here's how you do X"

No, not exactly. From [1], the XY problem is "you are trying to solve problem X, and you think solution Y would work, but instead of asking about X when you run into trouble, you ask about Y."

So person is trying to solve X, but asks

"How do you do Y?"

"Why are you trying to do Y?"

"Well, I want X and Y seems to be the solution"

"No, to solve X you want to do Z."

In this case the asker nicely gave their X as part of the question, but the bulk of the question was still about trying to move forward on their mistaken "solution" Y.

1. https://meta.stackexchange.com/questions/66377/what-is-the-x... (and note all the other answers saying the same thing.)

When you accuse someone of being wrong, it's helpful to provide justification, not insults.

Y: How do I strip unnecessary symbols after a decimal point?

X: How do I round a number to a desired precision?

It looks like the perfect example of XY problem to me. Please tell me how I'm confused.

Your Y is not a different problem, it's just a poorly worded description of X.

In an XY-problem situation, X and Y are different problems.

They are different problems. Stripping the final character from "3.14159" does not produce the same result as rounding to 3.1416.

The first action concerns a syntactic property of PI's decimal representation while the second action concerns a semantic property of PI's value.

A related example given on https://xyproblem.info is a user asking (Y) how to extract the final 3 characters of a filename string when he probably intends to ask (X) how to retrieve a filename's extension.

Agreed. IMO StackOverflow would be greatly improved if they implemented a rule barring these responses unless they first answer the actual question at hand.

To build off of another user's example, answering

>"If I eat pizza in the morning, why does it taste metallic?"

with

>"Because of a chemical reaction due to blah blah blah... which raises the question- why are you eating pizza for breakfast anyways? Avocado toast is superior for <reasons>"

Would be far more productive of a response for both cohorts of users; those searching for the direct answer and those searching for a better solution of what to have for breakfast.

> people pick away at stuff that's orthogonal to the issue and won't resolve anything.

Sometimes it’s a case of “I don’t know the answer to that, but I’d like to help. I’d like to know more about why you’re doing that so I can perhaps suggest something that doesn’t require doing that anymore.” Which indeed may be very helpful because even if you knew the real answer to your problem, in many cases you’ll need a workaround since the behavior probably won’t be fixed anytime soon.

I like to read the reasoning behind things that appear to be spurious because sometimes it turns out they have a good reason for doing it that I haven't considered, and it might be something I want to incorporate in my practices.

> this isn't one of those cases.

Reading all of the replies and context around the solutions, you're right that it doesn't appear to be one of those cases. But I don't think it was that clear cut from the original question.

The counter to this is, I believe, much much more common on SO: someone comes up with an odd solution to a problem in their head, tries to implement said solution, runs into challenges and comes to SO asking for help making their solution work. That then leads to over-engineered workarounds for hurdles they would not have encountered had they asked for help with their actual problem.

It can be useful. I remember asking about a problem I ran into in a convoluted solution and someone suggested I should try a much simpler solution which was actually a better idea.

It can be, if the people involved are knowledgeable. In this case and a distressing number of other cases, it's clear the people piping up have no idea what's going on (and some aren't even paying attention when reading the initial question). They're imitating what knowledgeable people do to feel like they're contributing.

I think it is something like the old joke where you tell the doctor that it hurts when you do something and the doctor says stop doing it. Often, we can eliminate the problem by factoring it out altogether. In this case people observed the problem to be solved as a failing test case. The author could have avoided being ask why by just stating the facts and not trying to explain why he observed it.

No, they'd ask twice as many questions in that case.