When is it acceptable to answer a question that wasn't asked?

Question

I, personally, am a fan of answering the question asked verbatim, whether or not it seems to be the best approach to the implied problem.

For example, a question asked for a regular expression to validate emails, and tagged that question with php. The second answer clearly explained that one should not use regex in PHP, and should always default to filter_var, and had a score of +8/-0.

My opinion was that the user didn't ask, "should I use regular expressions", he or she asked, "help me build one." I answered the question by providing them with some solid regular expressions meeting the requirements, and received a score of +2/-2 with a comment "this is sad..."

Edit: Let's not focus too much on this specific example. We have all seen the questions that ask for a solution in Javascript and mention jQuery, or want a way to search the file system in Java and someone says, "this makes more sense for a scripting language."

Now I understand that there comes times for answers like this. If my son (no, I am not yet a father) asked me, "Hey Dad, which body part should I press against the stove to see if it is hot?" I would answer, "You shouldn't, you should hold your hand over it." And I would be a good father to do so, rather than "the back of your hand; it's the most sensitive to heat". We want to see people avoid being burnt by going down the wrong route to solve a problem.

But we're not all children here. Perhaps the OP had a perfectly good reason to need a regular expression instead of filter_var. Maybe they came upon some strange edge case that they didn't bother to explain because it wasn't relevant to the question. Should we not trust them to handle the stove as they see fit, and answer the question asked? In my opinion, the suggestion to filter_var is a comment, not an answer.

But alas, I am the one with a net score of 0, so I ask -- when is it appropriate to answer a question that wasn't asked?

Answer 1

There are cases where the answer to a question is "You are doing it in the wrong way," followed by an explanation on what is the correct way.
For example, if somebody asked how to change the content of a module that comes with Drupal, I would say that is not what should be done, and I would explain how to override a function provided by a Drupal module. I would not answer to the question asked, in that way, but I would tell the user what s/he needs to do.

In your example, a better answer would have shown a regular expression to match a valid email address, explained if there are valid email addresses that are not validated from that regular expression, and said that the better way is to use filter_var().

It is true the user could have a good reason to use regular expressions instead of filter_var(), for example because the code needs to run on a site still using a PHP version lower than 5.2.0; if that is the case, s/he should have stated that in her/his question.

The fact the answer saying to use filter_var() has more up-votes is probably because users prefer not to use regular expressions when it is not necessary. Imagine if a user wanted to know a regular expression to remove spaces at the beginning, or end of a string; the natural answer would be to use trim(), and that answer would probably be up-voted more than an answer suggesting to use a regular expression as requested from the OP.

Answer 2

I think that most of the time, when question askers are aware that they are not using the best tool for the job, they explain why not. And so it's tempting to generalize and assume that when they do not explain why, they aren't aware -- or just haven't put enough effort into the question.

However, given recent introspection across the stack exchange community about creating more welcoming environments, I think that it's important to assume a certain level of background knowledge until the asker directly requests more information.

Striking a balance

My personal tendency is to answer the question as asked, because that's just how my mind works. I've been stung a couple of times as a result -- and that's probably a good thing. It is important that a site like this point out the potential pitfalls of a particular approach, and I've learned to read questions with a slightly more critical eye over time. I still generally answer the question as asked, but if I foresee a problem, I may add a caveat, or suggestion that the asker consider a different approach.

But when doing so, I do my best to remain aware of my own ignorance. It's often impossible to convey all of the relevant contextual details behind a particular question. It's unreasonable to expect askers to cover every possible source of confusion. And there are very few "bad" approaches that are truly bad in every imaginable context. Taking a moment to acknowledge that you don't know everything about the asker's situation shows respect, and actually makes it more likely that they will consider your advice, even if they conclude that it doesn't apply to them.

Thinking about the long term

When I first answered this question -- quite a long time ago -- I ended with this statement:

More generally, I'd say that on balance, I'd rather have a site that is a bit pedantic at times than have a site that encourages the proliferation of bugs!

My position has changed because I see more clearly how this kind of pedantry works to exclude and belittle some audiences. And the notion that pedantry might somehow save us from a world full of software bugs seems quaint and romantic to me now.

These days, I would rather someone use the wrong tool today and come back tomorrow than use the right tool and never return. In the long run, we'll get a lot more good work done.

Answer 3

The best solutions to this I have seen is to answer a users question first and foremost and then warn them about the dangers of what they are doing.

Take a look at this answer I gave a while back about a MySQL driver call on PHP. First I attempted to answer the posters question as directly as possible.

After I did that, I mentioned the dangers of using MySQL driver queries and recommended MySQLi driver, or PDO_MySQL driver. For all I know this user was under a very strict environment, and could only use the MySQL driver; if they aren't, then (hopefully) they learned why the MySQL driver is bad, and moved onto MySQLi, or PDO_MySQL.

Answer 4

One important thing to note is that users asking questions here aren't actually looking for an answer to a question so much as a solution to a problem. The question the ask, the way they word it, and the details they include/exclude all amount to their representation of the problem.

A good answer does more than just answer the question explicitly asked. To write a good answer, you must understand the problem the asker is actually facing, and be able to solve that problem. A lot of times, the question lines up well with the problem, and a simple answer to the question is a valid solution.

However, there are many cases where the question doesn't communicate the problem clearly, or the problem the question addresses is actually a piece of a larger problem. The first order of business in these cases is usually handled in the comments on the question - clarification of what the asker is trying to do, how they are trying to do it, and why (as well as why they are taking that approach).

The next step is to provide a solution to the problem the asker is actually facing. That problem may very well be the user's lack of understanding. If I suddenly started working on a Ruby project tomorrow, I'd have little to no clue what I was doing. The first problems I encounter are more likely to be caused by me not understanding the language or the frameworks, and not knowing the common tricks and pitfalls. An explanation of a rule or pattern is more helfpul than a simple "change this foo(x.y) to bar(y.z) and it will work".

The Stack Overflow community understands this. That's why your answer, while addressing the explicit question, is scored lowly, while the answer that says "don't do that" has been upvoted. People vote based on the quality of the content, not how accurately an answer aligns to a question.^*

_{^*This statement is a generalization. I don't really know how each and every user on the site votes.}

Answer 5

We cannot control how people vote on question answers. That is completely up to them. Often times the best answer to a question is a question that better informs modes of thought. Not all questions require a response that concretely answers the question asked. In your specific example, I don't believe that you should be penalized for offering this answer, but neither do I think the person who was less concrete should be penalized. But that just brings us back to my original statement that despite your answer being valid, we can't control the way users vote.

Answer 6

That question provided a problem and the technology available. The other answer provided the correct solution. Your answer provided something that will break and lead to problems without pointing out the issue it raises. I think it deserves the down votes as not helping people learn the solutions to their problems.

In general answers should address the problem as asked, but the other answer fits into the scope provided by the question. The didn't say "use jQuery", they used the technology specified in the way it was designed, alerting them to an issue they need to be aware of.

Answer 7

The user might not always know what they are asking for. Sometimes people just add "buzzwords" to their question, but don't always know what they are talking about. Or, sometimes people's "requirements" are just wrong.

If you answer solves the issue, feel free to post it. Alternative solutions are always welcome, though you should describe why that solution is better than what the OP was trying to do originally.

Answer 8

Context matters for this, I don't think it's a generally-answerable question.

If I believe my answer is significantly different than what they're explicitly asking, I might ask via comment if they're entertaining alternative solutions, or I might ask why they're trying to do a certain thing, as there may be a significantly more appropriate way of going about it.

If I believe the OP might not be aware of something, I might answer with an alternative solution, but clearly label it as an alternative solution if such-and-so condition doesn't actually apply. For example, a user asked about a Struct, I replied with an alternative using OpenStruct.

Otherwise I'll generally stick to the asker's script, but may interject alternatives in both technology and approach "below the fold". With only occasional "If you were doing this in jQuery you'd be done"s.

Answer 9

IMHO First answer the question asked, then address issues, not the other way around.

If I ask how to Foo the Baz, don't tell me it is easier to Bar it until you've addressed the Foo solution (perhaps by saying you're not going to attempt the Foo approach).

Of course, there's also non-answer related topics, which need to be careful not to receive upvotes before true answers do, otherwise the actually unanswered question receives fewer views. Clearly they should be comments, but if important or detailed enough to be answers, preferably added after true answers have been upvoted or accepted.