11

The "don't ask" page of the Help Center contains one sentence that I've always felt captures the essential ingredients for a truly great question on the network:

You should only ask practical, answerable questions based on actual problems that you face.

I find myself quoting this line a lot on beta sites and when leaving comments in the First Posts queue.

But today it struck me that, although I feel like I understand what it means for a question to be "answerable," I don't know that I could sit down and explain what it means in simple English to someone who doesn't intuitively understand the concept.

What specific, explicit qualities distinguish questions that are answerable from questions that are not answerable?


...now comes the (possibly tedious) part where I go over my efforts to find an existing explanation and, in doing so, risk distracting from the focus of the question.

I'm going to link some questions and answers in this post, purely for illustration. Keep in mind that the ultimate goal in answering this post isn't to pass judgment on the examples or to rehash past discussions,1 but to pin down the concept of an "answerable" question in the context of SE, and provide some updated, explicit clarifying language.

The best I could dig up concerning "answerable" questions was the Real Questions Have Answers (RQHA) blog post. Jeff's point there boils down to:

real questions have answers, not items or ideas or opinions

Well... okay, while that's helpful for someone who can mind-meld with Jeff and intuit the distinction that he's expressing, it's not as clear or explicit a statement as the tone suggests. It's "X has Y, not Q or R or S" where sometimes Y contains Q or R or S, none of which is very narrowly-defined.

It's also more than 3 years old and seems to run counter to the following evidence of what questions and answers have been exceptionally well-received on the network sites.

Answers, not items

Legitimate questions can easily inspire fantastic answers (or sets of answers) constituting, or organized in terms of, items (and even outright lists, as Shog implies in this answer). This is particularly true for conceptual questions, practical questions that are particularly complex, and problems for which there's no known, singular solution. Here are a few examples:

Answers, not ideas

I'm going to gloss over this one because the word "ideas" is semantically slippery, but I did poke around a bit, and this top Physics question was too fascinating not to mention.

Answers, not opinions

This one's easy; there are many examples of openly opinion-based questions that have been well-received in terms of having generated lots of views, favorites and illuminating answers. I know that's been a bit controversial, and Robert discussed this topic at length in an earlier blog post, Good Subjective, Bad Subjective (GSBS).

That being said, here are some examples I collected while trying to avoid sites known to be inherently subjective (e.g., Code Review):

Maybe these are the sorts of questions Robert had in mind in GSBS; maybe they're exceptions to the rule;2 maybe you think they're literally worse than Hitler. That's not important. As he points out in the blog post:

Even the definition of what is too subjective on Stack Exchange is somewhat … subjective.

What I love about Robert's post is that he immediately follows this admission with an in-depth explanation of six explicit guidelines (the same guidelines listed on the don't ask page) for distinguishing "good subjective" from "bad subjective" questions. That's what I think would be useful here -- a set of explicit guidelines that we can apply to determine whether a question is or is not answerable.3

Stack Trek III: The Search for Truth?

I recently came across an interpretation of Jeff's statement (quoted above) from RQHA:

Although this is compressing a twelve-hundred-and-some-odd-word long post into a single sentence, I think the message is rather clear: The difference between a question and a discussion or poll is that a real question is seeking to discover some particular truth. The other type of "question" is merely seeking to learn what other people think.

...

"Real questions" don't necessarily have practical answers, but they do have authoritative ones.

Is the expectation of one or more authoritative answers one useful metric for judging whether a post is answerable? (This is not a request for an ontological treatise.)


1 If this is a dupe, let my final words be recorded as, "I hate Mondays."
2 Zipping back to the previous section for a moment, there was some meta discussion about the "mass of a coin" question being one such lucky exception to a rule.
3 In the context of Stack Exchange, of course -- because if the internet has proved anything, it's that given enough time, Google will index "answers" to just about any question you can think of.

  • 5
    I wonder if this question is really... Nope. Not gonna type it! ;) – Andrew Barber Aug 18 '14 at 20:16
  • 1
    Comes down to the individual sites. The Jeff post is about StackOverflow, where not a question has some pretty clear guidelines (at least if you meld--not lists of things, discussions, or sharks fighting bears). Other SE sites have decided to be a bit more lax... – user1228 Aug 18 '14 at 20:37
  • @Won't I don't know how close the relationship was between SO and the overall network back in early 2011, but the blog post is filed under [stackexchange] so I interpreted it accordingly. – Air Aug 18 '14 at 20:45
  • @AndrewBarber I hate Mondays – Air Aug 18 '14 at 20:45
  • Just because you put a question mark in a sentence doesn't mean it is a question? – Travis J Aug 18 '14 at 23:12
  • 1
    Give the question, can solution be produced in a bounded amount of time. – Aron Aug 19 '14 at 12:07
  • 2
    @Aron: That interpretation has been previously suggested and rejected. – Air Aug 21 '14 at 15:58
3

In order to determine whether a question is answerable, Pretend that you know the answer. This shouldn't be unreasonably hard if you understand the question, because you should have some idea what shape the answer will take.

You don't have to make up a real answer, you need only imagine an abstract answer that fits the question. This might be easier for those of us who have a programming background(which many of us do.)

If you can imagine a potential answer

  • That is unambiguously correct (I.E. the OP gave you all the information you need, and the answer is not an opinion.)
  • That doesn't leave loose ends. (This covers the problematic list questions where there can always be more items to put on the list)
  • That can be written without an unreasonable amount of effort

Than the question is probably answerable.

  • This is a great heuristic. I'm not sure I would be able to apply it to a "Good Subjective" type question, though. This comment about the "rankability" of answers is a great insight that might be helpful in those cases. What do you think? – Air Aug 22 '14 at 22:58
0

"The difference between a question and a discussion or poll is that a real question is seeking to discover some particular truth. The other type of "question" is merely seeking to learn what other people think."

We deal with this a lot on the more subjective Stack Exchange sites. The difference between an answerable question and a discussion post is indeed based on truth vs. opinions.

This is extremely important for Stack Exchange visitors, those who do not have an account on Stack Exchange. The people who find our sites from search engines may not know that, for instance, bob123 usually gives great answers. Many inexperienced Stack Exchange users struggle with this concept and take it personally when asked to improve their posts. The users, those of us who are here a lot and who post, tend to know who is more trustworthy and who is not. But we're not the audience...

When a visitor comes to the site, we want that person to leave knowing that they received a solution to their problem. If that person must go and independently verify information they find on Stack Exchange, then it's possible either the answer is incomplete, or the question isn't answerable.

On Stack Overflow, it's fairly easy to tell if an answer is good or not. Code either works or it doesn't work. Sure, there may be better solutions, but they're all pretty much easy to test very quickly.

But on a site like The Workplace SE, it can be tough to verify a claim. A work relationship that goes sour due to some bad information on the site isn't something that can be easily undone. So on these sites, we ask people to provide some evidence that what they are claiming works and in which scenarios it does and doesn't apply.

For me, that's the key: If you're making me have to go verify information you've posted, then either the answer is incomplete, or the question is unanswerable.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .