The text for "Primarily Opinion Based" is, across all SE sites:

Many good questions generate some degree of opinion based on expert experience, but answers to this question will tend to be almost entirely based on opinions, rather than facts, references, or specific expertise.

I'm mostly familiar with Stack Overflow, but I've tried to write this question in a way that doesn't exclude other sites. I'm looking for a fresh perspective.

There are so many questions that just want a solution*. And they get it, often without any explanation of why that solution is good.

I don't want just any solution. I want the best one and I want to be convinced. But nobody tries to convince me, so I rarely accept any answers.

I want to ask that elusive "good question" that asks for expert opinions based on facts. In a programming context, that might be a comparison of two algorithms, for example.

It seems like the divide between a "good question [generating] some degree of opinion based on expert experience" and something that needs to be closed as primarily opinion based is a slippery slope. At least that's what I've seen. If there are two sets of facts that lead to opposite conclusions, should the question be considered "too opinionated"?

How can I go about creating that "good question" without getting closed?

* On Stack Overflow, the term we use for these questions is "gimme teh codez" questions. I am sure that there are a number of "homework" and "do my work" questions on other sites.

  • 2
    There's an entire blog post about the subject... blog.stackoverflow.com/2010/09/good-subjective-bad-subjective It's linked to on every "Don't ask" page. – Catija Apr 27 '16 at 19:06
  • 1
    Explicitly ask for rationale and justification? That's what I do. – Dan Bron Apr 27 '16 at 19:08
  • @Catija I read that already. Unfortunately, the blog posts are not community written, and do not fall in line with the opinions of the masses all the time. – Laurel Apr 27 '16 at 19:10
  • 1
    Interesting. Can you be more specific? What kind of topic would you theoretically like to ask about? Different sites within the Q&A network deal with totally different subject matter, and the way each community handles the nuances of subjectivity without going off the rails varies accordingly. – Ana Apr 27 '16 at 19:11
  • The "opinions of the masses" vary from site to site... other than the SE policy on the subject, this really is something you'll get a different answer from everyone on... or at least every site. If you have a specific site you're interested in, you should consider asking this on their meta rather than on the SE meta. – Catija Apr 27 '16 at 19:11
  • @Catija I considered asking elsewhere, but I decided that I wanted to get different perspectives. I assumed that there are several sites that are somewhat technical like SO, and therefore have similar philosophies on "opinion based". – Laurel Apr 27 '16 at 19:19

It seems like the divide between a "good question [generating] some degree of opinion based on expert experience" and something that needs to be closed as primarily opinion based is a slippery slope. At least that's what I've seen.

I moderate a site that deals with this (The Workplace, pretty much everything there is "opinion"). In my experience, here are some good guidelines about characteristics of good questions:

  1. Scoped.
    • Don't ask "how can I figure out ~huge problem~." Questions cannot require books to answer.
    • Don't ask "how can I figure out ~problem without enough detail~." But you need enough information to answer the question.
  2. Actual questions.
    • Validating your perspective makes a bad question. "Am I right?" or "please tell the other person they are wrong" are common XY problems. A lot of bad subjective questions fail this category.
    • Do you have a question? Or is it a discussion. This is where a lot of others fail. Can you summarize your question in an actual question or is it basically "what do you think?" or "what are your opinions?"
  3. Short(ish).
    • Generally longer questions tend to convey too much information, which means either it's super specific to your situation. Not always the case, but often a good sign.
    • A few paragraphs is ok. And probably good. But 10 paragraphs? Not good.
    • Super short questions normally fail the "Scoped" category. Not always the case, but often is that way.
  4. Focus on the problem.
    • Do NOT focus on the solution you expect or you will get answers about that. Focus on the problem.
    • This is really, really, really hard to do.
    • Ask about the "how to make decision" or at least contain words about this in your question.

I want the best one and I want to be convinced.

And how are you convinced? By facts or just someone else's opinion? By facts of course. If a question only attracts opinionated answers, it is not possible to distinguish the right answer from the wrong one.

Every site has its own line on where 'primarily opinion based' starts. It is impossible to draw a line for every site at once.

  • Should a question always be closed for being opinion based when there are two answers that are opposites? – Laurel Apr 27 '16 at 19:21
  • Not if they can be validated on facts. – Patrick Hofman Apr 27 '16 at 19:22
  • If you think that's true for all SE sites, then you should add it to your answer. :) – Laurel Apr 27 '16 at 19:26

What SE generally looks for is answers that are convincing to a reasonable person, and questions that bring those. A reasonable person can often be convinced by specific experience in something, so those answers work out well. For example, on RPG, if someone has actually played a campaign with a particular houserule, that's generally a good basis for an answer. Presumably, on DIY, someone who has actually tried installing a particular brand of appliance can give good answers on the details of making that work. And so on and so forth.

A reasonable person can also generally be convinced by a logical argument. If there are specific assumptions and proven statements to build on, it can often be practical to justify good practices using nothing but sound reasoning, without having to actually try it oneself.

Often these go together; someone's specific experiences in certain things can be fit together with other observations that are objectively proven to make a good argument. Or more general experience can be backed up with some well-chosen statistics that show it's not just someone's vague hunch.

So, of course, questions that aren't specific enough to be brought down to either a chain of reasoning or a specifically-applicable body of experience are not going to work well. Questions that end up with contradicting answers that both (seriously attempt to) justify their differing results are fine: likely, one is wrong and one is right, but there should be enough there for good voters to make a sensible judgement call and end up with appropriate scores. Or sometimes the situation really is so complex that the state of the art has not settled on a single definitive answer... but has collected some useful arguments that can be learned from and applied in perhaps slightly different situations, or based on one's own leanings.


I think that restricting the question to a specific field area along with a given definitions for why you're asking this question might not get it closed.

Don't just ask, Should I use Angularjs or jQuery? or Which is better, Angularjs or jQuery? This (IMHO) implying that the user is lazy and didn't put any effort in research before asking that question.

But if you define the field in which this question is relevant, then you might get some answer and not just instant downvotes:

  • What are you currently working on? ("I'm developing a SPA")
  • What are my requirements? ("I want to toggle a DIV on click" VS."I need to separate my business logic and my display logic")
  • If you have any doubts you want to consult? (I've read that... but i'm not sure if...)

This is how i'm asking broad questions and I did get my answer without getting it closed as being too broad.

You must log in to answer this question.

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