At this point, it seems like there is subset of users who draw the line on "primarily opinion based" at a place that is very different from my expectations based upon the history of the site. In fact, arguments I'm seeing today as justification for closure seemingly apply to questions asked long ago that were well-received.
Some examples of questions that appear to be varying degrees of subjective, yet which were not only accepted, but actively embraced by the community:
As some of you might know, Are all Stormtroopers as poor soldiers as the ones in the movies? has one of my all-time favorite pieces of content on this site. Yet it is also an example of a question that I believe is likely to get marked by some users as "primarily opinion based", save for the last sentence asking for canon sources. In fact, the overwhelmingly popular answer doesn't even answer the question, but rather subverts the assumptions of the question altogether, demonstrating that the premise that the Stormtroopers in the movies were all "poor soldiers" was completely inaccurate. Other answers lack canon references at all, or include n-canon.
Could the terraforming techniques from Red/Blue/Green Mars work on any other planets in our solar system? was called out as an example of the types of questions that might be considered to be showing the characteristics of a great sci-fi question. Granted, it was called out by the user who asked the question, but both the meta answer, and the actual example question, were well-received.
Was Yoda fully sane on Degobah? - A very popular question (and answers) that would seemingly be asking for an opinion.
Why exactly do the Death-Eaters serve Lord Voldemort? - Also very popular, and again, seemingly asking for an opinion.
Is archery underestimated in the wars of a Song of Ice and Fire? - My answer, which is both the accepted answer and the highest voted, has no basis in canon. It has received no downvotes (as of the time of this post, at least). I agree that TLP's answer is excellent, and I have upvoted it. Yet the question, as posed, seems to be little different from things like "Why don't Jedi use the force more often?" or "Why don't we see more dual-phase lightsabers?". Yet those questions are listed as example arguments as to why "canon-only answers, please" is "cheating", or "violating the spirit of the law", when the archery question only addressed canon by appending "Is it a fact archery are little used in Westeros and Essos?" to the end.
It seems, going by historic acceptance, that there is a place for some level of subjective questions on this site.
Yet clearly some sort of subjective questions are problematic, and need to be fixed (or, failing that, closed). The preference should be fixing it, of course, but sometimes either that isn't possible, or the OP ignores the advice. We don't want questions that lead to discussion, or where every answer is equally valid.
The definitive description of "good subjective vs. bad subjective" for Stackexchange remains, as far as I know, this blog post by Robert Cartaino:
Guidelines for Great Subjective Questions
Great subjective questions inspire answers that explain “why” and “how”. The best subjective questions invite explanation. If you’re asking for a product recommendation of some kind, you want answers to contain detailed information about the features and how they can be used, and why you might want to choose one over the other. “How?” and “Why?” has more lasting value than a bunch of product-feature bullet points or a giant enumerated list, no matter how extensive. In contrast, the bad subjective questions let answerers get away with hit-and-run answers that maybe provide a name and a link — but fail to provide any sort of adequate explanation, context, or background.
Great subjective questions tend to have long, not short, answers. The best subjective questions inspire your peers to share their actual experiences, not just post a mindless one-liner or cartoon in hopes of being rewarded with upvotes for being merely “first.” Sharing an experience takes at least one paragraph; ideally several paragraphs. If I’m asking about how to bake cookies, don’t give me a list of grocery items: milk. butter. vanilla. eggs. There is virtually nothing I can learn from a short, static list of grocery items that make up a recipe. Instead, tell me what happened the last time you made cookies from that recipe! Share your detailed experiences, so that we all might learn from them.
Great subjective questions have a constructive, fair, and impartial tone. The best subjective questions avoid the all too seductive route of ranting and flamebait. They set the right tone of constructive learning and collaboration from the very outset, by emphasizing that we’re all here to learn from each other, even if we have different viewpoints or beliefs about the right way to handle what are inherently subjective decisions. We’re not here to fight each other; that’s an enormous waste of everyone’s time. There is always more than one right way.
Great subjective questions invite sharing experiences over opinions. Certainly experiences inform opinions, but the best subjective questions unabashedly and unashamedly prioritize sharing actual experiences over random opinions. It’s more useful to share with us what you’ve done than what you think. Everyone has an opinion. It takes zero effort or imagination to have an opinion about anything and everything. But people who have done things, real things in the world, and have the scars and arrows in their back to show for it — now that’s worth sharing. You should be uniquely qualified to have your opinion based on the specific experiences you had. And you should share those experiences, and more specifically what you learned from your experiences, with us!
Great subjective questions insist that opinion be backed up with facts and references. Opinion isn’t all bad, so long as it’s backed up with something other than “because I’m an expert”, or “because I said so”, or “just because”. Use your specific experiences to back up your opinions, as above, or point to some research you’ve done on the web or elsewhere that provides evidence to support your claims. We like you. We want to believe you. But like wikipedia itself, {{citation needed}}. And good subjective questions make this clear from the outset: back it up!
Great subjective questions are more than just mindless social fun. The best subjective questions avoid the social pitfalls of “Getting To Know You” (GTKY) and mindless entertainment. Sometimes people just want to poll a community for ideas that might help solve a problem (best book, best approach). These can be okay when there is actual knowledge in the collection of answers. What isn’t okay are the social bonding questions which are designed just to impress others, such as “What is the coolest/stupidest/weirdest/funniest thing you saw/did/tasted today?”, or questions where the site’s actual topic is tacked on as a token afterthought, such as “Favorite food for programmers.” If you removed the “for programmers” part of this question, is it really unique to our profession? Could an average member of our community reasonably be expected to learn something that makes them better at their job from this question? If not, then it’s a bad subjective question.
So, there you have it: the difference between a good subjective question and a bad subjective question — expressed as six simple guidelines. If you’re wondering if a particular subjective question is worthy, wonder no longer. Apply the six subjective question guidelines and see how it scores. If the score is low, close it. If the score is high, vote it up.
Criteria #1 seems to cover a lot of the justifications I've been seeing for closing things as "primarily opinion based", and I feel that's wrong. Questions asking "why is x seemingly underused in title y?" are explicitly called out as being one of the criteria of good subjective. We shouldn't be using that as justification for closure, let alone the sole justification.
Item #3 is the one that I feel highlights the warning signs for our site. If the question violates this impartiality (it is based upon or promotes a pet theory, or focuses on criticism rather than seeking explanation), then it needs to be edited to be impartial. If it can't be edited to be made impartial, then closing as "primarily opinion based" may be the correct action. Note that discussion of pet theories does not necessarily indicate it is "bad subjective"; if there is evidence to support a particular theory, it may be okay.