Short version: Is this the same as a question "without a definitive answer"?
Long version: Personally, I understood "subjective" to mean, "without a definitive answer", yet I keep seeing questions on this subjective question site closed because they lack definitive answers. I understand the requirement that not all answers be equally valid to mean that there should be some reasonably obvious criteria for determining if a certain answer is "more valid" than another answer. This could be "correctness", but since subjective questions often defy correct / incorrect assessments, I would think it could also be some other reasonable criteria, such as "thought-provoking", which maybe should be specified in the question. I think a good litmus test for this is, "Could you downvote something for being less valid, and support your claim based on the question?" E.g., "Down-voted because this is common sense and not thought-provoking." Does this seem like a correct understanding? Can you provide me with a better interpretation, so I don't run afoul of this rule again? Should this be clarified in the FAQ?
Commentary: I suspect a non-official reason people understand this requirement as "needs a definitive answer" is that it stops threads where everyone has something to contribute that can easily grow very large. People enjoy these questions, perhaps partly because everyone can contribute meaningfully, so more people read and respond and they get a lot of votes. While I understand frustration when this is blatant "rep-farming", I think there is at least one class of questions which are best answered by accumulating a lot of answers and letting the cream rise to the top. This becomes a sort of process of "community inference", and lets the community explore topics that maybe aren't yet understood well enough to have a "right answer" yet, or maybe are too complex to ever be fully answered. While I agree that the reputation system has an undesirable result for these questions, I feel like the problem is with the SE system and not with the questions.
I am especially frustrated because I find that many software test questions fall into this category, and right now programmers.se seems to be the best site available for these questions (I've voted for the QA Area 51 proposal, but it's stuck in the "commit" phase). Lots of the work that testers do requires inferential analysis, and being able to draw from wider ranges of experiences with questions like these is a big help!
Disclaimer: I did post a question that falls into this category. My goal wasn't to farm rep, and I considered if it was appropriate to mark 'community wiki', but decided against it based on this meta thread on the purpose of 'community wiki'. I was frustrated when it was closed but glad to see discussion over the closure (followed by it being re-opened), and thought the discussion was really good and the general topic deserved a full thread on meta.