Based on the recent, upvoted answer to the old character tag question:
If suggest reading the full post for background information, and some of the other, many discussions on character tagging.
If it takes an acrimonious discussion for every tag to establish whether or not it's appropriate, the test you're using is broken. You're letting the perfect become the enemy of the good... Find a simpler test, one that folks can apply in common cases (release of a new book / movie) without having to tear each others' throats out here on meta.
This statement is made because our currently policy is fairly unclear. It uses terms such as "lots of questions", "significant fraction", "do you think", which lend themselves towards subjectivity, and the arguments we have.
@Shog9 clarified further in his comments:
Tags are a utilitarian form of organization, @Aith; they're messy by design. We have some rules in place to help avoid complete chaos, but utility should always take precedence over organizational ideals that can't easily be achieved in practice. This is my problem with the status quo: folks are apparently wasting more time arguing about which tags are unnecessary than they are on adding those which are useful.
A folksonomy is only useful if the meaning of a tag emerges based on how it's used; you can't assign a meaning other than what emerges naturally... and if none emerges, you can't force it to. As with language itself, you have to be willing to examine how people use it to understand what they're saying.
Based on that last part, I would think it could be a good idea to design the test around how people actually use tags, or want to use them, if possible.
I'm not aiming for this to be a discussion about whether or not we should have a new simple test. I'm aiming to see what types of solutions we can come up with towards our problem. Perhaps later we can vote as to whether/when to implement a suggested change. Let's play it hypothetical and constructive to start, and see where that gets us.