Are general tag discussion threads ever useful?

Question

I've seen “tag spring cleaning” meta threads on many sites, where people are invited to raise problematic tags, one tag per answer. They usually end up in a mess: after the first few, nobody pays attention to anymore, and being able to discuss things only in comments is stifling (is this tag really problematic? should it be disambiguated? should the meaning be clarified? …).

Such threads are pretty common on private betas, too, owing to a superficial reading of the blog post that's linked on private beta meta¹. Those threads usually end up in a mess too, for exactly the same reason as the spring cleaning ones.

Is there ever a case where a general tag discussion thread is useful², or should tag discussion threads always focus on a specific tag or a specific family of thematically related tags³?

¹ _{Which should be this clearer and more up-to-date post instead, but that's another matter.}
² _{I confess to doing it myself once, when I had less experience of private betas, and it didn't end up in a mess — because it didn't end up with one tag per answer. But that was rather unusual.}
³ _{e.g. “should we have tags for individual countries/authors/species/…”}

Answer 1

I think trying to handle more than a single tag or a small group of closely related tags in a single Meta Q&A is of very limited use and should be avoided.

At GIS Meta I asked a tag question for a different purpose and it morphed into becoming the repository for all tag requests on that site before I started to haul it back to its original purpose. You can see some of that at Make suggesting synonyms privilege that kicks in at reputation of around 1,000?

Out of that experience we created Guidelines for proposing tag synonyms which basically says that for each tag proposal (covering one tag or a group of related tags) there should be a separate Meta Q&A which starts with someone describing the tag problem that they have identified in a neutral way, and then lets the answers be used for any proposals on how to resolve it, so that they can be voted on.

I think the way we do it now is working well.

Answer 2

Discussing a general philosophy of tagging is fine. For example, "Should we have character tags" or "Is it ok to create a tag for a specific API if people are asking questions about that API?"

Discussing practical etiquette for tagging (perhaps with a few examples) is good too. Things like, "Should I add a language tag when my question is merely about solving a problem using that language".

But when your goal is to discuss the merits of specific tags, it helps to be... Well, specific.

In addition to the problems you note (lack of attention, difficulty in interpreting the results), they tend to suffer from something akin to riders - tags added late that are treated the same as the rest of the group, even if they're actually different enough to warrant their own discussion. You might recall a couple of high-profile instances where the discussion started out focused on problems endemic to a small set of very similar tags, but a few months down the road readers were cheerfully applying the same treatment decided on for those to a dozen or so other tags (at best tangentially-related) that'd been tacked on at some point.

Over the years, we've seen a lot of time wasted on such discussions. It's hard enough to get folks to sit down and debate the merits of a single tag; trying to hash out a consistent result for multiple unrelated or barely-related tags tends to drive away everyone who isn't hell-bent on just getting rid of the whole lot. For that reason, I explicitly forbid "two-fers" in my guidelines for burnination.

I would strongly recommend avoiding these whenever possible.