Does the tag system need revamping?

Question

I think one of the less refined features of SE is tagging. Even here on meta:

tag-blacklist-request === burninate-request ?

There are loads of questionably useful tags on Stack Overflow, here's just a few random ones that took no effort to find:

future height dynamic coupon combine words styling function old on-the-fly once dogfood

Newer users often tag questions almost the same way they wrote their title, something like:

Where should I put my config file?

[php] [where] [put] [config] [file]

It was my understanding that for a tag to deserve its existence, it should be able to stand on its own. Is that too narrow of an assessment?

And of course there are so many tags that are nearly identical.

It seems like a lot of time is spent dealing with synoymns, retag requests, blacklist requests and the like. I generally feel like most of these are a waste of time.

In general I think the tag system works well, but could be better. One idea I had was using one set of "tags" for the language and another for the more minor details or sub-categories, for example:

[javascript] [html]
_{[onclick] [closures] [variable-scope] [textarea]}

Another thought I had was peer review for creating new tags. It seems even 6K is not enough to know how to tag properly. (see revision 2)

I don't know if this is really a problem, or if it's localized to Stack Overflow. I can see some of these ideas being overkill on smaller sites.

My big question is: Is there anything that could be done to improve the tag system, or is everyone happy with the way it is? If there are no changes or desire for change, I'd feel a lot better doing janitor work with the tags. Any plans on the back burner from the dev team? Any ideas or suggestions from the community?

Answer 1

Tags are often bothersome.

Thing is, the faq already suggests doing something like what you propose:

At a minimum, try to include at least one very broad tag (i.e., java or c#) and one other tag to narrow the topic down within that broader category (i.e., strings or garbage-collection).

And I don't think it's good, because it gives rise to maximum (hey, I'm in the top ten answerers in maximum, I'm an expert), and all the sillyness. If anything, tags should be restricted to languages, but that begs for the question what's a language? What about frameworks? What about things like jsf? (JSF is a good example, because it often occurs together with the Java tag, and while I do know some Java, JSF is absolutely foreign to me.)

The current system is really very simple, and that's a good thing because of flexibility and ease of use. Introducing subtags as a feature (as opposed to a faq suggestion which is hardly ever read) would introduce an added complexity to the site which would make it somewhat less accessible for newcomers, and it would add another step in asking a question: "Now select a language. No, that's not a language, you should select a language. Good. Now describe the area of your problem in narrower terms." -- it wouldn't be very pretty, and the situation would remain the exact same anyway, as people would still tag badly.

The current system, despite the many hiccups, works fairly well because people care about tags and do janitor work on them, and it all kind of averages out pretty well eventually. Yes, it has many problems, and it's often really frustrating, but I don't think the problem can be solved: we are morons, we use wrong tags, and nothing could ever change that.

Now, to be a bit more constructive: peer-reviewing new tag suggestions would indeed help, and I'd totally support that.

(Under the current system, I actually see a reason for wtf, as questions about arcanely mysterious phenomena in the vein of wtfjs.com. Having a badge in wtf would mean you're well acquainted with these. But I digress.)