Good idea, and it sounds like this can be done now by moderators and also given as a job in the future to this user.

By the way, to find out which ones to keep, here are some examples of retagging from Stack Overflow:

• The Great Tagging Reorg
• Tags: foobar versus foo-bar?

Continuing the previous topic, I think we could now try to write the Tagging FAQ, to be shown to 250--500+ rep users (I think once is enough, as with regular FAQ, no need to nag them all the time). Shall there be some wiki?

Update: a relevant discussion + explanation of tagging philosophy

• Could the tagging system be enhanced to support tag synonyms?

I'd be happy for homology and cohomology to be merged with each other but not with homological algebra. I would tend to skip past homological algebra questions but would avidly devour homology and cohomology ones.

I don't think homology/cohomology is the same as homological-algebra. I would use homological-algebra for questions about the machinery of Ext, Tor, long exact sequences, spectral sequences, etcetera. I would use homology/cohomology for questions about the homology of topological spaces, coherent sheaves or some other specific field in which homological techniques were used. Of course, there is an overlap, but I don't think they are the same.

A modular tensor category is indeed a monoidal category but one with approximately a zillion extra properties or pieces of structure. A typical example of a modular tensor category is the category of finite-dimensional representations of a finite group. A typical example of a monoidal category is a monoid viewed as a category with only identity morphisms. (I think you may have been confused by what is a pet peeve of mine, namely, people using "monoidal category" or "tensor category" when they really have a lot of extra structure in mind, such as enrichment in some abelian category. To a category theorist, such categories are the exception rather than the rule.)

I thought supersymmetry was some sort of physics thing that I don't understand. Super linear algebra is simply generalizing linear algebra to work with a funny twist in the R-matrix. I would avoid posts about supersymmetry and read posts about super linear algebra.

@Anton, re: community user

what you meant is that when somebody tries to use the tag [graphs], it is automatically converted to [graph-theory]

My idea was that Community User, a bot, can be given an automated task to continue to retag the questions that were right now retagged manually, but now I think synonym system would be better if people were told about them when posting.

Here's the feature request in case somebody wants to look/upvote.

What I meant is that the bot would continue with "fourier-transform" --> "fourier-analysis" and the other tags listed above. (I now think other solutions are more efficient for this issue)

Yes, they seem to be. I tried comparing 3 Shift-Reloads on summary pages and of questions with jsMath and the results are

Summary: 4s 6s 6s
Questions: 7s 9s 10s

I would say "jobs" sounds like actual job advertisements (which may be MO shouldn't have at all) while "career" for the general process of growing up in the profession, so, yes, merge the existing tags.

I think job advertisements are definitely not appropriate. just putting that out there.

Should [odes] and [pdes] be merged into the more popular [differential-equations]?

Yes, though probably someone will create them again.

Actually, I don't understand the [odes]+[pdes] merger at all. Here are two major areas of mathematics, big enough to have separate top-level codes in the MSC, lumped into just one tag, while much of the rest of the tag cloud seems pretty fine grained from my perspective. It makes no sens to me, unless it merely reflects the fact that there aren't a lot of differential equations people active at MO (yet). But I am not convinced that popularity should decide the granularity of the tagging system anyhow. Well, to a moderate degree perhaps, but this seems extreme. Actually, I find quite a number of tags which have apparently never been used at all, so popularity isn't an explanation. I don't know what it is, then.

@Anton: They are, but don't tell anyone because it's a secret.

Seriously, they are if you allow infinite dimensions but Picard only works in finite dimensions so if you like the classification should be:

ODE <-> DE+Picard
PDE <-> DE-Picard

More seriously (and back to the topic), I guess a reasonable policy would be to merge tags where they are used synonymously, rather than making it a statement about those fields of research, which seems to be your explanation for this case. As always, it is important to make the policy clear.

[complex-geometry] already exists, but [complex-manifolds] does not.

Should [complex-geometry] be merged into [cv.complex-variables]? Should [physics] be merged into [mp.mathematical-physics]?

I'm considering creating a [remove-this-tag] tag for the purpose of merging unwanted tags into it. For example, today I realized that the [newbie] tag exists. I'd really like to delete it (so that new users can't use it), but there is no way to do that, and I can't really think of a way to make that a feature-request because it could leave some questions with no tags at all. Does anybody object to [remove-this-tag]?

I want to get a mathematical physicists opinion on [physics]→[mp.mathematical-physics] before I do it. I'll try to find one today.

But they already have lots of ways to request retags if they can't do it themselves: they can leave a comment on the post, or (since they are the author) they can just ask in the body of the post. I would be pretty surprised if a 500+ rep user didn't respond to such a request within an hour.

I feel like [please-retag] indicates that there is a problem with the other tags (it sounds like that's actually what you had in mind), but [remove-this-tag] very clearly only refers to itself.

@Scott: Does adding a tag require less rep than retagging? I haven't seen that mentioned anywhere. (More, I see no way that I can add a tag to an existing quiestion, and I have almost enough rep to retag by now.) If not, I fail to see your point. [remove-this-tag] seems good.

But actually, I am very surprised not to see a feature that I thought would be an obvious one: When you enter a new question, there is no easy way to pick a tag from among the existing ones. You have to open the tags page and search it, then enter the name of the tag (correctly!) in the tags field. I think it ought to be harder to enter a new tag than to use an existing one.

@hanche: adding a tag requires the same 500 rep as retagging. The tag-suggestion feature should work all the time, but it might be slow for two reasons: (1) you're on a shaky internet connection, or (2) Fog Creek is getting a lot of traffic. Fog Creek has said that once SE is out of beta, the'll be doing much more optimizing for speed, so hopefully, that second possibility will eventually never come up.

I agree that [ap.analysis-of-pdes] isn't so good because it doesn't utilize the tag-suggest feature. I vote for [ap.partial-differential-equations], but I don't have strong feelings.

I see there are a bunch of questions labeled sg-symplectic-geometry. I assume it should be sg.symplectic-geometry.

Did [differential-goemetry]-->[dg.differential-geometry] and [sg-symplectic-geometry]-->[sg.symplectic-geometry]. Could somebody confirm or deny hanche's suggestion of [von-neumann-algebras], [c-star-algebras] --> [oa.operator-algebras]? My understanding is that [oa.operator-algebras] is a strictly broader tag, in the same way that [ag.algebraic-geometry] is strictly broader than [moduli-spaces], which would suggest that we should keep the more specific tags, but I don't know anything about operator algebras.

You're right that [oa.operator-algebras] is strictly broader than [von-neumann-algebras] and [c-star-algebras]. I guess I am still confused about policy after being puzzled about the [differential-equations] thing. But then, at least, everything that's tagged either [von-neumann-algebras] or [c-star-algebras] should be tagged [oa.operator-algebras] as well, right?

just did [quantum-algebra]-->[qa.quantum-algebra].

@hanche: I think the tags [odes] and [pdes] were a fine idea (I created them, after all), but people just weren't using them. Tagging is something that the community has to develop a standard for. Of course, we can influence tagging conventions a great deal, but ultimately the smartest thing to do is to look at how most people are actually using the tags, and merge/rename in such a way that it becomes easier for everybody else to follow the de facto conventions. I'm certainly not absolutely sure that merging [odes],[pdes] in to [differential-equations] was the right thing to do. If [odes] and [pdes] get recreated and people start using them meaningfully, I have no intention of interfering.

I don't think you created it, unless something else just got merged. There are 13 other questions with that tag.

Edit: This was in reply to someone who later abruptly deleted many of his posts. It seems we can no longer delete our posts here, but if a moderator would like to remove this one, feel free.

@hanche: done.