Many people use tags to filter out questions they view. (For example, using favorite tags, ignored tags, filters, by regularly visiting some tags or following them via e-mail notifications, by viewing only questions in selected tags, etc.)

One of the effects is that some questions might go unnoticed because they are only tagged by some obscure tag and they are not using any of large tags with many followers. In other words, correct tagging can influence how many people actually see the question.

Do you think that it would be useful to have a set of tags such that every post on math.SE should contain at least one of these tags? If yes, what would be a reasonable candidates for top-level tags?

Apart from the possible problems with low visibility of questions using niche tags, top level tags might also help with better organization of the questions on the site and with searching. (For example, if all questions on are also tagged and you are looking for a question which definitely belongs in (functional-analysis) but it might be also tagged (lp-spaces), you would only have to look in the top level tag.)

I do not think of this as some formal policy, which would be enforced by software. I simply suggest, that if we have a list of such tags, users who retag questions could add one of the top-level tags if they are missing.

At MathOverflow they have a list of top-level tags and it is recommended to use at least one of them in a post on that site. Some of those tags are available also here, at least as synonyms.1

Probably some of the tags in the among the most popular tags might be reasonable candidates for top-level tags.

1I use the name top level tags in the same meaning as on MO. I do not know whether this phrase has some special meaning on some other sites in SE network.

  • $\begingroup$ I've been thinking about this for some time too, yes (but the problem of making a good list of toplevel tags...). $\endgroup$
    – Grigory M
    Commented Jan 1, 2015 at 10:57
  • $\begingroup$ ...and maybe it should be enforced by software — many new users tag posts only with one very narrow tag... $\endgroup$
    – Grigory M
    Commented Jan 1, 2015 at 10:58
  • 14
    $\begingroup$ @Grigory: Or they tag with number theory, logic, real analysis, abstract algebra and integration when they ask about solving the system of equations $x=y,\ x=0$. $\endgroup$
    – Asaf Karagila Mod
    Commented Jan 1, 2015 at 11:20
  • 1
    $\begingroup$ I think this would go and lead a life of its own, to the point where the solution is going to be worse than the problem it tries to solve. $\endgroup$
    – Lord_Farin
    Commented Jan 1, 2015 at 11:33
  • 1
    $\begingroup$ The list would be very long. It would be very hard to make it exhaustive too, what happens when a legitimate question doesn't fit in the list? There are some things to sort out too, like the distinction abstract-algebra vs algebra-precalculus (or number theory vs elementary NT), are they both top-level? I also don't see an easy way to deal with small but unique tags, like learning, teaching... $\endgroup$ Commented Jan 1, 2015 at 12:04
  • 2
    $\begingroup$ @NajibIdrissi What happens when a legitimate question doesn't fit in the list? In such cases we either find out that another tag should be added to the list or we simple let the question be an exception to the rule. (You may notice that not all question at MO have a top level tag.) To me this seems similar to the question: What if someone asks a question for which we do not have a tag? $\endgroup$ Commented Jan 1, 2015 at 13:48
  • $\begingroup$ @Martin In this case I don't really see what's the difference between a top-level tag and a regular tag, if there's no software enforcement and we can have exceptions / new top-level tags. $\endgroup$ Commented Jan 1, 2015 at 14:18
  • 1
    $\begingroup$ Since this is tagged "discussion" and not "feature-request" voting should (IMO) be on the merits of having the discussion (as opposed to the proposal it self). $\endgroup$
    – quid Mod
    Commented Jan 1, 2015 at 14:24
  • 3
    $\begingroup$ @NajibIdrissi Exactly as you write, top level tag is just a regular tag. But having a rule that almost all questions should contain one of these tags (if a good list of such tags is compiled) might help better organization of the questions on the site, searching and visibility of the questions. $\endgroup$ Commented Jan 1, 2015 at 14:29
  • 3
    $\begingroup$ As an aside: some questions in (lp-spaces) belong to (real-analysis) rather than (functional-analysis). One can use the notation $L^p$ for a particular class of functions on real line without any concern for the structure of this class, considering only the properties of individual $L^p$ functions. $\endgroup$
    – user147263
    Commented Jan 1, 2015 at 15:55
  • $\begingroup$ @quid I agree that voting on this post might be unclear (some user might use voting to indicate whether they consider such discussion useful, some users might voting as indication of whether they agree with the idea of top level tags). But there's not much to be done about it - voting is different on meta and both on main and on meta people are going to upvote/downvote for various reasons. $\endgroup$ Commented Jan 1, 2015 at 16:26
  • $\begingroup$ Would it be useful to have a "meta meta" question about the tag system as it enters a new phase of maturity? We have questions like this one, and also about tags which reflect the level of mathematical sophistication. Or is the evolutionary model the best? For example, should every post ideally have at least three tags - one being of this kind, indicating a general area; the second being elementary, first degree, postgraduate; the third a specific subject designation. And how would this relate to the tag badges. $\endgroup$ Commented Jan 4, 2015 at 22:03
  • $\begingroup$ This older posts are a bit related: Should MathOverflow tag prefixes be used? $\endgroup$ Commented Apr 10, 2019 at 12:08

1 Answer 1


To set the context, let me just mention that I am one of the main supporters (and "enforcers" as time permits) of the top-level system on MO.

Further, the list of top-level tags on MO was not created on MO but "imported" from the arXiv, the top-level tags correspond essentially exactly to the math categories of the arXiv.

This is a blessing and a curse. The blessing is there is a hardly any discussion what is and is not top-level, might better be or rather should not anymore...it is fixed externally and that's that. The curse is that it does not fit completely the needs of the site. But for MO it is quite alright; it is about research math, as are the math categories of arXiv. So pretty much any piece of math relevant to MO should have some correspondence; indeed, there is "General Mathematics" as "last resort;" and, even, for not completely math-y questions, there is "History and Overview."

For this site, that list would be rather unsuitable though. To come-up with the list could be an issue, and I doubt there will be enough consensus to make this feasible in practice. In brief, I think it is a good idea in principle but not feasible in practice.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .