Not signed in (Sign In)

Vanilla 1.1.9 is a product of Lussumo. More Information: Documentation, Community Support.

    • CommentAuthorWillieWong
    • CommentTimeJul 12th 2010

    On Qiaochu and sigfpe chimed in.


    To my mind, tags are only useful if the keyword of the tag can't be found in the question. Any reasonable question I can think of which you might want to tag "inequality" probably already has the word "inequality" in it.

    When I read this, I was fairly convinced. Then I read a rebuttal from sigfpe:

    But the purpose of tags isn't simply to provide words to search on, in addition to those in the text. If I click on a tag I get a list of questions that have been tagged that way, not a list of questions containing that string. So tagging a string already in the text serves a useful purpose.

    For the time being, ignore the fact whether "inequality" is an appropriate tag for that question (personally I think "no", but for different and specific reasons from the general statement Qiaochu suggests). The question is, what is the purpose of the tags?

    A few days ago I removed the "yang-mills-equation" and "yang-baxter-equation" tags from a couple posts. I justified what I did to myself thus:

    1. Each of the posts were the only ones using those tags

    2. Each of the posts had the phrase ("Yang-Mills" or "Yang-Baxter") in their titles.

    3. Tags like that seem to be overly specific in light of the fact that a post can have 5 tags. A question on Yang-Mills, for example, can reasonably go under gauge-theory or dg.differential-geometry. And Yang-Baxter can go under statistical-physics and mp.math-phys. (I should note that those tags were already present on those questions.)

      3'. To encourage overly specific tags makes it difficult for people to follow/filter areas.

    Now to ask an actual question.

    Should tags on MO be more like keywords or more like categories?

    • CommentAuthorAndrea
    • CommentTimeJul 12th 2010
    More like both. Specialized keywords may be useful for searches (but in this case they are not necessary if they appear in the title), while generic ones are useful for filtering (say following the RSS feed for ac.commutativealgebra)

    Since searching in MO is so primitive at present, I would suggest allowing tags even when the words appear in the text. Of course, there is at present a limit of 5 tags, so it's best to choose them judiciously.

    tags provide a more flexible (and occasionally more annoying) structure than categories (which almost always never quite work the way one wants them to). So I'm more for the keyword approach, and with somewhat redundant tagging.

    Removing a tag just because the tag text appears in the question is a bad idea. Remember, many people use tags to mark questions as "interesting" or "ignored", not just as searches, and there's no way to reproduce this based on the question text.


    I still agree with Willie's original assessment that the tags about specific equations were too specific.


    Scott makes a good point that I hadn't thought of. However, I can't see a tag like "inequality" ever being used for filtering; whether you do or do not want to look at a question about an inequality probably depends strongly on whether you're familiar with the methods likely to solve it, which will probably depend on the other tags on the question.

    My "tagging philosophy," such as it is, is that the point of MO is to efficiently redirect expert attention to questions where it is needed, and tags should allow experts to most easily find the questions they are best in a position to answer. So general tags are great, and specific tags should be used when they correspond to specific expertise. (Then, of course, there are tags like "soft question" which some people may just want to filter out.)

    • CommentAuthorWillieWong
    • CommentTimeJul 12th 2010

    To address both Scott's and Qiaochu's points: the filtering based on tags afforded by the software only becomes effective if we have a good definition for each tag. The tag 'inequality' can potentially be a well-defined one, but as it stands it is completely overloaded. Very few of us here on MO will say on our research interest that "I'm primarily interested in inequalities" without further clarification. :) And here we are grouping elementary inequalities together with inequalities from classical/harmonic analysis, functional analysis, a priori estimate from PDEs, combinatorics, theoretical computer science and more all under one roof.

    A propos the equation specific tags: I seriously doubt anyone following Yang-Mills will filter dg or gauge-theory or vice-versa. Having such overly specific tags actually, I think, makes it hard for people to follow/filter things. Why? This goes back to the 5 tag limit again. If we encourage the use of specific tags, a possible trade-off may be the disuse of the "general" tags. I do not want to have to follow 14 different tags just to see all of differential geometry, and I especially don't want to filter 25 tags just so I don't accidentally wander into the wrong place and expose my ignorance of algebraic geometry.

    Also want to say that I generally agree with Qiaochu's tagging philosophy.

    • CommentAuthorsigfpe
    • CommentTimeJul 15th 2010
    I completely agree with Qiaochu's philosophy, just not with his original reason given for removing that specific tag.

    > I do not want to have to follow 14 different tags just to see all of differential geometry

    I wish I could put a relation on tags so that picking certain tags automatically picked the transitive closure eg. so picking differential geometry would automatically pick subdisciplines and subsubdisciplines.

    sigfpe, could you elaborate? I am looking at the list

    of all questions tagged "inequalities," and they're all over the place. I do not think this is a particularly descriptive tag.

    • CommentAuthorsigfpe
    • CommentTimeJul 15th 2010
    If I didn't make it clear, I agree it's a bad tag. But not because it duplicates text in the question. It's bad for the reason you gave in this discussion: it doesn't efficiently direct people.