2
$\begingroup$

While it is nice that we have MathJax to render $\TeX$ and $\LaTeX$ in our questions and answers, it may not be a good idea to use this in the title, for the reasons I have noted on CS.SE meta:

  • If a title contains 'plaintext formulas', such as Is O(n^2/2) <= O(n^2)?, do not edit it to replace it by Mathjax (which you should do if this occurs in the body of the question). You may replace it by 'unicode math': Is O(n²/2) ≤ O(n²)?, or remove the formulas completely What is the effect of halving in big-O notation?.

  • If a title already contains 'unicode math' (and is fine otherwise), don't touch it. I've seen people editing to replace unicode math with Mathjax in titles. Stop that.

  • If a title already contains Mathjax, you may wish to replace it by unicode math or remove the formula as in the first part. Just watch out for people willing to replace it by Mathjax again. Direct them to this answer, if you wish.

However, this policy (if any) differs a bit across the network, Mathematics tends to be neutral, in that simply being aware of the potential issues is a good idea.

I haven't found any relevant discussion elsewhere, but I can always add it if someone else has found something.


For an example, see this revision. Since Mathjax isn't rendered in the 'diff' markdown, this immediately shows the effect of non-rendered Mathjax in a title


So, what should our stance be on this? Encourage the usage in titles? Discourage it? Or neither?

| |
$\endgroup$
  • 2
    $\begingroup$ Do you have an example where this has been a problem? $\endgroup$ – Robert Cartaino Apr 12 '18 at 14:45
  • $\begingroup$ @RobertCartaino I've only seen it on this question so far, where I 'fixed' it: quantumcomputing.stackexchange.com/posts/1660/revisions . The main reason I ask this is because of the negative response and arguments on this post, of which I was a bit surprised (since the same suggestion was met positively on here). I wasn't sure whether it was viewed negatively because it was considered in the wrong place or form or as a bad idea, so I'd rather ask here to know for sure. $\endgroup$ – Discrete lizard Apr 12 '18 at 16:09
  • $\begingroup$ Discrate lizard, thank you. That's helpful. $\endgroup$ – Robert Cartaino Apr 12 '18 at 16:10
  • 1
    $\begingroup$ Tex is also used in the title of this question $\endgroup$ – Mithrandir24601 Apr 12 '18 at 16:29
  • $\begingroup$ @Mithrandir24601 Thanks, I suggested an edit. $\endgroup$ – Discrete lizard Apr 12 '18 at 16:47
  • $\begingroup$ ... which got rejected. Ah well. $\endgroup$ – Discrete lizard Apr 12 '18 at 17:52
  • 1
    $\begingroup$ @Discretelizard I've cleared out the LaTeX from the title of that particular question, and replaced it with the unicode symbol for lowercase Pi. I hadn't noticed this particular meta thread when you had suggested the edit. $\endgroup$ – Sanchayan Dutta Apr 13 '18 at 17:42
5
$\begingroup$

Generally speaking, you should try to avoid markup in titles wherever possible. The main reason is search. Titles are an important part of finding your content, so including the words/phrases folks will most likely type into a search box will help internet searchers find your questions.

So in order of preference:

  1. I would first try to entitle your question in plain-text English if at all possible.
  2. If an equation or other symbolic notation is necessary to make your question readable, I would look for a standard UNICODE solution before resorting to markup.
  3. Then in that rare rare case where the title simply does not make sense without very specific formatting, using TeX may be your only solution. This seems highly unlikely, although not impossible.
| |
$\endgroup$
  • 1
    $\begingroup$ A resource for math unicode is this: xahlee.info/comp/unicode_math_operators.html . Although searching for "unicode + [name of math symbol]" usually works as well. $\endgroup$ – Discrete lizard Apr 12 '18 at 16:49
  • 1
    $\begingroup$ "The main reason is search" is a pretty brazen claim coming from SE, given how absolutely broken the Stack Exchange search engine is, and how far behind it has fallen compared with search on multiple other places on the web. $\endgroup$ – Emilio Pisanty Apr 19 '18 at 10:15
2
$\begingroup$

This depends on the context and the details of the question at hand. A title's job is to describe as concisely and succinctly as possible the very core of the question. If that requires math, then cutting out the mathematics makes the title less descriptive.

Sometimes the core question can be described without pulling in any fancy symbols, in which case LaTeX notation should be cut. However, in many other cases, the core of the question centers around an object that really is best described in mathematical notation, and trying to dance around that just makes for a longer and more confusing title.

In that vein, here are some examples from the Physics Stack Exchange site,

and here are some examples from Mathematics,

as well as

from this site.

All of these titles compactly encapsulate in a few symbols of mathematical notation what would take a long sentence of text (and much more mental processing time) to describe.

Other sites, notably CS.SE and TCS.SE, take an opposite view, and discourage mathematical notation inside titles, which is to a large extent reflective of the fact that the content itself in (T)CS is much more amenable to textual description (with maybe a few complexity classes thrown in) than physics and mathematics. This site sits somewhere on the boundary between the two ─ which means that it does stand to get questions that are best described using mathematics in the title.

Furthermore, I contend that the primary purpose of titles is to be used for navigation within the site, i.e. they should be optimized with a primary use of display in question lists for both browsing and searching within this Stack Exchange site, where MathJax works perfectly.

That said, though, there are a few relevant points to be mentioned:

  • Use of titles by search engines. This is a red herring: the claim that the use of markup inside a title will somehow make posts more difficult for search engines, both internally at SE and externally, to index and process. That claim simply doesn't hold up, as the crawlers also have access to the full text, and it is that full text that should make a full clear textual description of the problem at hand and which forms the backbone of keywords that queries are matched against.

  • Display of titles in external search engines. This is the only real concern, and it is definitely worth considering and acting to minimize effects on. Most of Stack Exchange traffic comes from external Google searches, which display the un-mathjax-ed title and a snippet of the text; if the un-mathjax-ed title is unintelligible then that is a problem that needs to be weighed carefully, taking into consideration that (i) most people who care about the question are quite likely to be LaTeX-literate enough to get the gist of the title from the bare LaTeX, and (ii) the convenience of Google searchers needs to be weighed against the clarity of the title's display on internal listings within the site.

    So, if a title has mathematics, should you remove the MathJax in favour of clunky Unicode rendering? Well, not particularly, no. Here's a massively under-appreciated fact: MathJax is a rather broader standard than pure LaTeX, and it will happily render a bunch of unicode symbols. This means that if you want, say, $$|\psi\rangle\otimes |\phi\rangle$$ in your title, you can typeset it as $|\psi\rangle\otimes |\phi\rangle$, but you can also copy-paste the rendered output back into the text with dollar signs around it, $|ψ⟩⊗|ϕ⟩$, and it will render as $$|ψ⟩⊗|ϕ⟩,$$ which is mighty accessible and often indistinguishable from the "real thing" that some TeX purists might insist on. (A point of warning, though: the output can change depending on the chosen renderer, so have a play with right click > Math Settings before you make up your mind. Also, not all unicode displays perfectly so this does not apply uniformly.) Generally, the added friction of an extra $ or two in an expression like $|ψ⟩⊗|ϕ⟩$ in a title as displayed in Google is a really minimal price to pay for proper on-site rendering, particularly given that the audience is already LaTeX-literate.

    On the other hand, keeping the mathematics but removing the MathJax in favour of a half-hearted attempt in unicode is completely the wrong solution - it's the web 1.0 way to try and patch things before we had stable and robust solutions to displaying mathematics online. It sacrifices on-site readability for extremely dubious gains in external displays (which are mostly to an audience that can parse LaTeX to begin with) and there's really no real need for it, particularly when compared to the mathjaxed-unicode alternative.

  • Display of titles in other technologies, particularly RSS readers and the Stack Exchange iOS and Android apps. This is another red herring: there's nothing stopping RSS readers from incorporating MathJax rendering, and if folks really want to pull RSS content from mathsy sites without MathJax then that's on them, frankly. Also, this issue isn't unique to SE - there's plenty of MathJax'ed content from mathematics blogs that'll need proper rendering way before you need it for SE question titles. The same, but worse, goes for the SE apps: if they're not rendering MathJax, then that's a bug with the app, period.

In terms of actual actions, should LaTeX in titles be discouraged? That's really the wrong question, honestly: titles should describe as concisely and succinctly as possible the very core of the question, which often means that LaTeX markup is superfluous anyway, but when it isn't, it should be included. And, if LaTeX is included and the MathJax'ed-unicode displays equally well, then that's preferable.

And as for editing, the default should be to err on the side of inaction, i.e. do not edit other people's titles unless you are actually making them clearer and more descriptive of the core question being asked, much the same as the standard for other aspects of editing in titles.

| |
$\endgroup$
  • $\begingroup$ I'm not sure if I completely follow, do you say that there is no significant disadvantage of a title with MathJax versus a title without MathJax, all other things equal? And that therefore this isn't (shouldn't?) be an issue? $\endgroup$ – Discrete lizard Apr 19 '18 at 14:49
  • $\begingroup$ I disagree that "A title's job is to describe as concisely and succinctly as possible the very core of the question". IMO, a title on SE has two jobs: 1. to attract someone with the same problem as in the question and 2. to attract someone who can answer the question. Although what you say can sometimes help for 1 and 2, it ignores the point that a title should also 'sell' it's content. To achieve the latter, it is at times (I don't want to imply how often) needed to make a title slightly less specific, to make it simpler to read and get people to read the actual question. $\endgroup$ – Discrete lizard Apr 19 '18 at 14:51
  • $\begingroup$ I do agree that the default of editing is to do nothing unless there is an significant improvement, but note that improving a title (in clarity, 'cleanness' (grammar, style, general attractiveness)) is often significant. (However, it seems we do disagree on whether the action in question would be an improvement) $\endgroup$ – Discrete lizard Apr 19 '18 at 14:53

You must log in to answer this question.

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