Retitling Questions: a modest proposal

I'd like to suggest that after a Question receives an Accepted Answer, some consideration be given to revising the title (if appropriate) to reflect what the real issue turned out to be.

It seems to me users often pick titles when first posting a question that are uninformative and are worth revisiting once the actually point is identified. Improved subject lines will help in searches, and by extension help with identifying duplicates.

High rep users, it seems to me, are not excessively shy about making changes to tags, which helps with later searches somewhat. But I think a good title trumps all that, and the SE search engine can use all the help we can give it.

Discussion?

• Now that I posted this, I realize it is not so specific to Math.SE. Maybe I'll look at meta.SO to see if it has been discussed there. – hardmath Mar 12 '13 at 1:30
• I think it is MSE specific. It related to the culture and behavior of this particular community. It is a good proposal for other sites as well, but it is a specific one too. – Asaf Karagila Mar 12 '13 at 1:32
• @AsafKaragila: Fair point. Math.SE is a nearly ideal case (IMHO). The review queues stay empty or nearly so much of the time. Our community is closely knit enough that departures are noted. But we do have a problem with finding duplicates quickly enough. As André Nicolas commented recently, it may be easier to repeat an answer rather than run down a duplicate one knows well must exist. – hardmath Mar 12 '13 at 1:46
• Yes, I agree with your proposal, which is why I voted it up; and I agree with Andre, and even made similar comments myself. I think that we may also want to push for some community norm that if someone points that something is a duplicate, people will try to find it rather answer the question. Especially if it is an obvious duplicate. – Asaf Karagila Mar 12 '13 at 2:12
• Shall we eat our children? – cardinal Mar 12 '13 at 2:28
• In addition to re-titling questions to be more descriptive of their contents, we should probably also make a concerted effort to expand the number of questions tagged faq. This would also aid in the search of duplicates. – user642796 Mar 12 '13 at 11:52
• @cardinal: What about those of us without any children? – Asaf Karagila Mar 12 '13 at 20:09
• I, for one, am sometimes uncomfortable with simply retitling someone else's question due to my lack of experience. Is there any mechanism by which users can suggest a retitling of the question, and 1) the original asker can confirm the retitling; or 2) other users can vote up / down on the suggested retitling? Perhaps, if necessary, we could restrict this to high-rep users, for example those with at least 5k rep. – Vincent Tjeng Mar 14 '13 at 4:17
• @AsafKaragila They shall be eaten in lieu of their nonexistent children. – Did Mar 14 '13 at 12:09
• @VincentTjeng If the OP resents your edit, they can always revert it. But in uncertain cases I usually leave a comment like "Wouldn't sometitle make a better title?" - the OP will then either adapt or ignore it – Tobias Kienzler Mar 15 '13 at 13:47
• Concerning the dupe-search, if you know it has been answered by you (or some user you remember), search user:me is:answer some keywords, or replace me be the specific user's id (the number in the username link) – Tobias Kienzler Mar 15 '13 at 14:02
• Another approach would be to somehow motivate OPs to leave a comment saying something like: "Feel free to change the title of this question if you have something more descriptive", when appropriate. That way the OP can signal his own hesitation in choosing a title. – becko Mar 20 '13 at 3:31
• I feel like... Re-titling this question to reflect the real question better: Retitling questions to reflect the real question better. – user21820 Feb 19 '17 at 15:25

I think we should be more conservative about changing titles than retagging.

It's reasonable to suppose that someone searching for an answer to a similar question will be thinking along the lines of the OP's original wording, and not necessarily in the language of the answer. I can't see why anybody would object to revising a title after some discussion has clarified the language of the original question; retitling questions to reflect answers, however, may make it harder for users to search for answers to questions they are about to ask.

We already have too many users who post questions without first searching to see if their questions have already been answered. To keep the search process as easy and effective as possible, I think we should avoid revising titles to the point that they contain substantial language outside of the original question.

• It seems we have been more conservative about changing titles than retagging. My suggestion is to aggressively retitle questions, once an accepted answer is posted, where needed to make the subject line clearly reflect the question's actual topic. Of course it would be possible to edit titles in a less informative way, but I hope this thread will produce useful community guidelines. I searched meta.SO and found some ideas to relate. – hardmath Mar 12 '13 at 17:53
• I think accepted answers should serve as an indicator what the questioner was asking for in that should be the content of the title. But we have too many questions with titles like "Question on Analysis". – Michael Greinecker Mar 12 '13 at 18:08
• To clarify, I have no issue with editing titles like "Question on Analysis." But these sort of titles can be clarified to reflect the content of the question as soon as the question is posted. Simply put, I'd prefer to avoid new titles that reflect ideas introduced in the answers. – Brett Frankel Mar 16 '13 at 21:32
• It may well be timely to revise a title, based on the OP's clarification of a Question well before an Answer is accepted. I meant the presence of Accepted Answers as an endpoint, signifying that the OP felt "understood", so also a criteria for post-hoc querying, and not as a proposed source of words to put in the OP's mouth, so to speak. Yes, very often the misunderstanding of the original post is what makes a Question worth being more easily searched out, and we should avoid title edits that replace a cognizable fog with the bright glare of an articulate Answer. – hardmath Nov 8 '13 at 14:02

I'm for it. I have been making an effort lately to do this myself. I am sick of seeing titles like "Proof about a semigroup" (an example taken from the front page right at this moment). This provides absolutely no information, and there's no reason why the title can't be "In a semigroup, $ab=ba\Rightarrow (ab)^k=a^kb^k$." In fact, I don't see any reason to wait for an answer to be accepted to change it.

I agree with Brett's question about revising original wording, though. I think we should not change the title if it includes mistakes or bad wording which are descriptive of the question's content. As he points out, the whole point is to make the search better.

• My motives for retitling contingent on Answered Questions is threefold. First, it helps to ensure the nature of the Question has converged, something we've all seen I feel sure with new users or others whose formulation of the problem evolves with feedback. Second, it focuses on Questions and Answers that are worth finding, and thus improves the search process by boosting recognizability. Third, it provides a threshold window for the OP to alter the title themselves (saving us the effort and perhaps giving a result close to what new Questioners might expect, per Brett's note). – hardmath Mar 15 '13 at 14:00

I am going to use this post to catalog words that should generally not appear in question titles:

I'm not sure anything should (or could) be done about this, but it might be worth contemplating.

Most instances of quick (74 228 results) occur in conjunction with "question"; where they don't they seem to have somewhat higher quality—misspellings of "quicksort" and the like.

• Somewhat realated: Should we retitle posts whose titles contain “interesting”? – Martin Sleziak Mar 13 '13 at 17:29
• I am slightly tempted to make a robot that retitles all such questions, replacing "interesting" with "uninteresting". – MJD Mar 13 '13 at 17:34
• Just be sure to put a failsafe in case the software becomes sentient and decides to rename itself to SkyNET. – Asaf Karagila Mar 13 '13 at 18:38
• I can solve that problem by pre-emptively naming it SkyNET! – MJD Mar 13 '13 at 18:44
• In all fairness, the word "very" appears next to "ample" in some of these titles. – Brett Frankel Mar 16 '13 at 21:48
• Solution in the spirit of "modest proposal": announce that in one month, all questions matching these patterns (except "very" and "quick", which have legitimate uses) will be batch-deleted. If someone thinks they have valuable content posted under these titles, let them edit... – user147263 Jul 26 '14 at 20:09
• By the way, over on Meta Tavern they have a "Smoke Detector" bot that posts links to new questions with egregiously bad titles, network-wide. (Typically, all-caps, or with spam/offensive keywords). I wonder if something like this could be added to Math main chat room... – user147263 Jul 26 '14 at 20:19
• @900sit-upsaday, one problem I can imagine is that there are potentially many good questions of the form why is UVW hard/difficult? and why is XYZ interesting?. – Antonio Vargas Jul 26 '14 at 22:08
• @AntonioVargas I never saw a good question with such a title myself, but I'll take your word for it. Those words could be excluded, they are not the most frequent offenders on the list. – user147263 Jul 26 '14 at 22:14
• Are you really suggesting all these words be inserted into all titles??? ("...should not generally not appear...") I like that! – gnometorule Aug 12 '14 at 13:41
• @MJD Your robot could also replace "easy" (and possibly "hard") with "homework". – user1729 Aug 12 '14 at 14:00
• "Initial value problem" is a common example of a potentially good use of "problem" in a title. – Jonas Meyer Aug 12 '14 at 14:44
• Or Cauchy problem, Dirichlet problem, Neumann problem, n-body problem. Problem is a bit of a problem in that list. It may indicate a problem, but has a lot of false positives. – Daniel Fischer Aug 12 '14 at 14:57
• @DanielFischer Sure, but also see Gerry Myerson's response below. – MJD Aug 12 '14 at 15:05
• Don't want to bump the post again, so adding a comment: basic has 1220 appearances in titles. – user147263 Aug 20 '14 at 23:03

There are many many questions with titles such as "combinatorics problem", "about a combinatorics problem", "doubt in a combinatorics problem", and so on. I have just retitled three of them. I'll stop there (for the time being).

Searching meta.SO for "retitled questions" gives only eight hits (but some relevant ideas), while "editing title" returns more than 1,000.

One thread discusses What do you look for in a question title? and broaches the appropriateness of making changes.

Another thread, I've found a particular set of “question”s with bad titles; how should I proceed? sparked not only a discussion but a clean-up "posse". One target was (as in Michael Greinecker's example) titles containing the word "Question". Some work was broken out by tags, so one feels at least a tiny corner of the realm has been cleaned up.

I think priority should be given to reworking titles to improve not only the chance of a question appearing in search results, but the recognizability of the specific topic to someone glancing quickly down a search results list, where the title is prominent and accompanied by about four lines of text from the question or answer.