30
$\begingroup$

This is a proposal inspired by the overwhelming number of questions with uninformative titles on the site. Such titles tend to contain words like question, anyone, help, doubt, easy, exercise, homework, problem, stuck, etc. There are many of those...

Call for action

Consider improving the titles of the questions that you have answered. The best time to do this is right after you answer the question, since the answer bumps it anyway.

Awful titles get overlooked much too often. As an extreme example, I still remember a question titled "Proof or Theorem" (sic) which received three answers, all from 2K+ rep users, none of whom edited the title or commented upon it. (It's been edited since).

Having already put the work into writing an answer, adding a little extra to help your answer be found makes perfect sense. Also, coming up with a title for a question that you already read and thought about (even a while ago) should not be very hard.


Older questions

This query presents the list of titles that likely need improvement:

Questions with (likely) uninformative titles answered by a user

(UserId is the number in the URL of your profile page.) The query also shows current tags. When editing a title, it's best not to forget the tags and formatting of the question, either...

As noted above, the ideal time to fix the issues with title, tags, and formatting of a question is right after answering it, when there is no extra bump involved. But better late than never. Usual caveat applies: don't flood the front page with rapid-fire edits. Also, consider the quality of the post itself. If it's at or below the current average of the front page, then the editing effort is better placed elsewhere.


Inspirational statistics

Data Explorer shows the percentage of questions with the word "question" in the title. Among all SE sites with at least 30K questions in total, which site is on top? Drumroll please...

  1. Mathematics 4.27%
  2. MathOverflow 2.67%
  3. Physics 2.05%
  4. English 1.8%
  5. Statistics 1.54%
  6. Electrival Engineering 1.16%
  7. Programmers 1.09%
  8. TeX 0.5%
  9. Server Fault 0.41%
  10. Sharepoint 0.38%
  11. Drupal 0.34%
  12. Stack Overflow 0.33%
  13. GIS 0.28%
  14. Wordpress 0.26%
  15. Ask Different 0.24%
  16. Ask Ubuntu 0.23%
  17. Gaming 0.17%
  18. Unix 0.16%
  19. Super User 0.12%
$\endgroup$
  • 6
    $\begingroup$ "better late than never." Not sure I agree. I wouldn't want to see the front page flooded with old questions that have no business being there, other than that someone improved the title. I think retitling old questions should be done very sparingly. $\endgroup$ – Gerry Myerson Jul 27 '14 at 0:01
  • 4
    $\begingroup$ @GerryMyerson Being bumped to the front page brings about an hour of fame at current rates... but I added the usual caveat nonetheless. $\endgroup$ – user147263 Jul 27 '14 at 0:04
  • 9
    $\begingroup$ Being bumped to the front page bumps something new off the front page. Bumping 50 old posts to the front page bumps everything new off the front page. The new posts deserve their hour of fame. Well, some of them do. $\endgroup$ – Gerry Myerson Jul 27 '14 at 0:07
  • 15
    $\begingroup$ @GerryMyerson Which is an interesting point, by the way. Activities that improve the content of the site should be done very sparingly, while activities that make it worse (e.g., posting a ton of repetitive hw questions) are carried out with abandon. Where does this lead? $\endgroup$ – user147263 Jul 27 '14 at 0:14
  • 6
    $\begingroup$ @900, it leads to the current situation, which is bad but nevertheless perhaps optimal. Some (most?) of the old questions that have crummy titles are repetitive hw questions, so bumping new hw questions for old ones is at best a marginal improvement of content. $\endgroup$ – Gerry Myerson Jul 27 '14 at 0:19
  • 6
    $\begingroup$ @GerryMyerson Actually, I disagree with "it leads to the current situation". So far it has led to the current situation. From the current situation it leads to a worse one, and from there to worse... until we may just as well be on Yahoo! Answers. ("Quick math question! 10 POINTS!?") It's more of free fall than equilibrium. $\endgroup$ – user147263 Jul 27 '14 at 5:08
  • 8
    $\begingroup$ The title of a question has a large influence over its search rankings. More people see and benefit from a question if it has an informative title. I think that bumps, when done reasonably, shouldn't dissuade us from improving titles. After all, if one wants to look at new questions only, one can check /questions instead. $\endgroup$ – Ayman Hourieh Jul 27 '14 at 10:16
  • 2
    $\begingroup$ Obviously, the real point is that SE shouldn't bump answered questions to the front page just for a title change or small tweak.. $\endgroup$ – Thomas Andrews Jul 28 '14 at 16:27
  • 3
    $\begingroup$ @ThomasAndrews The algorithm of the front page could definitely be smarter. In fact they use a smarter algorithm on Stack Overflow but unfortunately not here. (I use it for my own filtered front page, though.) In recent weeks, SE leadership talked a lot about redesigning the front page, but mostly in the context of Stack Overflow, again. $\endgroup$ – user147263 Jul 28 '14 at 16:34
  • 6
    $\begingroup$ So maybe we have settled on "Feel free to fix awful titles of new questions, and strategically fix awful titles of old questions if they are interesting enough to warrant the bump."? $\endgroup$ – rschwieb Jul 28 '14 at 17:50
  • 2
    $\begingroup$ @rschwieb Yeah, I think the paragraph with the caveat says about as much. Focusing on the questions one has answered is one way of approaching this strategically, since it's likely that one has found something interesting there. $\endgroup$ – user147263 Jul 28 '14 at 18:04
  • 3
    $\begingroup$ @rschwieb Another aspect, which goes a bit beyond the topic, is whether the questions that aren't worth fixing are worth keeping... $\endgroup$ – user147263 Jul 28 '14 at 18:06
  • $\begingroup$ @GerryMyerson I guess the system is built to bring old questions to the main page sparingly. $\endgroup$ – Felix Marin Aug 2 '14 at 8:54
  • $\begingroup$ It seems to me that the 'right thing' is to fix the titles and to improve the front page ranking algorithm. One hopes that improved titles will reduce the need for 'me-too' questions (involving $- { 1 \over 12}$)? $\endgroup$ – copper.hat Aug 2 '14 at 12:53
  • 1
    $\begingroup$ @900sit-upsaday: Oh well, I will focus on zapping the $-{1 \over 12}$ questions :-). $\endgroup$ – copper.hat Aug 3 '14 at 2:25
9
$\begingroup$

Some people prefer not to dig out old questions, and that's okay. There are plenty of recent questions that have the same problems. A quick way to find them is to use either

  • the bookmarklet Titles likely to need improvement (this is not the actual bookmarklet, but a link to the page from which it can be dragged), or

  • Chrome extension Review+ Page, which supersedes the functionality of the bookmarklet.

The bookmarklet can be executed from any page of the site, and it shows a list of recently active questions with questionable titles. Questions on this list may actually be old, but they had recent activity which bumped them.

What is captured by this script

  • "Smoke detector" words, which were taken mostly from MJD, with later additions from the commenters:

anyone, difficult, doubt, easy, exercise, hard, help, homework, interesting, math, please, problem, question, someone, struggl*, stuck, urgent, very, wrong ... (and others)

  • Titles with at most 3 words (precisely, with at most two whitespaces).

  • Titles that begin with a lowercase letter. Not that wrong capitalization is a disaster, but such titles tend to be sloppy, as the posts to which they are attached.

  • Titles that contain display math $$ or non-ASCII symbols.

  • a few other signs of weak titles, observed empirically

$\endgroup$
3
$\begingroup$

I have two specific suggestions.

First, I want to remind everyone to put the relevant formulae right in the title wherever possible! If your question is about calculating the limit of some function, put the formula for the function right in the title. I personally think this should be done even with quite long formulae, but everyone's mileage varies. If we're talking about integral calculations, you can tell from just the formula whether you (a) want to answer the question, and (b) are even capable of answering.

Second, I'd like to bring up one subtlety that I suspect might be often overlooked. Consider these two hypothetical question titles:

  • How can I calculate a derivative?
  • How can I calculate this derivative?

The former looks like a general methodology question. When I open it, I expect to be giving a conceptual answer with probably more words than formulae. The latter looks like a question about a specific example. When I open that, I expect to be giving an answer which is just a computation with some explanation.

Please keep in mind the distinction between these two types of questions. If it's not possible to include the specific formula in the title, remember that simply explicitly saying "this function" or "this problem" in the question title will remove the ambiguity.

$\endgroup$
  • $\begingroup$ ...or perhaps describe the formula in general terms: How can I find the derivative of this product of two logarithms? $\endgroup$ – user147263 Jul 30 '14 at 21:29
  • 1
    $\begingroup$ @900sit-upsaday Yes, but my point is that unless the product in question is 6 inches long, go ahead and throw it in the title. $\endgroup$ – Jack M Jul 30 '14 at 21:33
  • 2
    $\begingroup$ It's my understanding that on certain platforms formulas in titles make it harder to use the site. There were discussions in meta about this. For example, meta.math.stackexchange.com/questions/10617/… $\endgroup$ – Gerry Myerson Jul 31 '14 at 0:23

You must log in to answer this question.