I have several times used the search function to see if someone has previously asked a question I am interested in, with no positive result. I then start asking my question and then very quickly the suggested 'Questions that may already have your answer' finds more or less the answers I was looking for.

This tool seems really well implemented, I like that it keeps updating as you type so you can see instantly the different results obtained by changing words (useful for me since I come from a field that uses some non-standard terminology so I'm always fishing for the correct term).

I guess my question is why is the search function so clunky in comparison? I've simply given up using search and now launch straight into asking a question, even when I'm sure the question exists somewhere, just because it is easier to find answers that way.

  • 3
    $\begingroup$ I've noticed that myself. I don't know what the deal is w/ it; it does seem a little strange. Another approach is to click on a tag that is related to what you are interested in, which will return all the questions associated w/ that tag, that sometimes works better than the search feature also. I still do use CV's search feature sometimes. It can be helpful for finding a question when you remember the question loosely (less helpful in your case where you don't know). Google also works sometimes. $\endgroup$ May 29 '12 at 0:10
  • 2
    $\begingroup$ I use the search function and it seems to work well for my purposes. What's an example of a search you've done that didn't find what you wanted? $\endgroup$
    – mark999
    May 29 '12 at 9:57
  • 1
    $\begingroup$ @mark999 I thought someone would ask that... It's tricky, since you're proposing a post hoc test. I'm sure I could state any subject I was trying to search for and the appropriate search terms could be found that appears demonstrate that the search capability works just fine. But that doesn't demonstrate that it works well for an imperfect user before they know the secret combination of terms. My point is that on a number of occasions I have had much better and much more rapid success just typing in a question rather than using search. $\endgroup$ May 29 '12 at 23:38
  • 2
    $\begingroup$ Not sure, but items 3 and 4 in this answer might be helpful, if you haven't seen it already. $\endgroup$
    – cardinal
    May 30 '12 at 13:13
  • $\begingroup$ @Bogdanovist I wasn't proposing any sort of test. It seems strange that you criticise the effectiveness of the search function but you're not even willing to provide an example. $\endgroup$
    – mark999
    Jun 2 '12 at 9:42
  • $\begingroup$ I didn't make myself clear enough. I'm not saying "search doesn't work". If I was, I could provide an example "I tried to search for logistic regression and search returns nothing", which is an objective fact that can be confirmed or refuted. This is not what I am claiming. What I am reporting is that in my experience, I have found it quicker and easier to find existing questions by starting to ask them and waiting for suggestions than by using the search function. An example of why would involve reporting every dumb mistake I made using both before finding what I was after. $\endgroup$ Jun 5 '12 at 6:11
  • $\begingroup$ So what I am saying is that "I always find it much easier to find things by trying to ask a question than using search". That's irrefutable, since it's a question of my subjective experience, even if it is because I'm using search wrong. The main point is that SE have put a heap of effort into an awesome dynamic system for suggesting questions as you type, but stick with a older style query-response type search function which doesn't really encourage you to actually use it. $\endgroup$ Jun 5 '12 at 6:17

According to a SO meta post from 2010, the results returned when asking a question are simply from searches with "intitle:1" added to the query ( Why does intitle:1 show questions whose titles don't have all my search terms? ). However, the "intitle" operator appears to have been deprecated. Another 2010 SO meta post suggests pre-pending a query with "title:" will return results more similar to the "ask a question" search. For example:

  1. what is confounding https://stats.stackexchange.com/search?q=what+is+confounding returns 4,981 results.

  2. prepend intitle:1 https://stats.stackexchange.com/search?q=intitle%3A1+what+is+confounding returns 4,981

  3. prepend title: https://stats.stackexchange.com/search?q=title%3Awhat+is+confounding returns 950 results, which seem to be mostly in the same order as typing "what is confounding" into the ask a question page.

  4. prepend title: to each term https://stats.stackexchange.com/search?q=title%3Awhat+title%3Ais+title%3Aconfounding returns 940 results.

I'm not sure what the technical difference is between 3. and 4., probably slight differences in the weighting, but the results do not differ enough to justify prepending "title:" before each term. More information on advanced search operators is available: https://stats.stackexchange.com/search

And some more discussion of SE search from the blog:

To summarize:

Someone with a better understanding of the inner workings of search on SE is needed to provide a definitive answer. However, the probable reason that you have more success with the "ask a question" searches is because they are more limited in scope, searching only question titles. The main search function, while it gives more weight to hits in question titles, still returns many more results that have the query in the body of questions and answers. Incidentally, I think the main search tends to ignore comments.

  • 1
    $\begingroup$ I asked a somewhat related question on SO meta regarding the SE-wide search not returning results that are returned in site-specific search. The question has not gotten much traction, but I'm interested if anyone has any insight. $\endgroup$
    – jthetzel
    Jun 5 '12 at 20:58

You must log in to answer this question.

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