What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 131 Stack Exchange communities.

I've lost count of the number of times I've either asked a question or been answering a question only to have it closed as an exact duplicate. But searching for the question turns up a list of seemly unrelated results using the sites builtin search. Then if I switch to a general web search engine and limit its scope to just SO, SF or SU I find the question within the first two or three results.

Why is this?

I'm pretty sure this isn't an exact duplicate of another question though I did find some that were at least somewhat related (using google).

Update:

I also have gotten (somewhat) better results by simply typing a question and browsing the generated list of related questions. Still a search feature that doesn't return good results isn't going to be used by most people.

Common sense would suggest (though common sense isn't always the best indicator) that a targeted solution to a specific problem domain should be superior to a generalized solution. In most cases this appears to hold true.

share|improve this question
    
I was about to say Duplicate but decided to refrain. However not sure what the reason for the feature request tag is? –  Diago Jul 31 '09 at 13:49

5 Answers 5

Q: Why does [insert favourite trilogy site here] provide better answers on [site focus] than google's own answers?

A: Different purposes.

Google use complex weighting to choose pages, SO probably uses a fairly straightforward db search.

share|improve this answer
    
nowadays anyone can have a score-oriented, document-based search engine. –  Mauricio Scheffer Oct 26 '09 at 19:14
    
And google still has the best. You wouldn't expect a fish and chip shop to have better pizza than a pizzeria, even though they both do it, would you? –  Phoshi Oct 26 '09 at 20:13
1  
I disagree. Google deals only with html (in this case) while stackoverflow has all the exact semantic information for each question/answer/user, if they used lucene/solr/xapian there's a lot of search specific optimization they can do, yielding better results than Google –  Mauricio Scheffer Nov 5 '09 at 1:33

I have also switched to use the "Ask Question" page instead of the search box to search for questions. Some moments ago I was searching for "Reusable functionality in asp.net mvc":

  • the search box returns 17 pages of (mostly unrelated) questions
  • when entering that as the title on the "Ask Question" page, I get what I'm looking for (in the first position of the search results)
share|improve this answer

I think these comments are based on the performance of our search engine as of a week ago. Particularly your "wow, the /ask page search is great!" comment, which almost by definition means you don't know about the changes we've made.

There have been some substantial changes under the hood since then, try looking at it again:

http://blog.stackoverflow.com/2009/07/stack-overflow-search-now-61-less-crappy/

(note: I'm not saying it doesn't still suck, but I don't think your comments are based on current search results..)

share|improve this answer
1  
Actually this question was born out of frustration from having spent a half-hour of thought and research to answer a question on SU only to have the question closed as a duplicate before I finished typing. Granted I have no way of knowing whether the user that asked it bothered to even do a search at all. It was frustrating none the less. –  Kenneth Cochran Jul 31 '09 at 14:01
    
I always attempt to use the built in search first, despite its passed lackluster performance before trying a web search. I will say I have noticed the accuracy marginally improve recently. I hadn't seen the blog but I assumed some changes had been made. –  Kenneth Cochran Jul 31 '09 at 14:04
2  
well, asking users to never, ever post duplicates is kind of a difficult thing. Can't you add your answer to the question it was a duplicate of? Good answers are always welcome, on whatever question.. –  Jeff Atwood Jul 31 '09 at 14:10
    
Coincidentally, the user had only asked the question an hour ago and claimed they had searched for similar questions before asking to no avail. –  Kenneth Cochran Jul 31 '09 at 14:12
    
Please have a look at my response if you need a proof of how much the search box sucks (and how much better the "Ask Question" page is). –  M4N Oct 20 '09 at 21:34

Because google has spent a lot more (time, money, people) on improving search than Jeff and the guys have. Incidently, more often than once I have found a dup of my to-be-asked question when getting the "possibly related" list after typing in the title, although I hadn't found it even through google.

share|improve this answer

Because [insert favorite search engine here]'s prime responsibility is delivering good search results, while SO, SF, or SU's primary goal is to deliver an overall excellent Q&A system.

share|improve this answer
    
Doesn't this also entail connecting its users with the questions and answers they are searching for? –  Kenneth Cochran Jul 31 '09 at 14:08
    
@Kenneth, yes... but you can't expect to index better than companies that make billions a year on search alone. It's not perfect but it's reasonable. This is like comparing a go-kart to a ferrari. –  Ian Elliott Jul 31 '09 at 15:05
    
It is possible and fairly simple to implement. See Lucene, Solr, Sphinx, Xapian, etc. –  Mauricio Scheffer Oct 26 '09 at 19:15

You must log in to answer this question.

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