What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 134 Stack Exchange communities.

(Turning this answer into a feature request.)

Stack Overflow is a repository of great and not-so-great answers. Three major causes for the not-so-great is misleading question titles, outdated answers, and poorly-answered duplicates with no connection to a better original question.

Anecdotally, today, a new user who had arrived from Google posted this "not an answer" underneath a question of mine:

@Mewp - 2 years after you answered the question, I ended up here (via goggle) looking for a solution on mod rewrite and the tutorial site you linked to is GONE!! Now what?? [...] So annoying that this ranks so highly in google for a lot of technical searches but unfortunately the site is full of unhelpful answers!

While it can be debated whether that question should exist in the first place as it is a resource request, the issue is widespread and goes far beyond resource requests. As a solution, Stack Overflow should consider voluntarily withdrawing questions from the Google index that gather some form of negative feedback.

Ways to collect feedback could be:

  • Ask the googlers. When a user comes in from Google, ask them something like

    We see you found this question while searching for how to flobber the gargle. Was it helpful and relevant to your query? Yes / No

    If a question receives too many "unhelpful" votes, either remove it from the Google index through robots.txt, or add it to an "unhelpful" review queue specialized on editing titles and such, or both.

  • Use the existing anonymous feedback data (Jeff Atwood himself has indicated that anonymous feedback could be useful for the cause) and exclude questions from the index that get overwhelmingly bad feedback.

  • When a user comes from Google, measure the time they spend on the page. If they leave it immediately with no action at all, interpret that as an "unhelpful" vote.

  • Also, perhaps questions with two or more close votes (except of course duplicate votes) should be withdrawn from the Google index. That would go a long way towards cleaning the search results from trivial garbage, too.

share|improve this question
2  
"Ask the visitors" just means anonymous visitors, right? –  Geobits Dec 4 '13 at 14:11
    
@Geobits yeah - namely those who came through Google. The exact search terms can't always be determined but the fact that they came from the search engine always can (I think) –  Pëkka Dec 4 '13 at 14:12
    
Just wondering because I come in through Google pretty often, even though I may be logged in. –  Geobits Dec 4 '13 at 14:14
1  
@Geobits it could be argued that in that case, you deserve to cast a vote, too. –  Pëkka Dec 4 '13 at 14:15
4  
Typically if I find something helpful through Google, I upvote it. I'm not sure if that would have more or less effect than a semi-anonymous "thanks-for-your-feedback" type vote. –  Geobits Dec 4 '13 at 14:18
2  
Shouldn't we rather just (1) fix the title, (2) fix and/or delete those answers, (3) post new up-to-date answers (or put a bounty on the question to let other users do this), (4) link to better duplicate? –  Dukeling Dec 4 '13 at 14:30
4  
@Dukeling sure, absolutely - but those are big problems and no one has come up with a comprehensive solution for them yet. In the meantime, it's not fair (and not in the site's long-term interest) to pollute the search index –  Pëkka Dec 4 '13 at 14:41
    
One of the reasons while questions asking for online resources are deemed off-topic. –  РСТȢѸФХѾЦЧШЩЪЫЬѢѤЮѦѪѨѬѠѺѮѰѲѴ Dec 4 '13 at 15:23
    
@ŁukaszL. true, but the problem is by no means limited to resource requests. –  Pëkka Dec 4 '13 at 18:07
7  
I like this idea a lot. Even understanding, as I do, why questions are closed, this can induce a desire to tear out my hair. This is #2 on the rank of frustrating results while troubleshooting, right behind the "thanks guys, I figured it out." dead-end forum threads. I have a problem and find a SO Google result that seems to match my needs exactly, only to discover that it was shut down for procedural reasons or is full of bad/non-answers... gah! The more I think about it, this might even be more frustrating than the dead-thread encounter. –  Chris Dec 5 '13 at 0:09
1  
I think they should stay in the index, because people coming from Google might answer the question if they don't like the current answers. –  ike Dec 6 '13 at 16:28
2  
@ike people don't search for questions that they know the answer to. –  Cruncher Dec 6 '13 at 16:31
1  
@Cruncher I once searched for a question on scifi.SE, and when I saw it had no answers, I thought of one myself. Besides, sometimes a low-quality answer can give someone an idea that can lead to a high quality one. –  ike Dec 6 '13 at 16:34
2  
@ike I'm sure that occasionally happens, but contrasted against the vast majority of incoming Googlers for whom this is a negative experience, it's exceedingly rare. Especially in light of many searches where Stack Exchange results make up the majority of page 1 on Google –  Pëkka Dec 6 '13 at 16:40
3  
"Some form of" should read "overwhelming" negative feedback particularly anon –  Jeff Atwood Dec 15 '13 at 2:15

3 Answers 3

Yeah, I hate this feeling when I'm looking for a solution to a problem for days. Finally, I manage to type a query into google in such a way that SO pops out. I say to myself: got it! I'm very happy. Then I click the link only to see that I'm not the first one with this problem, and it's unanswered since like 1970. I'm very sad.

However, it makes me add the answer when I'll manage to find solution myself, like in this question from 2 years ago.

So although I've up-voted this proposal in the meantime I've managed to find a counterexample showing that unhelpful result from google led to a good outcome :)

EDIT: sorry, just read the comment under the bounty. Hope that timestamps on the linked question and answer, and my assurance that it was found via google are credible enough for this answer to be somewhat helpful ;)

share|improve this answer

Ways to collect feedback could be:

  • Ask the googlers. When a user comes in from Google, ask them something like

We see you found this question while searching for how to flobber the gargle. Was it helpful and relevant to your query? Yes / No

If a question receives too many "unhelpful" votes, either remove it from the Google index through robots.txt, or add it to an "unhelpful" review queue specialized on editing titles and such, or both.

I can see a couple of potential problems with this solution.
I would guess that a great many people searching on Google are using search terms like:

"How to flobber the gargle?"

When in fact what they are really looking for is:

"How to flobber the gargle in this very specific context?"

And while the first post they find on Stack Overflow may be a perfect explanation of how to flobber the gargle, they don't know how to make it work in their specific use case, so they mark it as being unhelpful.

Likewise we are likely to see a lot of googlers who are searching for things that are a little over their heads. They are also likely to start out searching:

"How to flobber the gargle?"

But they don't yet know about flobbering or gargling, or how to use them properly, so reading a perfectly written post on how to flobber the gargle is beyond their understanding and so it isn't useful to them until they go back and read the documentation or search for:

"What is a flobber?"

and

"What is a gargle?"

We should't give anonymous googlers too much control over what appears in the search results because we don't want them to end up removing good content simply because they are frustrated at not having found a tailor made answer for their level of understanding and use case.

share|improve this answer
    
This is certainly a point to bear in mind, although I'm not sure whether it's this serious. Either way, there should be tests - asking Googlers could be done first and then the results looked at. Also, there is already anonymous feedback that Jeff says has proven useful to determine answer quality. Perhaps that should be looked at first. –  Pëkka Dec 13 '13 at 16:17
1  
@Pëkka Don't get me wrong it would be nice to have some narrowing of the search results, but why not base that on existing criteria like views, votes, and close votes? –  apaul34208 Dec 13 '13 at 17:17
    
the suggestion does not exclude that. But our activity on the site doesn't catch many of the problems - lots of rampant duplicates go unclosed because no one bothers to find a good original for the 10,000th time. And bad titles are generally not a reason for downvoting, but highly detrimental for the results Googlers get. –  Pëkka Dec 13 '13 at 17:24

Problems:

  • Confirmation that other people have my problem is often very useful to me.
  • You seriously think I use SE's searches to tell if my question is a duplicate? Of course not. I use Google.
  • I've answered questions I've tripped on a Google search too. Often after researching the solution myself.
  • I've also added bounties to unanswered questions.
  • Fewer people will visit SE. I don't see any way that can feasibly run with SE's business interests.

Putting this together, let's imagine the new workflow.

  1. Google a question that appears on SE without answers.
  2. It's not on SE.
  3. Go to the address bar, type "stackexchange.com," and start to ask the question.
  4. Probably half write it up, and see the useless question appear as a duplicate.
  5. Route 1: Click on it in a moment of excitement!
  6. THEN realize it has no answers.
  7. Or route 2: Post it as a duplicate. (Sorry, I'm not writing a real flowchart in markdown.)

This is strictly worse. You get more time wasted in half the use cases and more duplicates in the other half.

"But we're just wasting time of people who would find the answer somewhere else."

Great. Then show them our unanswered question, and after they go find the answer somewhere else they will come back here and answer it. This is why SO is by far and wide the first place I will look for an answer over any other forum or Q&A site and about 90% of the time over the official documentation. And if they don't? Then sounds like we didn't really care about wasting their time in the first place.

share|improve this answer

You must log in to answer this question.

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