What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 130 Stack Exchange communities.

I made a simple web app that 'lints' SO questions. It basically points out a lot of common problems that lead to questions being closed, and potential solutions.

I think it should be linked in the comments of a question that's either on-hold or has a couple close votes. Hopefully it'll reduce the number of questions that are closed. The main goal is to be brief and targeted, as opposed to something like an SSCCE - Short, Self Contained, Correct (Compilable), Example.

I'm looking for suggestions for some rules to add.

I mostly hover in the webdev areas of SO, so I'd be largely interested in tag-specific problems, because I don't see/understand a lot of those questions. Also, any patterns that have been noticed by Stack Overflow users; especially users who are involved in the review process.

  • The most common reason I vote to close is _______.
  • A lot of questions are missing ________.
  • Questions with the ______ tag are often closed because ______.

Examples with Lint:

Example without Lint (and 6 upvotes at time of posting):

Also, if you're interested in seeing the current rules in code, and suggesting technical improvements, this is staticlint.coffee.

(I realize this is on the borderline of StackApp, but my question is about patterns of Stack Overflow questions, which wouldn't be able to be answered elsewhere.)

share|improve this question
1  
Not related to your question but your app seems to be slightly wrong; you can't always "complain" that there's too much code and also state that the OP didn't try to help themselves? EDIT: Just seen your "it's difficult to do"... fair enough. –  ben is uǝq backwards Jul 27 '13 at 11:31
4  
You're also not following the CC-Wiki attribution requirements as defined by SE in the Terms of Service... I like the general idea a lot though... if you can get something working it would be really good to be able to define automagically what exactly might be wrong with a question. –  ben is uǝq backwards Jul 27 '13 at 11:34
    
@FakeRainBrigand, I gave you an upvote, in principal, this is a an excellent idea and I take my hat off to you for your efforts to improve our site :) –  user223277 Jul 27 '13 at 11:34
    
@benisuǝqbackwards you should put your comments into an answer. I think you have made good points –  user223277 Jul 27 '13 at 11:35
    
Well, I for one hope that question closures are not nearly this formulaic. If they were, we wouldn't need a lint-style tool, we would just need to tighten up the question quality filter so that these questions didn't get posted in the first place. –  Cody Gray Jul 27 '13 at 11:38
    
@CodyGray yes I agree with you, but it could be altered to cater for different tags, as some language questions require code examples, some do not, they require an explanation –  user223277 Jul 27 '13 at 11:42
1  
This is a pretty cool idea. Interested to see how it works out. –  Pëkka Jul 27 '13 at 11:45
1  
I don't have an answer to the question though @Skippy; personally, I think that any single "rule" will not work as the type of questions are far too varied. I get told off for everything being to long on this question; it's almost impossible to shorten it. I would probably increase the amount of text "allowed" but in some circumstances that would be ridiculous; for instance this comes out okay... because it's short. –  ben is uǝq backwards Jul 27 '13 at 11:48
    
@benisuǝqbackwards, have a recommendation on how to comply? Would turning the title into a link satisfy the attribution? I could change it to a single code block exceeding X characters or Y lines. –  FakeRainBrigand Jul 27 '13 at 11:48
    
According to the legal page something along the lines of the following would be enough: small footer saying it came from SE. Question title as link back to the question; you also need to include the author though with a link back to their user-page. –  ben is uǝq backwards Jul 27 '13 at 11:50
    
@CodyGray, this will through out quite a few false positives. That's why someone needs to look at it, and decide whether the lint is accurate enough to post. –  FakeRainBrigand Jul 27 '13 at 11:50
    
@benisuǝqbackwards yes I understand, it is not a straightforward, nor easy thing to 'successfully' implement (if possible) but it sure is interesting. I also made the comment that the rules would need to very, depending on tag –  user223277 Jul 27 '13 at 11:53
    
I would be very interested in seeing the results of such a test, once you get this up and running. –  user226423 Jul 27 '13 at 11:57
1  
Just a +1 for the effort. It's a pretty cool thing you've created. If it only ever servers to inspire, you've already done well. –  Bart Jul 27 '13 at 13:59
1  
Did you enter any part of this in the Stack Exchange machine learning contest they ran last year? If no, you might want to write the team and see if this would be of interest, because this seems in line with what they might do with some of the entries from that contest. –  Brad Larson Jul 27 '13 at 20:40

You must log in to answer this question.

Browse other questions tagged .