We have a corpus of "bad questions" with justification. Have we considered making a spam-filter-like pre-filter that can test the question for it's badness and suggest to the question writer "this question is likely a bad question for reason 'x' so please re-read the requirements and try again". Homework questions come to mind.

  • $\begingroup$ There has been some initiatives in other communities: see the smokedetector bot in the tavern on meta chat; and a chat room specifically created to detect low quality questions: chat.meta.stackexchange.com/rooms/773/low-quality-posts-hq $\endgroup$ Oct 7, 2016 at 19:39
  • 2
    $\begingroup$ @Andre you pointed at where SmokeDetector rents, not where it/he/she owns. Charcoal HQ is Smokey's headquarters. You can pay us a visit and see what we do. $\endgroup$
    – M.A.R.
    Oct 7, 2016 at 19:54
  • 1
    $\begingroup$ I'm not sure this is really something we can do. It seems more like something the SE developers would have to work up & implement. $\endgroup$ Oct 8, 2016 at 1:15
  • 1
    $\begingroup$ SE does already have their own spam protection; it already catches thousands of posts per day. Also, in case you were wondering, SmokeDetector does check all posts that are bumped from all sites (including this one), and as such most spam is deleted in a matter of minutes. $\endgroup$ Oct 8, 2016 at 5:15
  • $\begingroup$ I suggested something similar here meta.stats.stackexchange.com/questions/4242/… $\endgroup$
    – Sycorax Mod
    Nov 1, 2016 at 13:44

1 Answer 1


On some sites, the "ask question" pages will pop up just-in-time help when users seem to be getting ready to ask about certain topics. Of course, the specific topics and exact help text vary from community to community, but one example is regexes on SO:

screenshot of SO "ask question" page help pop-up, with links to regex resources

It's not a particularly smart system, though. It's really just half a step up from keyword matching. Creating a full-blown machine learning system, on the other hand? Could be an interesting challenge, but there's no immediate urgent need for it, it would take a lot of effort and add a lot of complexity, and the amount of marginal benefit it would provide (over keyword matching or some other simpler system) is unclear.

So depending on what you're asking for, the answer might be anywhere from "post a more specific request" to "sure is a nifty idea, but not practical for the near future."

  • $\begingroup$ How many user hours per day are put into question quality? How do you measure the value of "the amount of marginal benefit it would provide"? $\endgroup$ Oct 31, 2016 at 15:15

You must log in to answer this question.

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