I recently posted Show that a metric space has uncountably many disjoint dense subsets if every ball is uncountable. and it quickly got up-votes, and then an answer with gaps was posted which got one up-vote. Thus the question wouldn't show up in unanswered questions searches. But the answer posted still seems to potentially have gaps even after editing, and still only has one up-vote, and so imo ideally the question should still show up in some form of unanswered question searches. Has there been any discussion on creating a "smart" ranker of questions for searching which simultaneously ranks according to time since question was posted (lower is better), number of up-votes for question (higher is better), and number of up-votes for the best current answer (lower is better)? This may attract attention of users who know a good answer to a popular question that has only received one iffy answer with one up-vote.

  • 1
    $\begingroup$ Interesting thought. But isn't the target fraction of posts adequately accessible by bounties? The case you mention seems a good candidate for one. $\endgroup$ – Lord_Farin Oct 10 '13 at 17:06
  • 3
    $\begingroup$ @Lord_Farin You're right, bounties do fix the issue I brought up. I will issue a bounty in two days if I don't get a good answer. But issuing a bounty means I have to give up reputation, and it would be nice if a search mechanism circumvented this need. Not everyone will want to issue a bounty. Basically, I think good questions without good answers (e.g. just one up-vote per answer) should be accessible to users via search. $\endgroup$ – user2566092 Oct 10 '13 at 17:11
  • $\begingroup$ I'll delete the question if it's inappropriate for some reason, can down-voters say why they down-voted? $\endgroup$ – user2566092 Oct 10 '13 at 20:21
  • 4
    $\begingroup$ Downvotes on meta usually mean nothing more than that someone disagrees with the proposal. In this respect meta voting is a bit different from voting on main. $\endgroup$ – Lord_Farin Oct 10 '13 at 22:03
  • $\begingroup$ To add to what Lord_Farin wrote: voting on Meta is explained on the Meta help $\endgroup$ – Willie Wong Oct 11 '13 at 7:34
  • $\begingroup$ I have a funny feeling that this is actually how SE does its ordering. I'll have a dig about to see - they have some sort of ranking system, which, for example, gives preference to tags you follow. It isn't just plain ol' "most recent post" ranking... $\endgroup$ – user1729 Oct 11 '13 at 12:25
  • $\begingroup$ Perhaps this is what I was thinking of: meta.stackexchange.com/questions/197907/… Incidentally, I think your idea is very good and I do not understand why it has such a low score. Perhaps someone can explain to me why they think this isn't a good idea? $\endgroup$ – user1729 Oct 11 '13 at 14:12

If you come up with a function that adequately combines the parameters you mentioned, the next step would be to head over to Data Explorer and code it into an SQL query. This is the easy part.

The hard part would be to get someone to actually use that query (or an equivalent feature of the search interface, in the unlikely event it gets implemented). People looking for questions to answer have $>28500$ question with no upvoted answers to choose from, many of which have no answer at all. The idea of looking through additional thousands of questions with low-voted answers (most of which will be correct answers that are underappreciated) is hardly appealing.


You must log in to answer this question.

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