A short look into the overview statistics have shown, that SE.AI has a poor percentage of answered questions.[1] Only 71% of the questions were answered in the past. A detailed list, which kind of of questions are remaining unanswered in the system is given by a second statistics.[2] It shows only questions which have no comments right now. That means, the OP has dropped a month ago his problem and nothing happens since then. No negative feedback, no positive comment, no nothing. The amount of postings with this criteria is around 20-30 per month.

A naive assumption might be, that the question quality was poor. Indeed, some questions are in the system which containing only of two sentence which is a bit short. But many questions in the list are very elaborated, are written in an academic style and enriched with additional pictures. But this doesn't improved the answer-probability.

Could it be, that with the questions and the user who have created them everything is fine, but SE.AI struggles in handling the incoming load in an appropriate way?

[1] Stackexchange all sites, https://stackexchange.com/sites

[2] Stackexchange Dataexplorer, Unanswered questions without comments, https://data.stackexchange.com/ai/query/548333/unanswered-questions-without-comments


I'm not sure how big of a deal this is, since out other stats are adequate.

That said, at some point, it might might make sense to start a deletion campaign of old, unanswered questions, not just to improve the metric, but to clear away the junk.

  • 1
    $\begingroup$ See Enable automatic deletion of old, unanswered, zero-score questions after a year?. $\endgroup$ – nbro Oct 14 '19 at 17:52
  • $\begingroup$ If i have understood the statement right, the idea is that unanswered issues get escalated to the supervisor by deleting them? $\endgroup$ – Manuel Rodriguez Oct 14 '19 at 20:42
  • $\begingroup$ @ManuelRodriguez I'm not aware of any escalation mechanism. $\endgroup$ – DukeZhou Oct 14 '19 at 20:46
  • $\begingroup$ @DukeZhou If a question gets a soft deletion (known as closed), the original poster (OP) gets a message in his inbox, that his request is too broad and primary opinion based. As a result, he will get upset and a conflict is there. Producing a conflict between SE.AI and the users who have posted something is equal to escalate an issue. It's unclear what will happen then, this makes it a risky decision. $\endgroup$ – Manuel Rodriguez Oct 16 '19 at 11:35
  • $\begingroup$ @ManuelRodriguez ic. In many cases, the unanswered questions may be one-shots, with no sign of the user, before or after, and these area probably safe to delete. But I do tread carefully when deleting questions by users who are or were active. $\endgroup$ – DukeZhou Oct 16 '19 at 20:12
  • $\begingroup$ @DukeZhou I see, a recent close vote to a question which is 10 month old was rejected. ai.stackexchange.com/posts/9317/timeline perhaps with the idea that the question doesn't need an escalation. $\endgroup$ – Manuel Rodriguez Oct 17 '19 at 8:30
  • $\begingroup$ @ManuelRodriguez I don't recall the MOCAP question specifically, but, looking at it, I likely wouldn't have used mod power to close since is seems to be a legitimate, and potentially useful, query. (Sometimes I "tip the scales" if 2 or 3 trusted users vote for closure, on a case-by-case basis.) $\endgroup$ – DukeZhou Oct 21 '19 at 18:46
  • $\begingroup$ The problem is, that before closing a question is possible, four people need to press the close button. This group based decision is standardized for all stackexchange websites. A single person can't delete questions by it's own, because it will irritate the opponent who has created it. The resulting question is not how to soft delete a certain question, but how to convince other voters in the forum to review deletion flags. The answer to the problem is to create a statistics how many questions were closed per day. This helps to predict, how deletion friendly the community is. $\endgroup$ – Manuel Rodriguez Oct 21 '19 at 19:13

According to https://stackexchange.com/sites, currently, 71% of the questions on AI SE are answered. However, similar or related websites do not have a higher percentage of answered questions. For example, on Stack Overflow, only 70% of the questions are answered. On Stats SE and Data Science SE, 65% of the questions are answered.


You must log in to answer this question.

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