82

I have the feeling that with the growing influx of questions, an increasing number of questions that are clear closing candidates (i.e. it is totally obvious and undisputed that they belong closed under one of the close reasons) slip through the net because they do not attract enough attention, and thus do not accumulate enough close votes, especially on weekends.

However, unless you start collecting examples manually, it is impossible to verify this because unsuccessful close votes can't be queried in the data explorer.

I think this is an important metric, and is going to become even more important in the future.

Can the data structure be changed to store these unsuccessful closing attempts in some form, so they can be queried in the data explorer?

Ideally, the following data would be queryable:

  • Number of total closevotes on a question

  • The types of the votes (not sure how this would have to be done properly - normalizing votes into a separate table sounds awfully complicated)

Note: The point of this is the ability to get stats, not necessarily for finding questions that may be closeworthy! However, Waffles points out that there are only about 10,000 questions that have more than one closevote that has withered away, so having these stats wouldn't really help much. I'm accepting his answer for that reason.

Improve this question
19
  • 6
    Important indeed, I see more and more questions stay unclosed with 3 or 4 votes. Maybe better request can be decrease the required votes to 3? Five different people to approve closing is too high in my opinion.
    – Shadow Wizard
    Commented Mar 13, 2011 at 12:11
  • @Shadow yup, exactly. Decreasing the required votes to 4 is what I would like to suggest if the numbers prove the feeling to be correct
    – Pekka
    Commented Mar 13, 2011 at 12:13
  • 1
    Another option is require X amount of reputation to close a question (or five different people) e.g. 100K reputation (and minimum two different people) so if 80K member and 30K member both vote to close, it will get closed, if 10K, 15K, 5K, 12K and 10K it's also closed as the 5 people rule is triggered. This is more "intelligent" way and give more power to the high rep members..
    – Shadow Wizard
    Commented Mar 13, 2011 at 12:17
  • 3
    @Shadow: I wouldn't like that. Having the need of multiple closers or a mod is a great way to avoid unjustly closed questions. I do like the "rule of 5", but Pekka is quite right about the danger of questions "slipping through".
    – Hendrik Vogt
    Commented Mar 13, 2011 at 12:20
  • 5
    @Shadow it would be great to have more power in totally blatant cases, but being a high-rep user doesn't protect from making the same errors in judgement as everybody else. All in all, having a group of peers cast the vote is probably the better way. Although decreasing the number to 3 if all voters involved are 10k+ or 20k+ users might be a good compromise.
    – Pekka
    Commented Mar 13, 2011 at 12:24
  • @Hendrik but don't you think that with experience comes better judgement?
    – Shadow Wizard
    Commented Mar 13, 2011 at 12:24
  • @Pekka OK, that's your request... I totally agree with reducing the required people as well. :)
    – Shadow Wizard
    Commented Mar 13, 2011 at 12:26
  • @Shadow: Well, of course. But humans make mistake, and that is independent of experience. Ah, I just see, Pekka wrote something similar.
    – Hendrik Vogt
    Commented Mar 13, 2011 at 12:30
  • Correct me if I'm wrong, but questions with close votes are already listed in the moderator tools. So shouldn't be possible to somehow query these?
    – Felix
    Commented Mar 13, 2011 at 12:41
  • @Felix nope, according to this, the votes disappear after some time if closing doesn't materialize.
    – Pekka
    Commented Mar 13, 2011 at 12:42
  • @Pekka: I see, ok then :) I didn't say anything :D ;)
    – Felix
    Commented Mar 13, 2011 at 12:43
  • 2
    @Pekka I'd love to cast a close vote with a weight based on my ranking on the question's tags. Parenting the [google-app-engine] tag and taking care of it with other few high profile users, it's depressing to see many bad questions not closed because there are not enough votes casted. I recently quit to flag bad questions in order to close them because I read that the flag feature is for other purpose and I agree with that.
    – systempuntoout
    Commented Apr 19, 2011 at 9:09
  • see also meta.stackexchange.com/questions/87979/…
    – Ian Ringrose
    Commented Apr 19, 2011 at 12:56
  • 2
    I am not against this, we just need a slight internal redesign to allow for it
    – waffles
    Commented Oct 19, 2011 at 23:36
  • 2
    @Pekka'sOrganicRepFarm - I only say it because every dang meta post you've posted since day 1, simply makes sense, its logical and it benefits the end users, and sometimes in most cases it ends up getting implemented. They need you man. You would make an ideal project manager for a software company. +1 to ya.
    – JonH
    Commented Jan 26, 2012 at 17:56

4 Answers 4

Reset to default
1
+500

I have the feeling that with the growing influx of questions, an increasing number of questions that are clear closing candidates (i.e. it is totally obvious and undisputed that they belong closed under one of the close reasons) slip through the net because they do not attract enough attention, and thus do not accumulate enough close votes, especially on weekends.

Well, close votes are only deleted IF the question has more than 100 views. The current close interface in review allows you to dig through all the history, further more these votes are all visible in Data Explorer.

I don't see what point there is in digging through the history, when it seems we already have 40,000 decisions left to be made.

If we can figure out a way of clearing out the current backlog on Stack Overflow, I would be open to including this additional information.

Traditionally, we only ship data that is visible in the UI. To get such close/reopen data you would need to scrape the site daily as a high rep user. I am not sure this fits into the current mould and am not convinced of the value this additional data would provide.

Further more, if we are allowing this, why not also send through standard deleted upvotes/downvotes/accepts. A similar argument could be made for releasing this information.

also for the record... there are less than 10k questions that have no close votes active, yet had more than 1 close vote active.

Improve this answer
7
  • 3
    If we could see the data in SEDE, patterns might become apparent that would point out questions that that could safely be closed with fewer votes -- just as SEDE queries have helped identify flag-worthy posts. Fewer votes needed, or potential new algorithms for auto-closure, could cut into that 40K backlog.
    – Awesome Poodles
    Commented Feb 3, 2012 at 3:15
  • Yeah, what Brock says. There needs to be a way to verify (or disprove) suspicions about the voting behaviour with hard data, otherwise suggestions in that direction will never be taken seriously.
    – Pekka
    Commented Feb 3, 2012 at 10:00
  • @Pekka'sOrganicRepFarm my points here are quite simple and, I think, reasonable. 1) help us figure out how to deal with the 40k backlog on SO first... it is the first problem we need to solve. 2) this opens up the door to requiring we release deleted up/down/accepted votes ... this worries me.
    – waffles
    Commented Feb 3, 2012 at 10:29
  • 1
    also for the record ... there are less than 10k questions that have no close votes active, yet had more than 1 close vote active.
    – waffles
    Commented Feb 3, 2012 at 10:31
  • @waffles I don't really see how the two are connected. I was thinking about suggesting lowering the number of required close votes from 5 to 4. If applied retroactively, that would help fix the backlog. But without any historical data, how am I to back up the suggestion?
    – Pekka
    Commented Feb 3, 2012 at 10:32
  • @waffles ah, okay, the point you just made is very convincing - getting stats for this wouldn't help much. That's kind of the information I was looking for. Have a delicious bounty. :) I'll edit your answer
    – Pekka
    Commented Feb 3, 2012 at 10:50
  • @waffles interested in answering this one with the data you have?
    – Pekka
    Commented Feb 3, 2012 at 22:40
10
+250

I don't have an answer to the stats suggestion, but I do have a suggestion for a simple new tool which would help solve the same underlying problem it seems you're looking to address, without making any extra data publicly available:

One possible work around to help avoid votes failing at 4 would be to have a tool (possibly in /review or the 10K tools) that lists "close/reopen votes which are about to expire". The list could be sorted by urgency, as opposed to the existing tools which sort by recency of vote or total number of votes.

This would help catch close (and reopen) votes that fail because of lack of traffic. It would be interesting to see these stats, particularly for the reopen votes since I suspect the 100 views requirement will usually have been hit before the vote to reopen has even begun.

Improve this answer
2
  • 2
    I have a soft spot for the 100 views thing making it impossible to close very old questions that already attracted a ton of views. open to changing it so we require 100 views after the vote was cast ... if it sounds good ... post a meta on it
    – waffles
    Commented Feb 3, 2012 at 1:37
  • @waffles - posted: meta.stackexchange.com/questions/121171/…
    – Flexo
    Commented Feb 3, 2012 at 6:26
4

In the meantime, this SEDE query proved useful to me. It finds 'possible duplicate' comments attached to unclosed questions. Of course you get false positives, but there are many expired suggestions that look valid. Now we have the /review pages, it would be great to get these in to the workflow.

(If anyone fancies checking some of the ones in the tag it would be great, as I re-flagged them six months ago, I can't again.)

Improve this answer
2

I understand where you're coming from. It seems to me that questions that should be closed will either get closed eventually or will not be noticed. But, if nobody notices them, then is it really such a problem as to worry about closing them? I don't know of any way to query for close votes, but if you really want to find bad posts to close, you can go to: https://stackoverflow.com/review

Improve this answer
3
  • 9
    This is more about statistics. If it can be proven that thousands of questions "get away" because they don't accumulate the necessary number of close votes, suggestions of changes to the voting system carry more weight.
    – Pekka
    Commented Mar 20, 2011 at 21:03
  • @Pekka'sOrganicRepFarm 40,000 questions are in limbo now, we already have enough decisions to make, if anything we need to find a way to age close votes more rapidly.
    – waffles
    Commented Feb 3, 2012 at 1:26
  • 3
    @waffles I'm not sure whether we really understand each other here. :) As said below, actual decisions are not my point at all. I want to be able to find patterns, not primarily questions that may need closing... So for example, if there are tens of thousands of questions with two or three "too localized" votes, that might support an argument for lowering the number of close votes required from five to four. It's that kind of stats that I'm looking for
    – Pekka
    Commented Feb 3, 2012 at 9:59

You must log in to answer this question.

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