Place questions with X net downvotes on hold automatically

Question

To resolve the frequent issue of poor question unfairly gaining answers, and thus paving the way for more poor questions to be asked

I Propose

EDIT: From feedback, have changed net vote threshold to 5 (to at least match that of the current 5 users at 3k rep) END EDIT

If a question has 5 net downvotes, the question is put on hold automatically, and no answers are allowed until the question is resolved/has been edited and net votes are > -5.

The net vote threshold could be perhaps different, to avoid grey area questions being on hold unfairly, but this also needs to be high enough to allow for early action.

On hold note, something such as:

This question has been put on hold via community voting on the question, and cannot receive answers. To resolve this, edit and improve the question.

This proposal would assist in stopping poor questions from receiving answers, as currently questioners not wanting to put in any effort know they will likely get an answer eventually.

This could also help improve site quality by making people improve their questions before they can have answers, so answers are only given to decent questions.

Answer 1

This is an interesting proposal... In particular, it's interesting because the stated goal is not to hide lackluster questions (which has been proposed before, and which is partially supported already), but rather to prevent them from being answered - in other words, it suggests the use of closing for a purpose that closing is actually well-suited to perform!

Others have raised some well-founded concerns with how this might play out, and I have my own thoughts that I'll get to in a moment... But first, let's do some simple data-analysis to see if this even has a slim hope of making a difference.

The nature of questions that get downvoted to a score of -5 today

I'm using internal data so as to include deleted questions and accurate vote timing, but the queries are simple enough and most can be run on the public data if you wish to see how they stack up on sites other than Stack Overflow. I'm using Stack Overflow because there are relatively few posts that score <= -5 anywhere else - Programmers has only 2809 questions meeting this criteria, an order of magnitude fewer than the number of closed questions - in practice, it's unlikely this would have much of an effect outside of SO.

select count(*) "Total questions scoring <= -5",
  count(case when AnswerCount > 0 then 1 else null end) "with answers",
  count(ClosedDate) "currently closed",
  count(DeletionDate) "currently deleted"
  
from Posts q
where PostTypeId=1
and Score <= -5

...gives us:

Total questions scoring <= -5 with answers currently closed currently deleted 
----------------------------- ------------ ---------------- ----------------- 
124441                        49035        104940           101272            

Ok, that's a lot of bad questions, and a not-too-shabby portion of them with answers that could in theory have been blocked. How many of them would have been blocked?

select count(*) "with answer posted before -5 threshold reached"
from Posts q
where PostTypeId=1
and Score <= -5
and 
 (select min(CreationDate) from Posts where ParentId=q.Id) < 
    (select top 1 CreationDate 
      -- inaccurate for questions that recieved a mix of votes or had funky retractions
      from (select top 5 CreationDate from Posts2Votes 
        where VoteTypeId=3 and PostId=q.Id and DeletionDate is null
        order by CreationDate asc) x
      order by CreationDate desc)

This isn't intended to give us a perfectly accurate answer to the previous question - just a quick estimate, gained by comparing the timestamp of the first answer with that of the 5th downvote.

with answer posted before -5 threshold reached 
---------------------------------------------- 
43275                                          

Ouch. A whopping 88% of the answered, heavily-downvoted questions already had an answer by the time that 5th downvote arrived. It appears we may be running into the same problem here as we did with close voting: it takes too long to get 5 people to spend the time to vote on questions, even (or perhaps especially) painfully-bad ones.

Not seeing a lot of potential here, I'm afraid.

Other problems

Now, you can argue (and have argued) that folks might be more willing to downvote if they had a bit more of a goal to work toward... But frankly, I'm skeptical: most folks don't seem to want to go out of their way to find and rate crap; we had to put an awful lot of work into the close review system to even get close to making that work - what's next, "downvote review"? We're essentially just re-building the vote-to-close system with a lower privilege threshold and lighter-weight criteria for reopening... Which brings me to my primary concern here:

The amount of work required to close bad questions is entirely too high

Not surprisingly, this is the same problem I see with the existing close system - it just wasn't designed with the current volume of questions in mind. Getting 5 people to collaborate on anything is a small victory; getting it to happen repeatedly for thousands of questions a day is pushing your luck. First, folks get careless... Then they just give up. After all, this isn't their trash - why should they log time every day to pick it up when the folks dropping it don't care?

Here's the kicker: it's a lot of work for purely arbitrary reasons. We make up these rules; it could be 5 votes or 15 votes or... 1 vote. Reopening could require an equivalent amount of votes, or... no votes at all - why not just reopen when someone submits an edit? If we want more people voting to close, we could just lower the rep threshold for the privilege and be done with it, no need to hook up downvote and close... But you do have to decide,

Who do you want deciding what is permissible to answer?

Remember, the original idea behind linking privileges to reputation is to capture a certain measure of trust and experience based on actions. There are probably lots of folks who know what a good question looks like - or a bad one - without ever having seen a single post here before... But there are also plenty who don't, or won't. So the system is set up to require you to prove it: if you want permission to decide what others can or cannot answer, you first must contribute a bunch of well-received posts of your own. Once you've become a "co-author" of the collaborative work that is a Stack Exchange site, you get to participate on the editorial board...

It can be awfully hard to see it when you're up to your eyebrows in dung, but there's a very real advantage to not letting everyone who happens to spend a few minutes on the site make decisions that require input from multiple other people to reverse. One of the more depressing things I observed while evaluating the potential for single-vote closing was the number of reopen votes from high rep users that just... age away. When the folks who've helped to build a site aren't empowered to protect it from upstarts who don't respect the same goals, it disrespects the work they've done and diminishes their ownership.

An alternative: out of sight, out of mind...

Let's go back to the original problem statement here:

the frequent issue of poor question unfairly gaining answers, and thus paving the way for more poor questions to be asked

Now, why is this a problem? Here are a few arguments...

• Poor, answered questions set an example that poor questions get answered. There's some amount of truth to this, but I suspect less because of the "broken window effect" and more because of how it selects for answerers: folks who insist on good questions are starved, while those willing to answer anything are fed. Faster closing doesn't help much here, even if that were possible - we'd just be starving more people.

• There's less motivation to "fix" answered questions. Why learn to ask better questions if less work produces the same results? If we're not educating the people who ask questions here, we're doing them a disservice.

• Fixing answered questions may invalidate existing answers. This is a particularly frustrating catch-22: the question must be fixed or deleted. Deleting also removes the answers; fixing makes the answers invalid. There's no solution that doesn't make someone unhappy.

• Answered questions are harder to delete, and waste the efforts of the answerers when they are deleted. Even when poor-quality questions eventually are closed and deleted (and as you can see above, tens of thousands are) if they've been answered then the work of those answering is wasted. This is contrary to the goals upon which these sites were founded.

The solution here isn't to block folks wishing to answer questions - surely that's why we're all here in the first place! Rather, we need to do a better job of making sure the questions they answer are more suitable for answering. I believe the proper solution here is a combination of changes:

1. Make low-quality questions less visible immediately. Don't wait until they get downvoted - figure out if they're problematic and immediately tuck them out of sight until/unless they're fixed. Not just on the homepage, but on tag pages as well.

2. Encourage both askers and answerers to fix questions they care about. We're taking baby steps in this direction with some new badges, but we could be doing so much more: how about a message shown to folks answering these questions warning them that their efforts may be wasted if they don't take a minute to also fix the question itself? And, combined with #1, a message shown to folks revisiting questions that are languishing, encouraging the same?

3. Edit or die should really be our mantra here; many low-quality questions could be fixed, but many won't be - and if no one cares to fix them, even if warned (#2), they need to be quietly removed as quickly as possible. This sets a better example for everyone using the site, while removing the noise that is increasingly a problem for those looking for a solution.

We're working on specific designs for all of these changes as part of the ongoing se-quality-project.

Answer 2

I was originally going to write a feature request for this myself, based on a discussion a few moderators and I had. While this might be controversial, I think it's an idea worth exploring for a site like Stack Overflow.

We all understand that the largest problem facing Stack Overflow is the increasing volume of low quality questions flooding the site. Close votes are an attempt to stop bad or off-topic questions from being answered, but in order for something to be closed, it has to fit within one of the stated reasons. Close votes can take a while to process (as evidenced by the close vote queue), letting bad questions linger on the site. They are unable to scale to the size of the problem.

If we were to do this, I'd suggest applying the automatic placement on hold to occur at a net score of -5 (to at least match the number of close votes we'd otherwise require). I say a net score of -5, because you don't want highly upvoted questions that happen to accumulate five downvotes to be closed in this manner.

A new close reason of "very low quality" could be used for these questions. Such a close reason should be applied by the Community user (to not expose the voter identities) and would include a clear statement that the question did not meet the quality standards of the community. It could have a nice link back to the appropriate section of the Help Center on how to ask a question here.

Substantial edits to improve these questions would still place them back in the Reopen Votes review queue. Perhaps we could present a notification to voters once such edits were made to give them a chance to change their downvote or vote to reopen. This would allow the people who actually make an effort the ability to recover from an initially poorly asked question.

Now, I have my doubts as to whether providing answers to bad questions actually causes more to be asked, but this would effectively stop arguments about that. On Stack Overflow, bad questions in popular tags can accumulate five net downvotes in minutes.

This might allow us to focus our efforts on closing down truly bad questions, make this process more effective, and allow it to scale in a way that close votes cannot. When coupled with the refinements to the question ban system currently being implemented, this might help deal with really bad questions and bad askers at the source.

Answer 3

My initial instinct upon reading this is that downvotes and close votes are orthogonal. Close votes typically indicate that a question is either off-topic or unanswerable… whereas downvotes instead or as well indicate that a question is poorly researched, formatted and/or phrased.

But on further thought I see no reason to add a closure reason that is, essentially, "your question sucks and does not constitute a net positive to this site" — your proposal adds this and automates it. Disabling help vampires and whatnot is a wonderful tangential benefit.

I like it.

I'd incorporate Care Bear's suggestion of allowing questions that have been closed in such a manner to be re-opened by a single vote from a 3K user. I think this is the best balance between providing a useful safety net against abuse of this power (though I honestly think that such abuse would be very rare; I see no indication that users are presently abusing downvotes to any sort of substantial extent) and over-complicating things.

Answer 4

No. This would give too much power to users with way too little reputation.

Currently it needs 5 users with 3K rep each to put a question on hold. With what you suggest, the threshold will drop to X users with as little as 125 points to do the same.

Only users who have had their share on the site should be able to close question.

However, I would embrace adding a confirmation dialog of some sort when answering a heavily downvoted question like "This question appears to be off topic or low quality, sure you want to answer?", same way like we get to confirm second answer on same question.

Answer 5

The problem I have with this is the alarmingly lower barrier for question to be closed.

• You currently need 5 users with 3k or more to close a question.
• You propose that 3 or 4 users with 125 or more will be able to implicitly close a question by downvote.

I propose a different thing: Every 2 points net below 0 score will count as one close vote from Community♦.

So for example, if a question has -5 and two close votes, if I pile on my close vote, the question will be instantly closed with: Put on hold by UserA, UserB, Second Rikudo, Community♦.

That means that without the intervention of higher reputation users, you'd need -10 to close a question with the power of just downvotes, which sounds more reasonable.

Answer 6

This would not make a difference. There are fewer than 7000 non-closed questions with the score at most -5 on Stack Overflow. When a question gets this much attention and this kind of negative reaction, it is much more likely to be closed than to stay open.

I think that this proposal by Shog9 is a better idea toward the same end: use the input from downvotes to delete answered questions, without bothering to close them.

Answer 7

Place questions with X net downvotes on hold automatically?

In the past I have proposed in chat to lower the threshhold for closing negatively scored questions.

This proposal may be the extreme of what I suggest, or perhaps taking it to a logical conclusion.

I think it would be consistent with the current moderation strategy of the sites to lower the close vote threshold by one close vote for perhaps every 2 net downvotes. I would still require a minimum of 2 votes to close/put on hold (which would kick in at a net score of -6).

Such a table would look like this:

Score     Close Votes required to Close/Put on Hold
  0       5
 -1       5
 -2       4
 -3       4
 -4       3
 -5       3
 -6       2
 -7       2
 -8       2
 -9       2
...

and so on.

This would put poor new questions on hold more quickly and give newer users a minor voice in putting bad questions on hold, while ensuring quality review.

Newer users are an untapped resource in this area. I'd like to put their ability to contribute to greater use.

Applied to reopening

I'd like to apply the same principle to reopening:

Score     Reopen Votes required to Reopen
  0       5
  1       5
  2       4
  3       4
  4       3
  5       3
  6       2
  7       2
  8       2
  9       2
...

This maintains symmetry in closing and opening, which is an important strategic goal for StackOverflow and sites. In my conversations with StackOverflow stakeholders, they have made it clear that this is important to them.

If we make it easier to fix and reopen closed questions, that's a small concession to more quickly shutdown the bad questions.

Note that this is still a higher threshold than gold tag badge holders who can close and reopen duplicate questions with a single vote.

Locked posts

There should be no impact to locked posts, as you cannot vote to reopen them.

Answer 8

Just wanted to collate and touch on some of the excellent community feedback, and my own further thoughts as a result, to make this a more well thought out idea.

---

TL/DR

For those not wanting to read, I think we know what is mostly needed and why, so this TL/DR addresses the most important issues raised so far.

The two most important issues raised seem to be:

1. Changing the current question voting system to include something else that would seriously alter why people vote (upvotes to stop question close etc)
2. Drastically reduced rep threshold for closing questions (from 3k to 125)

To resolve both of these issues (in 1, I'm that good):

1. Any downvote made on the question (using the current vote system, unchanged) from users with X rep will count towards a total close vote threshold

X can be something everyone feels is sane.
While it alters the current voting system to include something else, I don't think it seriously alters it (not with 1k rep needed for it to count).

This is simple enough for users to understand what/how happens.

If we say 1k rep required for the vote to count, and the total vote count to close is 4 net downvotes, I'm sure with the influx of additional votes this will bring, and requiring 1k rep, we'll be somewhere near..?

Sure more people will vote up/down knowing the affect, but 1k rep is a sane level I think. Plus, users with 1k rep are not likely to unfairly/wrongly manage to get a net of -4 downvotes, not considering there will also be upvotes if the other votes were unfair.

These thresholds can be debated, but the basic idea seems sound.
It's about getting that sweet spot whereby more votes are given to make close quicker, but not unfairly (hence 1k and -4 total net).

Note, this idea does not add downvotes to any flag downvotes, as this is still a new close reason separate to current ones in flag - i.e. "Your question needs work before it can be answered".

END TL/DR

---

Purpose of this proposal

1. Close poor questions more quickly than the current system (flags)
2. Stop poor questions undeservedly receiving answers
3. Make it effortless and more simple for users to close the most common poor questions

Expanded:
1 & 2 - Will work together, poor questions closed quickly wont receive answers, questions more likely to be improved, or better next time (not essential, closed and no answer is key).

3 - Closing being as effortless and simple as possible will hopefully bring about more votes, which will all also aid in the speed of closure.

What it's not for

1. Not to replace or add to the current specific close reasons
2. Not to be (directly) added to any part of the current flag system
3. Not to drastically lower the required privilege level to close questions (original proposal did, has been addressed here)
4. Not to stop arguably ok questions from receiving answers

Expanded
1 & 2 - This feature shouldn't be another flag option, that would defeat the strive for faster question close. Also, flags are confusing to some users (which one to choose), however are more time consuming to all users than a downvote. Flag = 4 or 5 clicks, downvote = 1.

3 - The poor questions this proposal is to close doesn't really need 3k rep. Dupes, off topic, etc, do require 3k rep and will still be handled with flags.
Perhaps slightly lowering required rep to close for this feature only (1k?) will provide more votes, thus closing questions quicker than current flag, but without lowering the required rep to a worrying level (again, this is not 1k rep allowed to close in the current flags).

This lower requirement can be done by anyone with X rep downvoting on the question and will count towards the close. Say -4 net downvotes by users with 1k rep or above.

Users with 3k rep are not necessarily better at making close decisions as 1k-2k rep users, certainly not in judging a simple poor question to downvote, which we entrust with 125 rep (downvote, not close).

4 - Questions which are arguably ok will likely not receive enough net downvotes to close it, due to up votes and not everyone downvoting grey area questions.
This proposal is more for closing the utter crap questions with no code, no effort, etc, than dupe, off topic, or others which require more site knowledge.
Users with 1k rep are not likely to unfairly/wrongly manage to get a net of -4 downvotes, not considering there will be upvotes if the other votes were unfair.

Technical

If we say total of -4 net downvotes by users with 1k rep or above closes the question, with message (eg) "Your question is closed as it requires work, code, or something else before it can be answered [link to help centre how to ask etc]".

Re-open is also by 1k rep users (why not?) simply voting above the threshold of -4 net downvotes.

I suggest no time time delay between total net downvotes getting to -4 and question being closed.
However a time delay from total net downvotes getting from -4 to above and question being re-opened. 10 mins?

Perhaps time delay is lowered for every increase in net votes from -4.
So at -4 question closed. 1 upvote from 1k rep user re-open in 10 mins, 1 additional upvote from 1k rep user re-open in 5 mins (or time delay as it is if time already below 5 mins, etc).
Something along those lines.

Perhaps the downvotes for the new close reason should be added into the queue somehow. I don't know enough about how the queue works to add input here.
I'm sure there are many other technical considerations (re-open from other flags - when question edited, etc)

And what happens if question is closed from this proposal, and there are flags for other close reasons? I'm sure that will be handled somehow similar to currently with different flags to close. A decision as to which one best fits etc.

I'll leave all that to the more experienced and site devs.

---

I think these ideas/thresholds will work well, but as always, chew thee fat...