How many votes should it take to close/reopen?

Do users therefore need to vote to “keep open” all questions? Sounds like unnecessary effort. Maybe “open” votes should only appear after the first close vote is cast.

What happens when “the vote is ended”? I don’t understand.

I think it means something like:

  • if a question is open, it is closed when nr_close_votes >= max(3,2*nr_open_votes)
  • if a question is closed, it is opened when nr_open_votes >= max(3,2*nr_close_votes)

This implies the order of close/open votes affects the outcome, e.g.:

  • 3 open votes followed by 3 closed votes => open,
  • 3 closed votes followed by 3 open votes => closed.

So a 3-close-vote 3-open-vote question could be either open or closed.

I’m confused about “3 more needed in either direction”. What state is a question in if it’s neither “open” nor “closed”?

1 Like

It genuinely concerns me that we’re going to this much effort for something that should be decided per community.

Closing/Re-opening: Requirement of 3, increases by 1 every time a question is reopened.

Bold numbers may be edited per community.

If there’s a controversial question, it will be consistently reopened and closed until a moderator steps in, which is exactly the purpose of a moderator.

2 Likes

I’d argue that it might not be the best start if we had the MVP, with the sole exception of not knowing the number of close votes to close a question on the first community we’re starting with :stuck_out_tongue:

I agree it should be decided per community, but this is about the MVP, so we’re not thinking that far ahead here. Also, what about more trusted users? If a user has a higher trust level, is their close vote worth more? Or do they have some extra privilege when it comes to these close/re-open wars? Perhaps we should require mod/meta intervention once a close vote is cast on a question that’s already been re-open and only once the mods have decided/there’s a meta consensus about the question should the question be closed? Or the other way round, where it gets closed until there’s a consensus about re-opening it?

Probably not MVP, but if we do implement some sort of vote-lock-until-broader-decision, I’d expect leaving it closed would be correct. A question that is closed is safe, more or less by definition.

2 Likes

Yes, I agree with you, and that’s how it was meant:
A user makes the statement “this is a duplicate of XYZ”, and then the community decides if that’s true. The difference is just that if 97 users think it’s a valid question, it’s not gonna be closed as invalid by 3.

1 Like

The only time a question should be closed permanently is with moderator (TL5) intervention.
It would surprise me if there wasn’t a post made on meta about whether the question was on-topic if something was as contentious as this hypothetical situation.

Yes, because those 97 will vote to re-open.


Can we come to consensus on this? I feel the discussion is far too large for something so simple.

You have it the wrong way around: The size of the discussion shows that this is not something simple.
It only seems simple if you don’t go into details, but that applies even to aerospace engineering (“rocket science”).

Even you yourself are still making arguments for one of the possible options.
In combination with that ending of your post it makes it seem like “let’s just go with my way already”, but I presume that wasn’t your intention.

1 Like

I was hoping your message would spark at least slightly more discussion or agreement from more people on a proposed default state.

I think we need to present how each of us thinks of the process in more detail than just “N votes sounds good”, because it looks like we have a number of different usage scenarios in mind.

The way I see the system used is so:

  • 3 users can vote to close a question, and these users cannot vote to close it for the next 2 weeks;
  • 3 users (could be the same first 3) can vote to reopen it, and they cannot vote to reopen it for the next 2 weeks;

The closed question would be a low quality first time question of a new user. The question then has a greater chance to be improved (by OP and editors) with help from the close plaque text and comments (speaking in SE terms, but I would rather see a better approach to this altogether on Codidact).

3 users who see it was edited and improved would cast reopen votes and let the question be answered again. No need to wait for additional 2 reopen votes from two more users.

In a good scenario, we benefit from a quicker cycle of close-fix-reopen, and users are happier than with 5 votes required where there is a good chance it will not find five pairs of eyes to reopen it. This also reduces the review queue time for reviewers.

In a controversial scenario, each time the post gets closed and reopened, the same voters who already cast a close or a reopen vote cannot cast it again for the next 2 weeks, so increasing the number of required votes is at this point unnecessary, IMO. But this is an even rarer situation than the above example, so I don’t think we should focus on this too much and make the more common process worse because of it.


But in general I don’t think the act of disallowing the question askers to receive any answers until some 3 users decide they’re “worthy” is the way to go.

I think the signal we could send to the OP is that due to the fact that 3 users looked at a post and think it’s not worth the time of experts, this post would not be seen by them unless they choose so by opting into seeing low quality posts. Just this is probably enough.

“Experts” here would be just a spooky word to make newbies think that they’re not getting the best chance they could, so it’s worth it to improve their question following the provided suggestions and guidelines (via some info box, like a close plaque on SE). In reality though, there is no such distinction between users as “experts” and “non-experts”, it’s just an option so see “closed” (deemed low quality by 3 users) questions or to hide them until 3 users cast “not low quality anymore” votes.

If we do it like this, we could even allow each user to pick how many “low quality” or “not low quality” votes they want to choose as their threshold for hiding bad questions.

Possibly even allow users to choose which specific reasons for “badness” they don’t want to see, like homework questions, questions with no code, no references, alleged duplicates, unclear or too specific.

(I suppose this could be worth a separate discussion, if so, maybe it’s worth moving/copying this post to another thread)

Decide the asker is worthy, or decide the question is worthy?

If your interpretation is the former (on either SE or Codidact), then that is a misunderstanding that is likely fairly central to our difference of opinion. It is only relatively recently that SE leadership has made much movement to a view of judging users rather than content; even now there’s still a lot of momentum behind the original content-focused moderation vision. But an idea of moderation that focuses primarily on judging askers or answerers as worthy is fatal to a real idea of quality.

Personal status and reputation should always be a second-class citizen, below correctness/usefulness of individual posts. This is counter-intuitive, usually deeply so, to almost all of the world, because most of the world is unfortunately tied up in games of status, rather than in improving each other’s lives, or even really their own.

(It should also be pointed out that there’s very little support here for starting questions as closed and requiring a full reopen cycle before they can be answered. So it’s a bit of a non-sequitur to jump immediately to “When a question is asked, N users are needed to ritually affirm that the asker has abased themselves properly and may now receive the boon of an answer”. @Olin suggested requiring a single upvote [which would be available at a lower trust level and be freer to cast] before allowing new askers’ questions, which is about the limit of what I think we’d seriously consider.)

The fundamental problem with this is that it enables a codependent end-run around the site’s quality standards by letting askers get low-quality, one-off answers from “experts” that don’t care about anything larger than a few quick upvotes. So things like security or maintainability concerns (on the computer sites), long-term or subtle relationship problems (on the personal sites), or rarely relevant but potentially devastating legal vulnerabilities will simply be ignored. But those are precisely where the difference between an expert and a mere competent speed-reader of docs is seen: detailed experience and understanding of entirely non-obvious considerations.

Answerers of such questions will resent any attempts to clean up the quality or point out problems, and as askers learn that they have a decent chance of getting an answer that seems sort of workable to them even without following any of the rules, more and more of the userbase will be drawn into opting into the low-quality side.

If the only thing closure does is forbid high-quality answers but allow low-quality ones, that is not merely useless, it is actively counter-productive, and in the most central site mechanism at that.

(An incidental problem is that it relies on telling new users some rather transparent lies about our userbase and practices. This is ethically questionable, and not especially pragmatic, either, since it’s so easy to figure out what’s really going on.)

2 Likes

Always judge a post on its own merits. Maybe it’s useful to mention this at times.

What I was talking about is how I see some, for lack of a better term, “privileged” users speak about low quality posts and how it’s “not worth their attention”, and I can see how that can be an unspoken truth among a large percentage of the regular SO contributors.

I don’t know what thresholds would be attractive to such users, as I have not seen such discussions yet. Maybe it makes sense to have them for larger sites (planning to reach the scale and daily throughput of SO), but I don’t think it’s going to be necessary for Codidact at the beginning for a while, just something to keep in mind. I just thought I’d put the general approach to an alleged problem out there for others to consider. Or maybe it’s just my misinterpretation :p

I honestly don’t have a very clear understanding of what closing does on SE and SO (I’m thinking the effects must be different between the biggest and smallest sites due to scale differences), so maybe what I proposed would be universally bad in every scenario. In that case, I’d like to learn a bit more about why and how it’s not a good way to solve the problems that disallowing answers is supposed to solve.

2 Likes

There’s an awful lot of back-and-forth here on a question that is nominally about setting a default value. Maybe it means our ideas about close/reopen are wanting in a more fundamental way.

Thought experiment: what would happen if a user with the appropriate trust level could vote “close” or “open” on any question upon seeing it? That is, you don’t have to wait for it to be closed to say “there is nothing wrong with this question”; you can weigh in when you see the question in your ordinary use of the site.

“Enough” close votes could generate a notification to people who’ve weighed in, and maybe even automatically create the meta post for the discussion (which makes the event visible to others who are invested enough to pay attention to meta). We’d need to think more about the criteria for actually closing the question with this voting scheme. And we’d probably still want “single-user close” (or -reopen) to be available at least for moderators and possibly for tag experts (when we work out the latter).

By the way, those passive votes would count as reviews, for purposes of gaining trust levels.

2 Likes

In the short term, the latter, but in the long term we need to do both.

The immediate issue is whether to close a question or not. However, if a user is found to consistently post bad questions, then some sort of action needs to be taken against that user. SE addressed this with a question ban.

A question ban is a heavy and blunt instrument that is slow to respond, and doesn’t deal with individual cases. This is why I proposed a simpler more automatic mechanism, as you noted:

For the very first question posted by a new user only. It would take a positive score for a question to be answerable, and the first question from new users starts at 0, hence any single upvote can release it to be answered.

For subsequent questions, the initial score is based on the recent average of questions. This means for “good” users, they never get a question held up again. However, for “bad” users, this ends up being a sortof question-ban-lite. You aren’t banned with a sledge hammer, but your questions have to be reviewed and found to be acceptable before they are allowed to pass. Any one question can still be released by getting up-voted. This even helps with the status of subsequent questions.

Well, as I see it, closure is intended to do two primary things together, along with some secondary ones:

  1. It incentivizes the asker, and also some potential or actual answerers, to work on improving the question to meet the standards so it can be reopened. That is, they are now working for the site’s quality, rather than, potentially, against it.
  2. It contains the immediate mess so that there are no additional answers to have to delete for low quality using the various interlocking mechanisms for that. (The actual effect of this varies, but in some cases it can really be quite obnoxious to have to clean up answers that are cheap upvote-bait but don’t belong on the site. This was most obvious on SO in the early days, and building a coherent site culture that frowns on these in general would tend to mitigate these worst-case answers quite a lot, at least for sites of less than around, say, 50 q/day.)
  • It publicizes and explains the quality standards for up-and-coming site regulars, and even for occasional visitors, so they have a better idea how to ask and also how to answer, and as their understanding deepens, how to moderate answers as well.
  • (On SE, at least) it serves as a helpful signal to feed into the q-ban.

A site at SO scale, or probably even Mathematics scale, can sustain quite a few answerers that are basically parasites: they aren’t invested in site quality, and will simply post whatever’s easy and gets upvotes, including spraying copy-pastes across a dozen similar posts, suggesting very general courses of action to take care of ambiguous error messages without trying to figure out the root cause, dumping bare links to software, and so forth, including some habits that closure doesn’t handle well (for which protection is sometimes useful instead). Even on ELL, with a measly 42/day, I was aware of at least half a dozen answerers that focused on churning out answers without much regard for real quality, and their efforts were somewhat restrained by managing to close questions in time, and would probably have been discouraged much more had we been faster at that.

Changing closure to allow answerers to ignore it if they like would tend to increase this population and make them more visible, so even on considerably smaller sites they would become a noticeable problem.

That’s true. I tried to leave that open without focusing on it much; the priority is to accurately judge post quality, and then, as needed, draw loose inferences about poster expertise or sloppiness based on that. (This includes giving experts various forms of recognition: it’s important, but secondary.)

1 Like