Use cases for duplicate, hold, and reopen actions

I’ve been creating use cases for the front-end team, and I want to draw the community’s attention in particular to the ones for question closure and associated activity. These use cases are informed by all the discussions on the forum and in chat and also include some new ideas, described here.

One of the overriding goals is to make this process less aggravating for all involved: the question author (who needs to know what the problem is and what to do about it), the curators (who need an easy way to moderate content and provide useful feedback), and the community (who want good questions and a minimum of thrashing). We’re aiming for a process that is fully constructive while not coming across as adversarial.

Key points:

  • Duplicates are a separate (related) workflow, not part of question closure. This means anybody can suggest duplicates (not just people with close votes) and the negative feelings that some see with “close” for duplicates don’t arise. There’s nothing wrong with duplicates. (Or, rather, the bad duplicates will get removed by the community.)

  • There are two other close reasons: “needs attention” and “off topic”. When someone (with the appropriate trust level) nominates a question for closure, we give targeted messaging to the author and other readers. The focus is on the feedback not who said it, so this feedback is anonymized.

  • While a question has pending close votes it can accept both “close” and “open” votes. Instead of going through close/reopen cycles, people can weigh in immediately, right there on the question. (MVP doesn’t have review queues but even if it did, we want direct access from the question.)

  • For both duplicate suggestions and close votes, we guide the user toward editing to address the issue, and we accept any edit from the OP as a remedy. This latter could be abused and maybe we’ll need to put some guardrails around it (maybe it’s a trust-level thing), but let’s start here. If there are cases of bad-faith edits, we have ways to deal with them separately. Remember that we are not going to be at SO scale on day 1; the smaller communities we’ll start with can moderate their communities effectively.

See the use cases for details. (The directory is named “flag” because there will be other stuff there too, including handling of spam.)


Two suggestions,

  1. For “needs attention” lets have a set of options to choose from with a possible text field, lots of people have a very distinct writing style and simply removing the name won’t help anonymize the message.

  2. For off-topic have a place to link the meta post where said type of question was declared off topic. It’s not fair to close questions without a good explanation of why its off topic and the comment section below a question really isn’t the place for the whole site to argue about whether something should be on topic. If the meta post doesn’t exist quite yet, then the community should be able to create one and then link to it in the close reason.


Nice work! I really like separating duplicates and closing.

I agree with this in theory. But it could be worth seeing this as a feature we’re testing and evaluating. Spontaneously, it feels like one of those things that works well in theory, but not in practice.

Use case “Suggest duplicate”:

I suggest one addition for the public page update. If someone starts to answer, some banner could show a text with something like

This question has been marked as a duplicate. Before you answer, please check if the question does not already have an answer there. If you feel that you have something to add to existing answers, please add them there instead of here.

But this is probably another use case to be added I guess. “Answering a question flagged as duplicate”

Use case “Respond to duplicate suggestion (as owner)”:

In the case where “If user disagrees” it feels like something is missing. At least some people, moderators and possibly also experienced users, need to be able to override this if it indeed is a duplicate. Maybe this is a new use case with the precondition: “Question have been flagged as duplicate, but OP disagrees”

“Should answerers be notified of the change in state?” - Definitely yes.

Use case “Edit a question that is on hold”

I see that only edits from TL3 users nominates the question for reopening. I can see the problem this came from on SO, which is questions ending up in a reopen review queue when just minor spelling errors have been fixed.

There are two issues with this. First, a substantial edit from a low rep user will not put the question in the queue, even if the question has been completely fixed. Secondly, a high rep user cannot fix a question partially without forcing the question into the reopening queue. This is similar to the problem on SO where the dup hammer is not optional.

I propose a different solution. Allow users who have edited to do an active choice between nominating for reopening or not. The alternatives could be something like this:

  • My edits have fixed the question so that it is answerable, on topic and follows the rules. I wish this question to be reopened.
  • My edits fixed spelling issues or other minor details or only partially fixed things that would make the question answerable. This question still needs work before reopening.

Non-mvp: Someone who repeatedly marks a question for reopening even though they’re only fixing spelling errors could get this choice removed temporarily.


This is going to blow up in our faces. If the OP’s judgment could be trusted, we wouldn’t need to close questions in the first place. Someone new to a site is not in a good position to judge what is on topic, well enough written for the site, includes the necessary information, etc.

I agree with @klutt in that the OP can optionally flag that the issues have been addressed in their opinion. This can then ping anyone who voted to close so that the close-voters can re-examine the post and decide whether it really does meet the site quality levels or not.

Here is a suggestion for how the above mechanics would work. When an OP edits a closed question, there is a check box associated with each close vote comment. They have to explicitly check the box to indicate each comment that has been addressed. When all the boxes have been checked, all the close-voters are notified.

This may still cause trouble, but it’s worth a try. On the EE site, I remember many arguments with inexperienced askers where a question was closed due to lack of a schematic. We often got snippy responses about how the circuit should have been obvious, despite the hand-waving description that could be interpreted many different ways. OPs just aren’t in a good position to be trusted to judge their own quality.


I totally agree that this will probably be necessary when the site grows. However, in the beginning I don’t think this will be an issue, because most people that will find Codidact in the beginning will be people who knows the drill.

Also, this will be a thing that will be relatively easy to fix. Changing this is not a major change that totally makes using the site fundamentally different. Furthermore, this sounds like a thing that could be affected by your trust level.


Yes, we should re-evaluate when we’ve seen it in action. We don’t have data yet, just a theory formed from observation of interactions around close votes at SE.

Good idea! I agree that’s another use case. (My first use cases were broad and multi-faceted, and people who know more about this than I do told me to stop doing that. :slight_smile: So there’ll be more smaller use cases instead, and it’s a work in progress.)

I see your point about community/mod override on duplicates, thanks.

Correct. However, people who can review edits can also vote to reopen. Instead of letting a low-trust-level editor have a (re)open vote that wouldn’t otherwise available, we should prompt the reviewer to cast that vote. If you approve an edit on a question that is closed or has pending close votes, you get a prompt asking if that edit addresses the problem – so you can say yes if it does (cast a vote) or no if it was just fixing typos and the edit shouldn’t change the state of the question.

Assuming that only new users ask questions that get closed. On the SE sites I was active on, high-rep users also had questions put on hold for not being clear or for being opinion-based or even for being off-topic. It happens. And new users are capable of responding to feedback (when actually given feedback) and fixing the problems in a question; I’ve seen it many times. New users are new, not stupid – if we tell them “please specify X and Y and clarify that thing you said about Z”, let’s give them a chance to do that, and then monitor it and see what we need to adjust.

It’s also reasonable to say that if the OP edits and that doesn’t fix the problem, and the question gains another close vote, then the presumption of effective edits is revoked – another edit from the OP is just an edit, not a reopen. Basically, the OP gets one chance for free.

Whenever somebody on SE proposes pinging voters on edits, some people complain that that’d be too noisy (discourage future votes) and others complain that it would be slow or ineffective (those voters have to go look, and others assume that they will so don’t vote themselves). We might end up there but let’s not start there.


Nice. But it would be good with an option to not put it in the queue to save others from even having to review it.

Actually, Olin’s argument does not assume that. (Well he does further down, but I’m focusing on this specific sentence, which is a pretty good argument) This argument makes no difference between high and low rep users. He just states that if the user X posts a question that gets closed, that in itself proves that X is not the right person to judge whether it should be closed or not.

Sounds like a very reasonable compromise.


Remember: no queues yet. Votes on the status of a question come from people actually visiting the question (or, with the modification we’ve just discussed, reviewing an edit to the question). But also, yes, I meant that the reviewer explicitly casts a vote or doesn’t, not that we assume that approval of a typo fix counts as a vote in favor of the question being open.

Bad wording from me. My main point is that I think that you never should lose any privileges because of gaining more reputation. (Change the words “privileges” and “reputation” for something else if you like)

Yes, I really don’t like the gold hammer on SO :wink:

Interesting idea, but I’m still worried about this causing trouble due to the majority of bad questions being written by people that don’t understand the site.

I could see this being flipped around, though. Being allowed to re-open your own question after an edit is something you earn based on your history. As before, you have to check something like this edit addresses all the close reasons, or check them off individually or whatever. If you do this a couple of times and the users agree by re-opening, then your questions get automatically re-opened based on your say-so. This privilege can come and go depending on your history. This is basically what you propose, except that the very first time the presumption is with the site, not with you.

This is similar to your question getting held for review when you first post. As you accumulate a positive question score, your questions are presumed good until voted otherwise.


What are some things that should be on this list? “Need more information” and what else?

  • Unclear title or title doesn’t match body.
  • Needs proofreading/grammar improvements.
  • Too broad.

Also if we are going to let the OP open their question by editing, then there should also be a way for higher trust level members to do the same.

It’s frustrating to do whatever trivial edit is required to make someone else’s question on topic and then have to multiple days for the question to be reopened.


I’m working on some changes to these use cases in response to this thread (which is what prompted my question). One of them is that if the OP is TL0 or TL1 the edit doesn’t automatically cancel hold/dupe suggestions and has to be reviewed.

One of my goals with this approach to hold/reopen is to reduce the cycles. Instead of getting a question closed and only then improving it and accumulating reopen votes, we push toward improvements as early as possible and collect “open” (which means either reopen or keep open, depending on state) starting as soon as there’s one hold vote.

Edits that need to be approved will now generate hints to the reviewers about casting open votes if they think the edit fixed the problem. Plus the author gets notified of issues requiring attention, which I hope leads to prompter edits from the OP before other people have to guess.

Oh, now that I’ve written the rest of that, I see what you mean here. Yes, thanks; I’ll figure out how to make that balanced.