64

I am aware of the reasons delete has not been restricted before, but the reasoning has begun to unravel on Stack Overflow, where a fairly large pool of users now have access to that privilege. I give you exhibit A

Close/delete thrashing

Yes, we can stop it by locking it (which I did), but locks are a poor tool here because it prevents ALL post interaction (which isn't fair to the user, who is not part of this "meta" argument). Moreover, motivated users will simply out-wait the temporary locks. I had to moderator delete another question to prevent the same problem. Neither solution is great. There's no real policy to quote to people doing this either, but it's becoming a more common problem.

I think the time has come to restrict delete and undelete for regular users (not including the OP) to once per post, just like (successful) closure. It's clear that motivated users with that privilege will continue to delete/undelete ad nauseum, and it actually diminishes the delete privilege because people who disagree will get more votes tomorrow to change it back the way they see fit. Add in Meta effect and it can prevent community consensus, leaving mods to fix it with nuclear weapons. With limited deletes on a given post it forces a detente at some point, and it's a level playing field for all.

11
  • 3
    Related, also here. I have not seen posts that have been deleted/undeleted six+ times and locked twice anywhere else on SE, maybe this should be SO-only?
    – Ollie
    Mar 10 at 15:30
  • 1
    @Ollie I suspect this happens elsewhere (many SE sites, including some older ones, are Beta, so they have a lower rep threshold on these privs) but it's not enough of a problem for them to complain... yet
    – Machavity
    Mar 10 at 15:32
  • 1
    Good point. Well, can't close-votes be re-cast after 14 days? What if that were the case for delete-votes?
    – Ollie
    Mar 10 at 15:34
  • 2
    @Ollie You can only recast close votes if they age away. Delete votes do not age away.
    – Machavity
    Mar 10 at 15:34
  • 1
    Some posts should be deleted because they're simply a poor fit. It looks like enough people here agreed on that. Why is this question so controversial in the first place? Changing the privilege system may be fixing the wrong problem.
    – Mast
    Mar 10 at 16:45
  • 3
    Machavity, the lack of aging away, in relation to Close Votes, was discussed at some length in The Tavern recently; with Cat and wisdom from Shog as to why it works that way.
    – Rob
    Mar 10 at 16:46
  • 10
    Just to make it explicit: I assume you mean limiting it to once per user per post?
    – Ryan M
    Mar 11 at 0:13
  • 3
    @RyanM Correct. Once per user per post
    – Machavity
    Mar 30 at 20:33
  • @Mast because the users undeleting don't ask for merging or reopening. It's such a fringe case that should instead be used to teach people how to properly curate content.
    – Braiam
    Apr 1 at 10:36
  • 4
    strongly related discussion at MSO: Rule proposal: one delete/undelete per post (currently featured, tl;dr: moderators try to somehow manually compensate for the absence of a feature requested here)
    – gnat
    May 26 at 7:31
  • 5
    Updated the status as this is on the devs' radar, and has been triaged into their backlog.
    – JNat StaffMod
    Jun 30 at 10:37
3

I had considered this a natural consequence of more users getting 10k tools.

but it's becoming a more common problem.

The initial tooling had a relatively smaller user base in mind.

and it actually diminishes the delete privilege because people who disagree will get more votes tomorrow

So as the Q&A repositories mature (and saturate) as more users get 10k tools the tendency will be for the incidence of such conflicts to also increase at least proportionally.

1
  • How does this either answer the question by offering a solution, or argue against it? This seems to simply suggest that situation should be accepted and the FR denied.
    – Rob
    May 27 at 0:11
-2

@Catija, one way to fairly decide how to move forward on an eight year old problem is that after a certain number (twice?) of rounds of opposing actions we:

  • Use the voter's Tag Expertise totalized, after four votes.

  • Instead of always five, in some cases fewer, votes to delete or undelete - we use the totality of tag experience for one way or the other, it needs to outweigh the opposing side, perhaps by 25%.

  • Subsequent reversals must be from a group of voters whose experience outweighs the previous group - that leaves things open/closed or deleted/undeleted, eventually, decided by those with the most expertise.

  • People of various experience can continue to vote either way (beyond 5 votes per action) until one side of the scale or the other tips (exceeds ~25%?) in favor of one outcome over another.

  • Each time the scale tips and an outcome is reached the top 5 experienced voters are retained as the users who made the decision, for the purpose of the banner and revision history.

After a certain number of reversals the ability of people to change the outcome will be reduced and the outcome will have to stand. Obviously bad tag edits not made in good faith can be flagged, and rollback wars handled by the same method.

What we end up with is the most experienced people making the decision, and people with less experience not being left out of having a say in the matter. This would reduce moderator involvement, particularly where they feel they have less experience to fairly resolve a dispute; and where locking is defeated by ardent followers and watchers.

This could also easily be applied to: Testing three-vote close and reopen on 13 network sites too.

3
  • 3
    This has 3 problems, 1º the voting history has to be complete for transparency and audit by the community (so your last bullet-point is not a good idea). 2º Some of the recent delete/undelete conflicts are by users with massive tag scores, that doesn't make them right and skews power toward a few (it actually makes the problem worse because integrity and score are not correlated in any way). 3º It introduces more business rule complexity something the company has avoided in the past when possible (and there aren't enough cases to justify more complexity). 1 vote per user is simple, democratic.
    – bad_coder
    May 26 at 18:59
  • @bad, None of those three problems occur, 1. it's available in the queue for some types and unavailable by design for others - if you feel it should be visible that's a new question. 2. It does "skew" towards those with high tag scores, so don't upvote them in the first place - with a high enough score they'll have a Hammer anyways. 3. No one is getting extra votes. --- Your own answer about how to resolve this would be more helpful.
    – Rob
    May 26 at 19:08
  • 3
    I did address those issues, tag expertise is just not a good indicator. Most questions have 1 language tag but it's the remaining smaller tags that indicate expertise, there's no way to distinguish based on that alone. Besides the abuse problem seems to be with users who are overinvested in the tag, so concentrating more power on them isn't going to solve the problem. I gave my answer, which is inline with the current proposal: 1 vote per user seems good enough.
    – bad_coder
    May 26 at 19:19
-14

I'm against this. It's such corner case that should be solved using existing tools. Either the question should be ineligible for deletion via reopening or, in the case of duplicates, merged with the canon question so all good content is centralized and not disseminated in several questions. This is the wrong solution for a problem which we already have tools to solve.

16
  • Note that I don't mention locking because that's the worse option. For me, locking is a temporary solution while better solutions are found, sadly implementation doesn't match this.
    – Braiam
    Apr 1 at 10:34
  • 3
    @Luuklag none. The ones undeleting are using the wrong tools. They should've using reopen votes or asking for merge. The ones deleting are doing things correctly: they believe there's no value in keeping the post around.
    – Braiam
    Apr 1 at 12:44
  • 2
    Well yes, this example is a rather poor exhibition of use of privileges. But using a broader view, in a more general context, I don't see the use of keeping things the way they currently are. Being able to use multiple (un)delete votes can only lead to the poor behaviour that was demonstrated in the example quoted by OP.
    – Luuklag
    Apr 1 at 12:53
  • 4
    @Braiam Reopening a deleted post doesn't undelete it, so I don't see what reopening would accomplish here. Some questions have value, but are not on-topic; the best path forward for those is to remain closed, but not deleted. Are you suggesting the community should change how they handle such questions (e.g. we should delete everything that is closed)? Also, merging is not appropriate for the vast majority of duplicates, even well-asked duplicates. Duplicate closure covers a broad case of "same question-itis", but merging only makes sense when answers on both Qs would make sense for either Q.
    – TylerH
    Apr 1 at 13:51
  • 2
    @Luuklag but I'm not talking about the example. I don't care about the example. The example is just the symptom to a bigger problem: people using the wrong tools. Why I have to repeat this again?
    – Braiam
    Apr 1 at 15:28
  • 2
    @TylerH I don't know how you came to those conclusions, because I'm saying that people undeleting should reopen or merge. Reopening makes the question ineligible for deleting, thus stopping the tug of war. Merging also stops the tug of war. If they believe that the post should stay, it should stay in the best state possible: open or. in the case of duplicates, merged. And yes, not all merges are appropriate because many duplicates aren't appropriated either. Read Dr. Strangedupe from Jeff. All duplicates should ideally be merged. (Also my post in MSO about this issue)
    – Braiam
    Apr 1 at 15:31
  • 3
    @Braiam I came to those conclusions by reading your answer. From your comment it sounds like you meant that the users casting undelete votes should cast undelete and reopen votes, not just reopen votes. Reopening will prevent delete votes, but what if the question is worth keeping around but not worth keeping open? The "best state" is not necessarily open.
    – TylerH
    Apr 1 at 17:29
  • 1
    @TylerH well, don't know you but duplicates has to be useful signpost. If it isn't a useful one, why keep it? If it's indeed a duplicate, why not merging?
    – Braiam
    Apr 1 at 19:21
  • 6
    I already covered why we don't merge duplicates most of the time in my original comment, but an additional reason is that multiple signposts are only useful as signposts if they're different, e.g. different titles or combinations of key words that a user might search for. If we merged duplicates then we wouldn't have signposts anymore, and that would open us up to even more duplicated questions being asked.
    – TylerH
    Apr 1 at 19:39
  • 1
    @TylerH what? The duplicates are so different that the answers don't apply to one another? Then they are not duplicates. Why are you making me repeat myself: not all merges are appropriate because many duplicates aren't appropriated either.
    – Braiam
    Apr 4 at 10:39
  • 4
    @Braiam Questions can be the same base question, but have different details in them. Their answers can reference those details, which would preclude them from making sense as answers to the dupe target/vice versa. They are still duplicates, but they don't make good merge candidates. I'm sure there are prior discussions explaining when to merge or not on Meta.SE or Meta.SO; search for them if you're interested in learning more.
    – TylerH
    Apr 5 at 14:06
  • 1
    @TylerH again, details either are irrelevant (2+2 vs 3+3) or they aren't (2+2 vs 3*3). In the later case, then the questions are not duplicates. Duplicates means: all answers for A are also answers for B and viceversa. That means that merging any way would achieve exactly the same set of answers. Why is that hard to grasp?
    – Braiam
    Apr 6 at 10:38
  • 3
    @Braiam Details are not irrelevant when they cause confusion in readers. Ex: "wait, why are you talking about a header here? OP is trying to center a body element? Where did this code come from" "All my code is specific to another question but the two threads were merged". Sure, we could merge these two questions that are both "how do I center some content using flexbox", because the solution to both uses the same properties, but then moderators would have to edit all the posts from one of the threads to remove stuff that no longer makes sense in their new context.
    – TylerH
    Apr 6 at 13:19
  • 1
    @Braiam And once that's done, you're likely to have answers that just copy each other. Merging all duplicates creates a lot of work, introduces a lot of harm, and doesn't actually solve a problem. That's why it isn't done except in very uncommon cases.
    – TylerH
    Apr 6 at 13:19
  • 1
    @TylerH well, for answers that are basically carbon copies of each other, we have deletion. All your arguments seems to hinge on the idea that duplicate questions can have different answers, which is basically the opposite of what duplicate means. Once you get out of that absurd framework, things actually make sense. In your current framework, they aren't even close.
    – Braiam
    Apr 6 at 13:53

You must log in to answer this question.

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