1
$\begingroup$

Delete/undelete reversal cycles can be prolonged (and the ultimate outcome altered) by double voting, in which users undo the defeat of their earlier votes by casting the same vote a second time.

For example, on this famous question that at one time generated all the Reversal badges, the recent cycle of deletion and undeletion was a repetition of the same tug-of-war that had previously been settled with an undeletion, after one double-voter helped push the question back toward deletion, and two voters he had reversed replied with undelete votes. There was no substantial change to the question, only a minor edit to change a tag:

https://math.stackexchange.com/posts/14879/revisions

The purpose of this meta question is to determine whether there exists support for the idea of

not casting the same vote twice (in the same direction, consecutively) on a question

Alternation such as delete/undelete/delete would not be a problem, in the sense that it does not increase an individual's total voting power, and also because people sometimes do reverse their opinion of a question after edits.

To be very clear, the question is whether there is a general preference (or a "social norm", whatever that could mean here) that double-voting be voluntarily avoided, although SE has made it possible, or whether it is viewed as (e.g.) a beneficial capability. It is not proposed that there be any kind of "policy" or enforcement mechanism to try to prevent or punish such actions, which are individual voting decisions that happen to be permitted by the software.

Notes:

  • It is currently possible to re-vote accidentally, by not remembering an earlier vote, and it might be justifiable to re-vote in order to reverse the effect of another user's double vote, but those would be practical exceptions that do not contradict the principle of not deliberately initiating a double-vote.

  • the point pertains primarily to questions that essentially retain their identity, if they are edited. For questions that bear no relation to the previous versions after an edit or receive drastic changes (such as defacement, spam, or asking a completely new question in the same posting) the preferred response would include a rolling back of the edit, so it is still not clear in that situation whether a repeat vote would ever be in order.

  • Correction of earlier version of question (discussing edits and open/close): for delete/undelete, edits play no role, SE does not check for double votes. For close/reopen votes there is a hard limit of one close and one re-open vote per user, per question, so there is no possibility to multiple vote.

$\endgroup$
  • $\begingroup$ Just to clarify, if all other users on the site are indifferent, then under this suggestion 3 users who believe a question should not be deleted would trump 5 users who believe that it should? $\endgroup$ – Scott H. Jan 8 '14 at 20:28
  • $\begingroup$ When double-voting does not take place, it is not possible for a smaller set of users to reverse a larger one, so I don't understand what hypothetical situation you have in mind where 3 overrule 5. $\endgroup$ – zyx Jan 8 '14 at 21:20
  • 1
    $\begingroup$ I think I don't really have a handle on how deletion voting works, but here is how this plays out in my brain: 5 think a question should be deleted, 3 think not, and nobody else cares enough to vote. The delete users start voting and after 3 votes are cast the question goes. The undelete users start voting and after 3 votes are cast the question is undeleted. The 2 remaining delete users cast their votes but without double voting they cannot muster the required 3 votes to delete. Thus, due to the voting order, the 3 undelete users trump the 5 delete users. $\endgroup$ – Scott H. Jan 9 '14 at 7:16
  • $\begingroup$ Users do not vote on the eventual fate of a posting. That poll is never offered, votes to Leave Same are discarded from the Close/Reopen review queues, and there is no way to vote for the status quo. So the design is fairly clearly not meant as a poll of user opinion. Instead, users vote only to change the current state of a posting. A smaller group of voters never has greater power to alter the state of a posting, than a larger group, and all state changes by the small group can be reversed by the large group. That is the sense in which no trumping can occur. @ScottH. $\endgroup$ – zyx Jan 9 '14 at 16:50
6
$\begingroup$

An interesting point. This issue was debated on meta.SO: Why is it possible to vote for deletion/undeletion more than once? It seems that SE developers somehow neglected to implement the same check that exists for close/open votes, and they do not (yet) consider this a problem worth fixing. They do appear to be open to revisiting this issue if delete/undelete wars become a noticeable problem. One argument from the SE side is that if an undeleted question does not get reopened, then it does not deserve to stay undeleted; they do not want sites to be depositories of questions that can't be answered. From this point of view, the fact that deletion is facilitated by allowing same voters to delete again is somewhat by-design.

Data

Not counting the question that prompted this thread, there are only 57 questions that were undeleted more than once and remain undeleted. I went through their revision histories: in the majority of cases it was the question owner playing with the delete button to see what it does. In the 3.5 years of the site, there have been only four tug-of-wars over deletions that resulted in undelete winning (again, excluding kalle-numbers):

Votes to delete

There could be as many (or more) cycles that ended in the deleted state; they are not found in SEDE. I can't say that I care about those. My point of view is close to that of Shog9 (as is often the case): if a question does not have enough support to stay open, I see no problem with it being deleted, by whatever means. I'd like to see more closed questions being deleted, not fewer. Hence I would not support a social norm that inhibits deletions.

Votes to undelete

If I were to suggest a social norm regarding undeletions, it would be: vote to undelete only if either

  • you believe the question should be reopened in its current state; or
  • you are going to improve the question so it will have a chance of being reopened.

In other words, do not undelete just so that "it's there". There are other ways to preserve content: Wayback Machine already does this for kalle-numbers, and 10K users can repost the deleted content outside of SE as they see fit. (In the public folder of a Dropbox account, for example.) The question owner can do the same. The point of deletion is not that the thread does not deserve to exist in the world, but that it does not fit the high signal/noise ratio model of SE.

Delete/undelete wars

As shown above, those are rare. I think it's best to deal with them on ad hoc basis, and not try to build a norm about something so abnormal. If the delete/undelete war over this particular question draws enough attention to indeed become a distraction, a mod can lock the post for a long period (say, a year) or even indefinitely. I think we are not there yet. It takes a lot to become a distraction on a site with 300-400 new questions dropping in every day.

$\endgroup$
  • $\begingroup$ To be clear, I'm not asking about delete/undelete wars, or viewing them (for this question) as a problem, or looking for a way of dealing with them. One can always request votes on meta, and as we both have advocated, the vote request thread should be generalized to cover more than reopen votes (I would include all collective actions like close, delete, undelete). Concerning vote wars I am only arguing that nobody should deliberately initiate a multiple-vote in such a war. Also, see the edit to the question for some benign exceptions, but which do not touch the basic principle. $\endgroup$ – zyx Jan 7 '14 at 7:02

You must log in to answer this question.

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