6
$\begingroup$

I saw this on my reputation changes for September 14, 2015:

enter image description here

so I checked the three questions where I should have unaccepted answers. Since after a whole day the situation didn't change and the reputation page still shows wrong data, I thought it better to ask here.

Yes, I still have the green tick; when I saw the three “unaccept” marks, I went to see and found the ticks were still there:

https://math.stackexchange.com/a/535404/62967

https://math.stackexchange.com/a/847173/62967

https://math.stackexchange.com/a/1204127/62967

Note that this is the complete report for the day and no “voting corrected” or “user removed” entry is shown, so this apparently is not something due to serial or anyway bad voting. If it is, then we have a different bug. In other words: if the “unaccept” is not a voluntary act by the user, I should be notified that this “unaccept” is a correction by the system.


In Normal Human's answer below, the picture shows “voting corrected” entries, which makes his/her case different from mine.

$\endgroup$
  • 2
    $\begingroup$ This might have something to do with the serial voting scripts. There are a couple previous similar bug reports on meta.se (one and two), and two of the three users who unaccepted your answer unaccepted other answers on that day. Quite odd all around. $\endgroup$ – user642796 Sep 16 '15 at 8:40
  • $\begingroup$ @ArthurFischer There is no serial voting involved in this case, as far as I can see. $\endgroup$ – egreg Sep 16 '15 at 8:48
  • $\begingroup$ That's why I'm hedging. (That and, like almost everyone, I have only at best a vague idea of how these scripts work.) The only similar reports I can find on meta.se seem to centre around the serial voting scripts, which is why I bring that up. $\endgroup$ – user642796 Sep 16 '15 at 8:56
  • 1
    $\begingroup$ A "voting corrected" entry is only generated if there are a certain number of votes between the users involved. I don't know what the threshold is. For a single vote, there's just an "un(up|down)vote" event created. $\endgroup$ – Daniel Fischer Sep 16 '15 at 11:53
  • $\begingroup$ @DanielFischer No, it's a different problem: one of the three users involved in this “mysterious unaccept” has no other accept vote invalidated and is still present in the system. I have had no “voting correction” in several days; if this user was involved in something, I surely am outside the matter. And no, you can see in Normal Uman's answer that even a single event has generated an entry. $\endgroup$ – egreg Sep 16 '15 at 11:58
  • $\begingroup$ The fact that these events happened at the same time as Normal Human's (among other users) leads credence to the notion that they are strongly connected. $\endgroup$ – user642796 Sep 16 '15 at 12:06
  • $\begingroup$ Yes, you are outside the matter, and only involved in so far as some votes on your answers were invalidated (there are others in the same situation). The "-10 Voting corrected" event in Normal Human's profile was not a single reverted vote (rep cap, or CW answers, don't know which, caused only 10 points to be deducted). $\endgroup$ – Daniel Fischer Sep 16 '15 at 12:06
  • 1
    $\begingroup$ @DanielFischer You seem to be informed about the business; I can understand SE doesn't want to disclose information, but I don't like to be treated as “keep silent, we know what we're doing”. So I'd like to know an “official” position about the matter. I have four events regarding removed upvotes and two regarding removed accept votes. I don't think the threshold is above three, is it? So I'd simply like to be informed that something was going on. Why one accept vote by a user is invalidated and other three (by the same user) aren't is something I don't understand. $\endgroup$ – egreg Sep 16 '15 at 12:11
  • 2
    $\begingroup$ I don't know what exactly is going on. I know a little bit about how things work, but the details are above my pay grade. At the moment, we're waiting for it becoming day in the USA so that the people with more shiny tools can shed some light. $\endgroup$ – Daniel Fischer Sep 16 '15 at 12:15
  • $\begingroup$ @DanielFischer Sorry, but I'd have liked something official, before addition of the “by-design” tag. There have been five hours between your last comment and the edit. From you adding the tag, I deduce you got information from the staff. I'd like too. $\endgroup$ – egreg Sep 16 '15 at 22:18
  • $\begingroup$ At this meta.se post, there is meanwhile an answer that says it's by design that the checkmark (and the accept status) remain, although the reputation for the acceptance is removed. Beyond that, it's exactly as Normal Human wrote. There was a group of users with highly unusual voting patterns. Votes to all who got several votes from that group (including people who just have answered questions from that group) have been invalidated. Investigations continue. $\endgroup$ – Daniel Fischer Sep 16 '15 at 22:51
  • $\begingroup$ The answer to the post you refer to (a similar question, answered after thirty-one months), is about extremely unusual voting patterns and is about invalidation of votes between two users (or more, I add). I'm sure I have no unusual voting pattern with the three users we're dealing with. Do you feel I'm too fussy about reputation? Maybe. But I also feel to be a not irrelevant contributor to StackExchange, who deserves, at least, some respect. $\endgroup$ – egreg Sep 16 '15 at 23:04
  • $\begingroup$ I'm sorry to have made a bad impression with the way I expressed things. The facts, as far as I know them, are that there was a group of accounts with highly unusual voting patterns, and that when the votes of that group were invalidated, the invalidation included all votes to users who received more than one vote from the group. That does not mean that such users are suspected to be more than innocent bystanders, but the circumstances were apparently such that this measure was deemed necessary. $\endgroup$ – Daniel Fischer Sep 17 '15 at 0:25
  • $\begingroup$ The further investigations concern only that group. I see now that that could easily be read differently from how I intended it, my apologies. $\endgroup$ – Daniel Fischer Sep 17 '15 at 0:25
  • 2
    $\begingroup$ I cannot possibly know why not all votes where invalidated (or even if this is the case). But, abstractly, there is a trade-off between concerns about the integrity of voting and affecting innocent bystanders; it seems quite reasonable to me to fix some cut-off. If you'd ask me why the cut-off is one vote rather than say two votes, I could not give a compelling reason, beyond something had to be fixed. $\endgroup$ – quid Sep 17 '15 at 18:18
5
$\begingroup$

This is an effect of vote invalidation. When votes between two users, or all votes of a particular user, are invalidated, the accept-votes are invalidated too. The answer keeps its "accepted" status, but since there is no longer a valid accept-vote, the 15 rep gain is gone. You can also see that hovering over the affected checkmarks shows "loading when this answer was accepted..." which is somewhat misleading and has been reported as a bug, but is intentional.

As Arthur Fischer noted, you are not the only user affected by this invalidation event. Here's my reputation tab: same day, same time.

my rep

So, there was serial voting involved; your reputation history wasn't affected enough to show this.

This does not look like automatic vote invalidation (which would happen around 3:00 UTC). I think it was triggered manually, by an SE Team member who was notified of an irregular long-time voting pattern (possibly involving multiple accounts) and pushed some buttons.

By the way, the effect was somewhat dramatic in the topbar notifications: lots of -10s and +2s (only a few fit on the screenshot):

notif

$\endgroup$
  • 2
    $\begingroup$ Sorry, but I can't buy this explanation. I was not involved in serial voting from those users, which is testified by no “voting corrected” entry in my daily reputation changes. So it's a bug anyway: either I should not be deprived of rep points or I should be notified that some correction about votes has been done. $\endgroup$ – egreg Sep 16 '15 at 11:44
  • 3
    $\begingroup$ Votes don't have to be serial to be invalidated. If an account is recognized as a sockpuppet, all their votes are likely to be invalidated, serial or not. I don't disagree that these particular events look odd. $\endgroup$ – user147263 Sep 16 '15 at 11:47
  • $\begingroup$ @Daniel Thanks for adding the important "c". I still hold that displaying "loading..." when nothing is ever going to get loaded is a bug, but that's a tangential point. $\endgroup$ – user147263 Sep 16 '15 at 16:58
  • 1
    $\begingroup$ I agree. But it's probably too low priority to get fixed in the near future. $\endgroup$ – Daniel Fischer Sep 16 '15 at 17:00

You must log in to answer this question.

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