17

Serial downvotes usually get detected and reversed automatically every night. This event is usually grouped under a header "Serial Downvoting Reversed" in a user's reputation history.

On various occasions however, I have seen such a reversal take place, yet appear as a regular "Undownvote". How do I know they are reversals and not the user undoing his downvote? They all take place at 3am UTC at the exact same moment. And in several such instances the votes would have been locked anyway.

One example is the reputation history for this particular victim for September 24th. All 3 undownvotes took place at exactly 03:00:04.

So I guess the votes get invalidated but not properly grouped/marked?

  • it is worth also noting that revision history doesn't show edits / ninja edits that would allow to revert a recent vote – gnat Sep 24 '13 at 16:17
  • How do you know that the downvoter didn't actually just undownvote them himself? – Sam I am Sep 24 '13 at 16:42
  • 1
    @gnat Still, 3 undownvotes in a row at exactly the same time (up to the second)? Exactly at the same time the filter runs? Unlikely. (And this is not the first time I've seen this behaviour) – Bart Sep 24 '13 at 16:46
  • @SamIam See my comment above, and the actual content of my post. – Bart Sep 24 '13 at 16:47
  • 1
    @Bart up to second is unlikely, agree (imagining 3 browser windows opened and trying to figure how quickly could I shoot at all three arrows in these...) – gnat Sep 24 '13 at 16:49
  • @gnat And I have a bit of an ace up my sleeve: chat.meta.stackoverflow.com/transcript/message/1541359#1541359 This question was posted after a discussion with a developer, upon his request. ;) – Bart Sep 24 '13 at 16:49
  • amount of mistakes I made as a developer is hard to count. :) With all due respect, I tend to rely more on revisions history that tells me there were no recent edits / ninja-edits that would make it possible to revert vote – gnat Sep 24 '13 at 16:53
13

This was literally a > that should have been a >= in a rollup algorithm somewhere. New code has been deployed and the 1,383 users this affects will be fixed over the next few minutes.

You must log in to answer this question.

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