17

My profile says I lost 2 points because "suggested edit approval overridden".

lost 2 points

When I checked the answer, when I checked the revision history I found that the community revision from 2018/07/18 was rolled back (#3) but mine was kept (#2).

So if my edit was kept, why did I lose 2 points?

Update: This is not a duplicate of this question. The indicated question is asking for power to overturn recent community review decisions. I am asking why, when the author reverted to my revision, I lost points.

14

August 2nd, 2018: I've corrected the underlying bug (votes are now only deleted when overriding approval on non-anonymous edits) and fixed the remaining 23 occurrences of this bug across 11 sites. Thanks for the report!


Here's the short explanation:

  1. When a suggested edit is approved, the system creates a revision for that edit and adds a special type of vote to the post. That vote, attributed to the last approver and targeted at the edit's author, is worth 2 points. This step is skipped for anonymous edits, as there's no user to target.

  2. When a suggested edit approval is overridden, the system looks for the last "approve" vote on the post and deletes it.

So the problem here is that when the system went to delete the vote associated with the overridden edit, there wasn't one - so it found the vote from your edit instead. And deleted it.

This has happened 23 times previously on various sites across the network.

I've fixed this particular instance; we'll need to find time to fix the other occurrences (and fix the underlying bug).

  • Why not issue a vote from the last approver to the Community user if an anonymous suggested edit is approved, instead of not issuing one at all? That way, the existing logic can work. – Sonic the Anonymous Hedgehog Jul 26 '18 at 3:32
  • Because that vote serves no purpose, @Sonic. Unless / until we allow folks to register and "claim" their edits, no one will ever get rep for them. Also... I have 23 votes to fix this way vs. thousands your way. – Shog9 Jul 26 '18 at 3:42
  • Um, can't you just implement my fix for all future edits only, and fix the prior 23 occurrences? – Sonic the Anonymous Hedgehog Jul 26 '18 at 4:12
  • If it only avoids the problem for future edits, then anyone overriding any of the fifteen hundred or so anonymous suggested edits currently sitting as the latest revision on posts with a previous, non-anonymous suggested edit would still break, @Sonic. – Shog9 Jul 26 '18 at 4:37
  • Ah, I missed that. What's the best course of action here, in case those get overridden? – Sonic the Anonymous Hedgehog Jul 26 '18 at 4:56
  • Easiest fix is to stop trying to votes that can't exist because there's no target user. Once that's done, I can manually fix the handful of cases where this has actually happened. – Shog9 Jul 26 '18 at 4:59
  • I think there might be another logic flaw in this algorithm. Say that there are two approved suggested edits on a post, one sometime in the past, and one the current revision. The final user to approve the current revision deletes their account, which removes the approve vote and causes the suggester to lose reputation (which has been described as intentional). Then, that edit is subsequently overridden by the author. As the last approve vote was for the prior suggested edit (since the last edit's vote was removed when the last approver deleted their account), the previous editor loses rep. – Sonic the Anonymous Hedgehog Aug 2 '18 at 15:45
  • It'll just try to delete the already-deleted vote in that scenario, @Sonic (which will do nothing, since the vote is already deleted). Votes are soft-deleted here: there's a DeletionDate column that gets set. The vote itself remains in the database and keeps everything consistent. That's how I was able to fix the 24 cases where this scenario occurred in the past: I simply cleared the DeletionDate column and recalc'd reputation for the affected users. – Shog9 Aug 2 '18 at 15:53

You must log in to answer this question.

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