-3

I have been recently looking into 'edit reviews' that I have voted on and voting didn't go unanimously.

My contested suggested edit reviews

Where I have discovered a few cases where code sections were deleted or changed in a ways that would give different meaning to question (basically vandalized), and they were voted 2-3 and (thank God) got rejected, but it was a close call. Which would imply that there are cases when questions are being vandalized and approved. I am not sure how much development effort would this require (would it be cost efficient), but I believe that I would add some additional protection if 'suggested edits' that have edits in code section, and have non white text changes:

  • Would be reviewed by more members (10 instead of 5).
  • OR members with higher rep (lets say 10k+).
  • OR those that never have been banned (or failed an audit).

Examples:
1 - 3 https://stackoverflow.com/review/suggested-edits/5157499
1 - 3 https://stackoverflow.com/review/suggested-edits/5157509
1 - 3 https://stackoverflow.com/review/suggested-edits/5237130
3(?) - 2 https://stackoverflow.com/review/suggested-edits/5245577 (error fixed in question by edit.)
1 - 3 https://stackoverflow.com/review/suggested-edits/5245886
2 - 3 https://stackoverflow.com/review/suggested-edits/5256250
1 - 3 https://stackoverflow.com/review/suggested-edits/5256316
1 - 3 https://stackoverflow.com/review/suggested-edits/5264448
2 - 3 https://stackoverflow.com/review/suggested-edits/5226830
1 - 3 https://stackoverflow.com/review/suggested-edits/5219793
1 - 3 https://stackoverflow.com/review/suggested-edits/5219209
2 - 3 https://stackoverflow.com/review/suggested-edits/5209635
2 - 3 https://stackoverflow.com/review/suggested-edits/5199644
3 - 2 https://stackoverflow.com/review/suggested-edits/5195383 (I see this as a fail)
1 - 2 https://stackoverflow.com/review/suggested-edits/5172041

2
  • This wouldn't work. SO already requires 3 votes in one direction, instead of 2 on other sites. Higher rep isn't correlated to better reviews. I'm an experienced suggested edit reviewer, albeit I haven't done much of it recently, and I haven't seen much of the behavior you describe: code edits to question that do anything but indent and the like are rare. Please post links. Commented Jul 19, 2014 at 9:55
  • @Gilles I have updated question, I do agree that it works most of the time, plus if we would factor in that it is a small set of only ~100 'edit reviews', and it can go ether way. Commented Jul 19, 2014 at 10:54

1 Answer 1

5

Questions

  • https://stackoverflow.com/review/suggested-edits/5157499 — The question was clearly edited within the 5-minute grace window, and the suggestion came during that code. The history is wrong in this cases, it's a long-standing bug. Whatever the edit was doing, it wasn't intended as vandalism: the part added after the signature wasn't present when the suggested edit was started. In any case this has nothing to do with the fact that the edit happened to touch some code.

  • https://stackoverflow.com/review/suggested-edits/5157509 — I don't understand the context, but do note that the author made the same edit a few minutes later, so it can't have been that stupid a suggestion.

  • https://stackoverflow.com/review/suggested-edits/5245577 — Yeah, too minor is arguable (the edit does leave some obvious problems), but accepting is arguable a well (the edit does fix several issues). This edit was approved by the author.

  • https://stackoverflow.com/review/suggested-edits/5256250 — This edit does two things: it corrects English and formatting in welcome ways, and removes part of the code. There's a clan that prefers to have full working code in questions and a clan that prefers to pare down code to the part that exhibit the issue that the question is about. Having full working code is prefered, but removing the packaging isn't “basically vandalism”.

  • https://stackoverflow.com/review/suggested-edits/5264448 — Another one from the 5-minute window.

  • https://stackoverflow.com/review/suggested-edits/5219793 — Again an edit that combines grammar corrections with paring down code. This time, the removed code is genuinely useless, not something that's necessary to build the program. So the edit does make sense — the asker should have posted that pared down code. It's risky to change code in questions (because you might accidentally change the problem that the asker was facing, or change what solutions are applicable), so it should be done sparingly, but here the edit wouldn't cause any problem, so it's ok. Stack Overflow tends to be very set against editing code in questions (and sometimes, for absolutely no good reason, in answers), so I'd expect a majority for rejection if this was debated on SO meta, but this edit isn't actually wrong.

  • https://stackoverflow.com/review/suggested-edits/5199644 — I don't know, is the reshape2 library needed here? This should probably have been a request for clarification in a comment rather than an edit, given that there is nothing in the question (at least to my ignorant-about-R eye) that suggests that the asker did include that library and just forgot to copy-paste it: the issue could have been related to the lack of inclusion of that library.

Answers


On the whole, I don't see any trend here other than the fact that there are bad reviewers and debated topics. There's certainly nothing special about edits in code sections: there are plenty of bad edit suggestions that get accepted, and of good edit suggestions that get rejected, and on the whole this isn't correlated with the fact that the edit happens to affect code. Stack Overflow does have a specific problem with too many edits that fix code in answers getting rejected, but not particularly with edits to code in questions.

1
  • Hi Gilles thanks for explanation, I hope you don't mind that I will be deleting Question as it has -3 ATM, and apparently I haven't put enough effort into research hehe. Commented Jul 19, 2014 at 16:03

You must log in to answer this question.

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