35

Today, someone came up with a rather clever way to game the badge and reputation system through the suggested edit system. They controlled two accounts, where one account continued to suggest trivial, nonsensical edits while the other (the question author) readily accepted them.

This process was repeated almost 60 times.

Of all the ways to game the system, that one never crossed my mind. However, since it's come to light, we should probably limit the number of suggested edits a user can make to a post. It would also be helpful to raise a system generated flag if that threshold is reached.

I can't imagine why a single user would need to suggest more than five edits to a post, so I propose we start the limit at ten. Additionally, I know there are a limited number of times you can vote on suggestions by an individual user, but does this apply if you're the author of the post being edited? If not, maybe it should.

17
  • 1
    I can't imagine why a single user would need to suggest more than five edits to a post, so I propose we start the limit at ten ? Commented Dec 27, 2012 at 10:50
  • 10
    I've never understood why any one user should get credit for suggesting multiple edits to the same post. Since edits are supposed to fix all problems with the question, seems like if you have to suggest more than one edit then you didn't do your job correctly the first time. Commented Dec 27, 2012 at 10:53
  • 4
    @psubsee2003 Mistakes can genuinely happen. We should limit possible gaming ways without punishing those who do it honestly. Although you're right, we need to try and make a good job the first time.
    – Alenanno
    Commented Dec 27, 2012 at 10:55
  • 3
    @psubsee2003 It keeps things simple, and there are cases where an edit is suggested, accepted, post edited again, and then another suggested edit to fix the other edit. We also work under the presumption that the majority of accepted edits are good (despite repeated discussion here about some not being so good).
    – user50049
    Commented Dec 27, 2012 at 10:55
  • 6
    @Alenanno Oh, I know mistakes happen.. I'm not actually suggesting limiting it to 12 suggested edit, I'm just saying you should onyl get the 2 rep points for the first suggested edit/ Commented Dec 27, 2012 at 10:56
  • 1
    What would prevent the user from gaming the system in the same manner but with 5 edits on 12 posts instead?
    – Sklivvz
    Commented Dec 27, 2012 at 10:57
  • @TinyTimPost ok, that's the hole in my logic.... if someone else makes an edit that needs additional fixing Commented Dec 27, 2012 at 10:57
  • @Sklivvz A whole lot of eyeballs. The incident today was detected rather quickly by the community (even prior to the meta post), despite being a holiday. The limits could actually be much lower and still not get in the way of most editors. This proposal more or less limits the amount of noise the activity makes for everyone else, and lets us know what's going on.
    – user50049
    Commented Dec 27, 2012 at 10:59
  • 1
    @Sklivvz There is also a limit to the number of edits one user can vote for another, but I don't know if this actually applies if the author of a post is the one approving. If it did, an attempt with more breadth would probably fizzle out sooner. It's a good question.
    – user50049
    Commented Dec 27, 2012 at 11:04
  • 1
    @BhuvanRikka I'm well known for my less than optimal estimating skills :P
    – user50049
    Commented Dec 27, 2012 at 11:08
  • 2
    @psubsee2003, I'm not sure there's a hole in your logic. If there is a mistake that the editor didn't notice first time then why should they get rep for fixing it a second? You say any one user, so limit the maximum rep from suggested edits to 2 per post and your first comment is correct (and a good idea). Commented Dec 27, 2012 at 11:14
  • 1
    I find quite often I need to make multiple edits to a post for new users. Fairly frequently after formatting code into code blocks I find the OP then makes an edit themselves and somehow manages to remove this formatting so it needs to be redone. Commented Dec 27, 2012 at 12:17
  • 1
    I thought this was about limiting the suggested edits one could have out at a time; a couple times I've seen one user suggest like 40 tag wikis at once (predictably they consisted entirely of the first paragraph of the relevant wikipedia article).
    – Zelda
    Commented Dec 27, 2012 at 12:42
  • @benisuǝqbackwards The hole is my logic is the review queue. If everything works as it should, then there probably is only very limited need to re-edit the same question more than once other than to fix mistakes you missed originally (so my original suggestion would work). But if crap edits get approved, someone might need to go back to fix something again. Commented Dec 27, 2012 at 14:21
  • Yes @psubsee2003, but there should be no need for a single person to edit it more than once. So, each person should only get rep for editing the question once. Multiple people can still suggest edits and get rep. Commented Dec 27, 2012 at 14:24

1 Answer 1

2

If the gaming is done to grant reputation to a sockpuppet account, then this could be handled as part of vote fraud prevention. In this particular case, the suggestor quickly received reputation due to the actions of one user, which is very similar to serial upvoting. Handling this case as part of serial voting detection would catch both cases where all the suggestions are on one post and cases where they're spread over many different posts. This would also catch the same thing done on tag wikis, where such a string of edits could easily go unnoticed.

If the gaming is done to bump the post, it's going to quickly backfire as it did, because someone's bound to notice and flag.

If the gaming is done to grant a reviewer badge to the main account, it's more difficult to detect. But that's a rather bizarre way to get a badge. Either it's done in a high-activity review queue (e.g. SO) and it's easy to find 20 naturally-occurring suggestions to approve every day, or it's done in a low-activity queue and some other reviewer is bound to notice the sudden influx of weird suggestions.

A single post with that many edits should have become community wiki anyway (why didn't this one?). 7 approved suggestions (enough to grant the sockpuppet enough reputation to vote) wouldn't do it, but an account whose reputation exclusively comes from 7 different events from the same source should be detected as part of vote fraud prevention. And there may well be cases where a single <100-rep user wants to edit a CW or tag wiki post many times (for example, the maintainer of a piece of software, who updates a tag wiki every time he releases a new version but doesn't otherwise participate on SO). So I do see corner cases where your proposed limitation would be legitimately be too restrictive. Besides this, it's an ad hoc limitation to catch a rare fraudulent case, which I think should be handled behind the scenes (à la serial voting detection) and not through built-in limits. All in all, I agree there's a problem, but I prefer a different solution.

You must log in to answer this question.