Although I think there have been many good contributions by anonymous users, would it make sense that a near 100% change not be allowed?
We have the review process, but it just seems that it probably should not get that far when this much of the original post is modified.
.
This was not an audit as I thought it probably was.
Sometimes someone will post a code snippet with no explanation or a link to something that's useful with nothing else. These answers are useful if expanded upon. Depending on the length of the original and the length of the additional information, this could easily exceed 50% increase in character count. 85% or 90% isn't unheard of for justifiable edits.
Additionally, the usage of an anonymous user isn't a clear indicator that the edit won't be justifiable, because if it was then anonymous edits would be outright disallowed.
Here are some downsides I see to this:
A few points:
On SE, we don't have anything against "anonymous users". We believe that users' contributions should be accepted optimistically, unless we have a specific reason to suspect a particular user of bad behavior. Since anonymous users are anonymous, we can't expect or anticipate anything of their behavior. Some people may think this gives them an unfair advantage, but actually it doesn't: the only way that an anonymous user's contributions could be met with prejudice is if that same user registered for an account, and persistently engaged in bad conduct on the site. Also, anonymous users miss out on a lot of benefits of registering and participating in the community.
Based on what anonymous users are allowed to do, their role is fairly limited, but useful nonetheless. They have a similar role to anonymous Wikipedia editors who see an obvious typo in an article and decide to go in and edit it. The "long tail" of drive-by contributions from the Internet at large has been shown to be a useful and significant contributor to overall site quality, both on Wikipedia and here, and even in software development projects such as the Linux kernel (although a little more time, skill, and effort is required to get a patch merged into Linux). When your audience is the entire world, it's OK if some of your users just want to contribute one little thing. Hey, we'll take it!
The very purpose of the review queue is to catch bad edits (such as ones that completely change the meaning or intent or opinion represented by the original content) irrespective of the extent of the edit. The percentage changed is a poor metric of determining whether an edit is acceptable or not. For instance, if the formatting is very broken in someone's original post, an anonymous editor could decide to clean it up with proper markdown formatting, which could show up as a very high percentage of changed text in the post due to all the whitespace and markdown changes, grammar, etc.
If none of the above have convinced you, consider this: Let suggested edit reviewers do their job. Suggested edit reviewers are volunteering to review edits by other people. We have a very good body of edit reviewers on SU, such that, as I write this post, we have 0 Suggested Edits in the edit queue. The only queue we have of any significant length is the Close Votes queue.
That's all I have to say. TL;DR: Let edit reviewers do their job, and we aren't bias against anonymous users' contributions in any way.
As others have noted, blocking edits that are too large can prevent legitimate edits from getting through. Example
The real issue is the one described at Improve filtering of spam suggested edits. Spambots have recently learned to make suggested edits to Stack Exchange post, and there's still some work for the SE team to do to detect and reliably block spam suggested edits without risking false positives. For now, human reviewers will handle this issue, and this is why we have the review queues in the first place.
An additional thing I've noticed is an email address in the edit summary. The spambot is probably treating the edit summary field as though it were the email field in the comments form of a blog post. We should reject suggested edits with edit summaries that that consist primarily or entirely of an email address.
I’ll be the lone voice in the wilderness supporting this proposal, possibly with some modifications. The phrase “anonymous users” seems to have struck a nerve. allquixotic says, “On SE, we don't have anything against "anonymous users".” Well, that’s an oversimplification. We do grant users privileges based on reputation points earned. Anonymous users have zero reputation (arguably, indeterminate, since they have no track record). An anonymous user is unaccountable; he could have submitted any number of acts of vandalism, and we wouldn’t be able to associate him with them. Every registered user has a positive reputation. So rather than single out anonymous users, let’s say that users who do not have a positive reputation have some slight restriction imposed on their ability to edit.
Raystafarian and DragonLord cite examples where a valid edit could increase the size of a post by over 100%. OK, good point. Let’s say that the metric should be simply how much of the original post is deleted by the edit. It’s hard to imagine a valid edit that deleted more than, say, 85% of the original text.
Multiple respondents have raised the concern that such a restriction would block good edits. Well, we have a rule that only users with at least 50 reputation can comment everywhere. I say that this rule is blocking good comments. I know, because I continually see good comments being posted as answers. They then get flagged as “not an answer” and usually deleted – the “convert to comment” option gets exercised way too infrequently, IMO.
Finally, there’s the issue of how such an (apparent) hatchet job should be handled. The easy answer is to accept them and then silently discard them. Perhaps more people would agree to a scheme where such edits are simply not accepted, and feedback is given. This would be similar to the way SE doesn’t accept a comment that’s less than 15 characters, or an answer that’s less than 30. Maybe vandals would just use this feedback to tailor their vandalism to fly under the radar, but maybe it would also discourage some of them.
The only reasons I see to implement an auto-rejection of this kind of edit would be if:
Some of these edits are somehow sneaking through the Suggested Edits review; or
There are so many of these invalid edits coming through that they are swamping the review queue.
I really doubt that (1) is the case, but it could be easily verified by a little analysis of past edit reviews:
If this analysis shows that the review process is failing (again, doubtful), then maybe this would be worthwhile.
As for (2), the suggested edits queue is rarely above single-digits at any given time, so this just isn't the case.
That said, barring (1) being true, there's really nothing to gain from implementing this sort of restriction on anonymous edits. The current procedures keep such spam edits from making their way into posts, and they don't threaten to weaken or over-burden these procedures.
If you're reviewing suggested edits and come across one of these, just consider it a gimme for your badge progress and move on.
