Automatically protecting popular questions after a given amount of time

Question

While discussing on chat that some popular questions get locked or protected to prevent unnecessary answers, I thought that it might be useful to automatically protect older questions which meet certain criteria. Ivo Flipse said that he's already been going around and manually protecting more questions recently which garner a lot of "thank you" and "me too" answers, and it would be nice to lighten his load and have this done automatically to some extent.

I am proposing the following general criteria; please respond in answers/comments about how they might be tweaked:

Auto-protection for questions which:

• Are at least 1 month old
• Have either an accepted answer with a score of at least 5, or at least 3 answers (none accepted) with a combined score of at least 10
• Have one or more answers with a score of 0 or less
• Have at least 5000 views (upped from 200 as per Jeff's suggestion)

These questions would have received enough views to be popular enough to perhaps need protecting, and have some high-quality answers already. By checking to see if they already have at least one low-scoring post, we know that it's already been the target of the types of questions we try to prevent by protecting questions.

This would lessen the work for moderators by reducing the opportunities for low-quality posts to be written in the first place, but would not prevent new users from contributing and starting to build their rep with newer questions.

What are people's thoughts? Would this be beneficial? Should the parameters be different?

---

Some stats

I ran a query with some of the criteria above (not all, because I didn't have time to play around a bunch, because I want to go to bed). Of the top 10 results sorted by descending view count, 8 have already been protected. This would strongly suggest that questions of this type ought to be locked, since some moderator's going to have to do it eventually anyhow.

Sorted by ascending view count, only 3 are protected (or locked), but many of the others had a large number of superfluous answers and maybe should be protected. It's less clear with these whether they should be automatically protected, though I don't think it would hurt any of those questions.

If anyone wants to put together a more thorough data query, please do so.

Related Meta Stack Overflow discussion: Should we automatically protect all questions with mote than N answers.

Answer 1

200 views is comically low. I would only suggest this happening automatically at 5k views or higher.

Post protection has also been proposed as a 20k rep ability.

Answer 2

Whilst I can see the motivation, this doesn't seem like a good idea to me. Whilst (as was said by Jeff) it is important to not set the range for inclusion too low, it is also important to make sure that the range for inclusion actually works.

I have flagged many 'thanks' or 'me too' type answers, and I don't think that even one was on a question that met all of these criteria. I'm also a bit concerned about pre-emptive protection -- if this site is supposed to be community driven, think of the hassle a new user will go through if they are trying to post an alternative answer. Technology, both hardware and software is ever evolving, just because an answer was the best answer that existed at the point the question was protected, does not mean that some time later an even better answer cannot come along. I've seen many new users come to the site simply because they want to help out by answering a question, and what constitutes a 'good' answer may change drastically over time. For example, if you asked me what was a good home operating system in 1989, my answer would differ greatly from my answer now.

Whilst I think this is a valid idea and I can see the good intentions behind it, these questions constitute a minority on Super User to the extent that I think the result will be more bad for users than good for moderators. If they require protection, let it be after someone makes that protection necessary -- a statistical likelihood of unwanted answers is not mutually exclusive to a statistical likelihood of good answers.

As an aside, the top 10 results by descending viewcount are far, far above the criteria specified. It is important to look at the lower end of the spectrum just as much.