1

I came across a question on serverfault today that is about 18 months old. The reason I found it is that a 1 rep user had just added an answer. I didn't notice the date at first, and added my own answer as well. This isn't a huge deal for one question, but what about the general case? Does it make sense that any question over, say, a year (or even just six months) should be automatically protected?

I know technology changes and we want the ability to keep older questions up to date, but remember that protecting a question only locks it for users with a small rep.

What about including views as part of the equation? A question that earns a "Popular Question" badge is probably getting enough views from low rep users to really need protection and enough views from higher rep users to get the edit and new answer attention when needed, but lower view questions maybe either don't need the protection or do need the edit/answer help from anyone who does happen to land at the question.

Update:
For context, here is a link to a query that shows actual answers that would have been blocked under this proposal, so we can judge the typical quality. Note that to accurately judge this you have to view them in the context of answers that are already available to the question at the time the answer is added: http://data.stackexchange.com/stackoverflow/q/93288/

And here is a query to show how many questions would be effected:
http://data.stackexchange.com/stackoverflow/q/93289/
Notethat this is a naive query: it only checks answer count rather than whether the answer has an upvote, and there's no way right now to know how many of these are already protected. Using the current data set, the result is 114039, or a little less than 1/10th of all questions, and because of the other limitations it's probably even smaller.

|improve this question
  • 4
    "We never free a mind once it's reached a certain age. It's dangerous, the mind has trouble letting go..." – The Unhandled Exception Feb 18 '11 at 15:37
  • The edit to include the view count is a nice addition to the idea. Though maybe then the age can be taken out (or be limited to, say, a few days rather than months)? – Arjan Feb 18 '11 at 16:13
20

I argued against this back before we had protection, and the similar thought on basing it by views.

The potential of a user to have the correct answer is independent of reputation. Spam comes from new users, but correct answers don't only come from experienced users. Age does not properly indicate that a question has been solved. And protecting an unsolved question is a double penalty - it denies entry to a potentially very constructive new user by blocking their ability to answer, and it prevents the question author from getting the answer that they may have waited months for.

Asking and answering is the heart of the site, it is the main thing we do as users and what we continually encourage new users to do. When users can't comment or vote or post bounties, we tell them to post great questions and answers to get enough reputation. But even 10 reputation is impossible to achieve if the one question you can provide an excellent answer to is protected, just because you were late to the party. And that's blocking the content that we want.

I had prepped this from when Diago proposed the 1k views thing, but here's as good a place as any.

Open, unanswered questions on Stack Overflow with over 1000 views

Open, unanswered questions on Super User with over 1000 views

Now, some of those are junk questions, but not all of them. On Stack Overflow, nothing's more recent than last July, while on Super User there's one from December. And this is just a search of all questions which are visibly unsolved. Views don't make a reliable measure. I still have a couple unanswered questions on Stack Overflow, which of course don't have anything resembling views, and I've progressed in the code past needing the direct solution. But getting an answer would provide some nice closure, and depending on the quality it may well make a substantial change to my workarounds.

|improve this answer
  • The queries are interesting... but what if we also excluded unanswered questions (ie, a question is not automatically protected until it has an upvoted answer)? – Joel Coehoorn Feb 18 '11 at 18:27
  • Also, here's a query that shows actual answers that would be blocked under the proposal, if you want to peruse those for a sample of typical quality: data.stackexchange.com/stackoverflow/q/93288/… Most of them seen decent, but don't forget to look at them in context of answers already available to the question at the time. – Joel Coehoorn Feb 18 '11 at 18:37
  • @Joel For disclosure, I started off with my first answer on Stack Overflow last February (I just got my Yearling yesterday ♪) being to something not only old, but with an accepted answer. I got my first 25 points that way, and it was very nice. In my scenario, it was only 2 months, but that's only by luck. Protection is meant to protect against spam and noise, but unless something is getting repeated noise, then there isn't as much need to protect it. I don't see pre-emptitive protection being all that posititve. – Grace Note♦ Feb 18 '11 at 18:48
  • As for the query, @Joel, does that take into account the user's reputation at the time of the answer? – Grace Note♦ Feb 18 '11 at 18:51
7

This is a bad idea, and I would not support it. See Grace Note's response.

We are allowing 15k rep users to protect questions, but I would only expect them to do so when the question ...

  • is at least a month old
  • has a reasonably high # of views (at least a few hundred)
  • has proven track record of generating clueless or spam posts as visible to 10k users who can see the deleted trail

In general, questions are innocent until proven guilty and should not be protected until there is reasonable evidence that the question needs protection. See above.

|improve this answer
5

I second Arjan opinion and just wanted to add my two cents: sometimes adding new answer to old question can be good thing, for example some new way to solve the original question.

|improve this answer
  • 2
    Any time you're saying something like "I second ___", that should probably either be a comment or a vote instead of an answer. – Joel Coehoorn Feb 18 '11 at 15:44
  • +1 for showing that protection won't stop folks from adding answers that maybe should have been comments ;-) – Arjan Feb 18 '11 at 15:50
  • 1
    A good example of that would be an answer I gave on a 2 years old question : stackoverflow.com/questions/41469/… I don't think the Google Book API existed when the question was asked, but now it exists so it's good that now people that find this question get to know it's there. There are lots of old question that do benefit from having new answer even if they are really old. – HoLyVieR Feb 18 '11 at 15:52
  • (Hmm, I will be deleting my answer now that the feature request has been enhanced. Trouble!) – Arjan Feb 18 '11 at 16:03
  • @status_completed - you have enough rep to answer that question, even if it's protected. Remember that I'm not talking about locking questions. You only need 150 rep to answer protected questions. – Joel Coehoorn Feb 18 '11 at 18:50
  • @Joel No, you only need 10 reputation, actually. Which is tiny, but insurmountable when it is blocking the avenue that is intended for earning reputation. – Grace Note♦ Feb 18 '11 at 19:00

You must log in to answer this question.

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