When to wield the "Protect Question" ability?

Question

With only 25 additional reputation needed, I expect that I will soon gain the ability to protect questions. I'm wondering what best practices are for this particular stack exchange in this regard.

It seems to me on the one hand, that this particular topic seems to and is likely to draw particularly crummy answers which seem to be associated with new and anonymous users and my inclination is therefore to protect any question which is eligible.

On the other hand, I also don't want to discourage new users, but expect that there will be enough unprotected questions available for them to gain the pittance of a mere 10 reputation needed to comment - not to mention there is always the association bonus.

What are other's thoughts on this? What should be the best practice in wielding this ability?

Answer 1

I look at three data points when deciding whether to protect a question:

1. The age of the question
2. Whether the question already has a good answer
3. Whether it has received multiple bad answers

Generally speaking I'm looking for some combination of a question that has been around for a while, already has a good answer or two, and is attracting numerous low quality answers as well.

For example, this question is a pretty textbook case (so much so it looks like protection for it was automated). The question was posted probably back in the first month of the site's existence, it has three strong answers, and it has four deleted answers and another three answers that aren't positive additions.

I view the "protect question" tool as a reactive rather than proactive tool. As you mentioned, the topic of our site probably lends itself to attracting poor quality answers more than others. In general, though, we want to be open to new answers, and especially so where our questions are new, are unanswered, or don't show a pattern of low quality answers.

Answer 2

Basically, I think Soldarnal has nailed it, but I thought it worth adding a few bits to elaborate the same position.

I think it is also important to understand that there is a StackExchange network-wide ethos here, that we (at BH.SE) ought to participate in. Some of the Mets.SE FAQ is reflected in a (typically) helpful post on the SE Blog by Shog which resonates with Soldarnal's criteria. I want to flag up two guidelines from that faq/post:

• Do protect questions that are attracting a lot of non-answers or very poor answers (spam, etc.) from new users.
• Do unprotect questions that aren’t currently attracting a lot of attention and don’t have a long history of unproductive answers.

I believe we have a growing number of questions that are prime for unprotection because they are (a) low-traffic; and (b) have very few answers whether live or deleted. (Personally, I refrain from protecting a question unless there are >= 3 deleted questions, and possibly some aspect of "junk magnet" aspect to the question, e.g., my last protection action).

Users who can wield protection powers can also see the list of protected questions, with the following data displayed (but columns only seem to sort on the page displayed, not across all data; there are six "pages" of data):

Question | Question Date | Protected By | Protected On Date | # answers past 30 days | # deleted answers

It is accessed from links at the bottom of the ("10K") "Tools" tab on the Review page, which itself displays recently protected Qs:

If anyone is interested, I've tweaked a query on the Data Explorer which should give all protected questions on BH.SE, with default sort descending on AnswerCount. (I think some rogue data is pulled in, but it's a start!) Note that "answer count" is net (i.e., only live answers). I couldn't work out how to include deleted answers, or give the "raw" answer count -- if anyone wants to fork and refine, please do!

Answer 3

Anytime a question gets a reasonable number of answers from new users. If you can see 4 or more answers from users with 1, 11 or 101 rep, and they aren't all excellent, then I think it's fine to protect the question. I don't have the rep yet to protect questions myself, but I frequently write custom flags asking if the mods can protect questions.

This site's topic is something everyone could feel qualified to answer questions in, which is a good thing, but new users frequently don't know our guidelines for answers, such as the need to show your work. Any question which appears to be attracting answers from new users who got here via search engines can be protected. No one wants to see yet another unsupported answer filled with opinions and proof texts, so protect away!

Answer 4

Please see comments below. This approach is kept for historical reasons, but is not necessarily recommendable.

Kind of new on hermeneutics.SE, but I have had the power for a bit on mech.SE and obviously have the power on Internet of Things (a site I moderate.)

From my perspective, protecting is best done as little as possible. In other words, if it's a question that attracts new users - great! Get them on board. In the beta period, we need more users who will stick around, post, vote, ask, etc.

Personally, I will definitely not protect a question before it has one deleted answer, and usually not before it has two. Bear in mind that protecting is essentially restricting access.

It happens from time to time that I visit a site where I do not have the rep to answer protected questions and see a protected question that I feel I could really add something to. Unless I see that I really, really have something important to add, it's rare that I'll bother sticking around to get the extra reputation required. Rather, I just move on and forget all about it.

On the other hand, if a question is generating a large quantity (I tend to go with 25-40% of answers) of NAA / VLQ flags, protecting just makes moderation easier and keeps the question cleaner. In that case, I personally say, "Protect!"

N.B.: This is just my personal bent; feel free to disagree.

Answer 5

As one new to this ability I would like to float the idea that all eligible questions should be protected and see how it is received. I think this tactic is the optimal course of action for the following reasons:

• First, the cardinal rule of etiquette is to never discuss religion or politics. The reason for this rule is that these topics are polarizing and compel people to 1) comment further on these topics and 2) to do so in a manner that is gutteral, reactive and not particularly academic, something which goes against our site distinctives.

• Secondly, this problem is unique to this Stack Exchange site. For example on a random day by my count of the first 50 questions of the homepage/list of active questions, the Christianity Stack Exchange had a 40% closure/on-hold rate for questions. Comparatively, Server Fault had a rate of 4% and Stack Overflow had a rate of 0%. A similar trend is seen in the answers to those questions which remain unclosed. To some degree this disparity reflects the discrete and binary nature of computers in which things either work or don't work and are either true or false. This leads to answers that are usually more appropriate to the questions asked and are less speculative. Our site's requirement that questions reflect a degree of academic rigor more apropos of a University than a Bible Study also contributes to this problem.

• Third, there will be less low quality answers requiring moderator attention

While I am sensitive to the fact that there is a network wide ethos established regarding this, not all subjects are created equal and I think the above factors are great enough to warrant an exception in the case of this topic.

Of course, the concern here is that protecting all eligible questions will limit a new user's ability to participate in the site. I would suggest, however, that this actually provides a desirable barrier to entry which will help to raise the overall quality of content on this site due to the following:

• New users will be more likely to post questions than answers. This will force them to see what good answers look and become more acquainted to the site, whereas they might otherwise skip the site tour before mentally vomiting onto the keyboard in the answer field. Hopefully they will be more likely to emulate these answers after seeing them modeled.

• New users will be forced to perform more due diligence and work harder to achieve the minimum 10 reputation to post. This will help encourage this trait in new users - a trait which is required for answers which show their work

But ultimately, this still provides very little barrier to new users. They would still only require a single answer up-vote or only two question upvotes. This is a significantly low bar of entry and despite any additional modest burdens protecting question may laden upon new users.