The recent discussion around the WebServer Benchmarks question has me thinking.

This question poses a generally useful query that doesn't have one true answer. Part of our mission is to understand the usefulness of "Good Subjective" questions to the rest of the community.

Our current "opinion-based" close reason is not up to the task of indicating that "Good Subjective" questions still have a home at ServerFault and the other StackExchange sites.

The current wording is:

primarily opinion-based

Many good questions generate some degree of opinion based on expert 
experience, but answers to this question will tend to be almost entirely 
based on opinions, rather than facts, references, or specific expertise.

This close reason should be re-worded to better convey this core principle and encourage new question reviewers not to close subjective questions that have merit and value for Professional system and network administrators.

Update 6/19/2014

I'd like to reboot this discussion, especially in light of our recent discussions on recruiting and retaining more talent.

Whether we admit it or not, a fair amount of professional system administration is subjective. Every org and/or admin has their own rules of thumb where there is no best practice or where best practice simply cannot apply. In lieu of A question about padding host names, but there are other examples as well, even ones that we cannot foresee. Yes, some answers can be opinion based, but answers can also be based on logic as well, and all answers are subject to voting.

So, should "Good" subjective questions be subject to closure? Should we take time to define "Good"? In my example above the OP took time to do some research first on his own. It is also a question relevant to professional system administration. The question can generate many answers, but they will not all be of the same caliber. Saying "We don't do that here, what's the big deal" is a terrible answer, and not based on facts, references, or specific expertise. My answer, however, is still opinion based, but there's logic behind it, as well as Pros and Cons weighed within.

Looking just now to copy the hyperlink, I see that question was closed. Pity. Guess what the closed reason is?

  • 1
    I disagree with your premise. That question does not pose a generally useful query. That question will not generate useful opinion based answers beyond saying that Benchmarks not specific to your environment are worthless 80-90% of the time.
    – Zoredache
    Commented Sep 12, 2013 at 21:31
  • 1
    Part of the problem with trying to boil down a hard and fast rule or two for 'good subjective' is that it's impossible to define what a 'good subjective' question looks like. You can say whatever you like about how to deal with a 'good subjective' question and we might all agree but that's not going to help when people aren't certain that a particular question actually is a good question. It seems pointless to over-think solving the first part when the second isn't solvable.
    – Rob Moir
    Commented Sep 13, 2013 at 7:25

2 Answers 2


I'm not certain that it actually is a good question. It doesn't have a right answer, and you can see from the answers that were posted to it that it's likely to generate nothing but shots in the dark and discussion. This isn't typical of "why do we do it this way" questions, but it certainly applies where the best way to do things varies wildly across scenarios and preference. In this respect, it's overbroad and not completely answerable.

That said, I would like to see the reason reworded somewhat, because the clause at the end is not correctly differentiating. It should read something more like:

Many good questions generate some degree of opinion based on expert experience, but answers to this question will tend to be entirely based on opinions as no objective criteria for correctness apply to this question as asked.

I think that is a little closer to differentiating what is constructive from what isn't - if you can't imagine a particular answer being marked as correct and any reasonable sysadmin agreeing that it solves the quandary presented (because it's too much a matter of opinion, disagreement, and circumstance), it ought to be closed as primarily opinion-based.

  • 2
    Personally I think the close reason should really reference Good Subjective, Bad Subjective directly. The point of this close reason is to close Bad Subjective questions, but it's inviting abuse as a hammer for any question asking for an opinion (even with this improved rewrite)
    – voretaq7
    Commented Sep 12, 2013 at 21:16

In regards to @BigHomie's edit:

One of the first things you have to ask yourself in regards to "good-subjective" is whether the question is answerable in a Stack Exchange context. The number of possible end results should be relatively small, with the devil being in the details. More than that and the Q&A becomes forum-like discussion with everyone piling on opinons.

A question with many possible solutions is acceptable if the number that apply in a professional context are few. If the professional scope doesn't narrow it down, or we could write a book about it, then the question is too chatty and not good for the format.

Your example about zero padded hostnames strikes me as borderline. On one hand, it's always good to plan ahead for future capacity. On the other, how many digits is a company specific thing, and speaking as a DNS admin who has worked at a few companies, hostname formatting discussions are right up there with religion and politics.

You must log in to answer this question.

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