There has been a shift in tone in ServerFault over the last few years. When it started, the focus seemed to be on providing an open resource for professional admins to help figure out problems. Now it's evolved into a place where people can only ask specific questions with exacting detail.

The impetus for that change seems to have been based in the fact that there were a large number of bad questions being asked that really didn't have any answers, because of the vast open scope of the problem, usually because it was a specific problem with an astounding lack of detail. So it was codified that questions should be well-defined.

But not all questions are so easy to nail down. Some problems that professional admins encounter are matters of preference, but sometimes we need help in making sure that we're seeing all of the points to be weighed. Just because a question is open-ended doesn't mean that it's inherently a bad question.

It seems to me that the whole point of StackExchange sites is to allow the community to decide whether a question is good or not. There is still a need for moderators to deal with spam, egregiously bad questions, and other administrative minutiae. But it also seems that some people with greater privileges believe that if a question doesn't match their ideal for a question, then it shouldn't be asked. And some of those ideals have been codified in the FAQ. And I think that that's wrong. I'm not looking to save my question (for one thing, I don't really think it's in danger of being closed), but it has prompted me to be more open about my concern for the community.

In my opinion, this exclusive focus on specific questions threatens to turn the site into (a less annoying) Experts Exchange. And I'm not saying that to be spiteful. (On a related note, I think there's not nearly enough focus on "Too Localized".)

Some concrete possibilities, as described in the SE blog post Good Subjective, Bad Subjective (and pointed out by voretaq7 in comments) are to have a separate SE for more subjective questions or have a specific set of guidelines for subjective questions. (The ones in that blog post seem to make sense fairly universally.)

However, I think we just need to refocus on topics that are useful discussions for the professional sysadmin community, and allow the community more leeway to moderate itself via voting.

Related: Good Subjective, Bad Subjective –  voretaq7 Jan 3 '13 at 18:19
Earlier… –  Iain Jan 3 '13 at 18:41
The desire to flag this as not being a question is almost overwhelming. ;-) –  Mark Jan 3 '13 at 18:41
More… –  Iain Jan 3 '13 at 18:43
Now that you've had your rant, what's the question? –  John Gardeniers Jan 4 '13 at 12:37
@wfaulk is my hero for the day. This has been a growing concern to me. (Evidenced, for example in here: –  gWaldo Jan 4 '13 at 15:49
If the goal has been to scare new people away from asking potentially naive questions on ServerFault, the mission has been mostly accomplished. But I think these questions have their place, especially for us mere mortals. Sometimes I need to be told that something is non-trivial, too general, or that I should provide more information in order to get a better feel for my own question. –  tacos_tacos_tacos Jan 9 '13 at 17:50
I'm surprised this hasn't been closed already as 'not constructive'.. :) –  Ian Lewis Aug 9 '13 at 10:59
add comment

4 Answers

By and large, the mod team on Server Fault does not "mod hammer" questions, despite our constant joking about it in chat. We certainly do not do so to questions with a large proportion of upvotes (though we may mercy-kill something racking up infinite downvotes).

We have a contingent of active users with the Vote To Close privilege, and many of them use it. Sometimes it is used a bit overzealously, but generally it's used fairly judiciously.
You've been around long enough to know that most of those users hang out in chat, along with pretty much the entire mod team. If you see things getting closed too aggressively feel free to knock heads together (or until you hit the reopen rep mark flag them and we'll knock heads for you).

The broader question of whether we should allow "open-ended" questions on the site is a good one.

I for one have no problem with design/architecture/best practice questions stemming from thorough research and practical concerns -- there may be no One True Answer to these questions, but the collection of answers they accumulate can serve as a valuable community resource (for example, our canonical question about compromised servers has many answers, each of which contributes something useful though no single answer is comprehensive).

Open-Ended questions that are "Good Subjective" also tend to be deeper questions, more likely to attract quality users and produce a useful pool of knowledge that will then bring in even more (hopefully high-quality) talent.
We certainly don't want to become an Experts-Exchange or Yahoo Answers style "do my work for me" site with one-liners answering relatively trivial (or even esoteric) questions -- Depth and context in the discussion of any issue is something Server Fault should always be looking for.

I didn't mean to imply that some people were injudicious in their use of the mod hammer, but merely that some people who have higher privilege levels (like vote-to-close privilege, which includes myself) have an attitude that I find to be closed-minded. Even then, they don't seem to abuse their authority. It's the attitude I have a problem with, and the people with privileges are the elders of the community. –  wfaulk Jan 3 '13 at 19:08
@wfaulk Oh no, I'm the one implying some people are (occasionally) injudicious in their use of close-votes :-) –  voretaq7 Jan 8 '13 at 16:34
add comment

wfaulk, I think your original question is a good one. It could have been a shopping question ("what's the best linux distro?"), and we know those are OT anywhere on SE, from the famous "QA is hard, let's go shopping" post.

However, that very post notes that while "What’s the best low light point-and-shoot camera?" is a product recommendation request and therefore OT, "How do I tell which point-and-shoot cameras take good low light photos?" is a good and useful question, which should provoke good and useful answers. Your question about linux distros seems to me carefully phrased to be an excellent example of the second class of question above.

Moreover, and I don't know that I'm unique in this, I extend more latitude to users with a decent amount of rep. If someone new to the site breezes in and asks a borderline-OT/NARQ/NC/TL question, I'll probably vote to close it then and there. If someone with a couple of thousand rep asks a similar question, I'll certainly think twice, maybe leave it and come back later, wait to see how the rest of the community takes it. In some cases I've voted to reopen even after the community took against such a question.

That's because I know that an experienced user will have thought carefully before posting a borderline question, and that they likely have a good reason for doing so. I have had that same courtesy extended to me, and although SvenW noted that my question was borderline, he answered it, and the answer was absolutely invaluable to me. I am very grateful that those who could have voted to close my question stayed their hands.

I'm not saying that all who have close privileges should think the same way, but that's how I approach the very real problem stream of duplicate, product recommendation requests, and incomprehensible and unanswerable questions we face every day, whilst still trying to let the good questions swim on their merits.

Extending more latitude to those with "decent rep", which I suspect most of us do from time to time, can be a dangerous thing and needs to be done with the utmost discretion. It can have the adverse effect of creating an excusionist environment if we're not careful. The last thing we need is users saying things like "why is it OK for a high rep user to post it but it's not OK for me?" –  John Gardeniers Jan 8 '13 at 0:45
I like your thought process here, but I think that we should extend the same level of courtesy to all users, regardless of rep. –  wfaulk Jan 8 '13 at 15:41
What MadHatter and wfault said. I'd expand on this to say that when a new user is greeted with downvotes and closed questions, he's likely to leave. It's important for mods to give more leeway to newbies, and to help them to refine their questions. Someone who asks poor questions and does not improve needs to be shut down, politely. –  Jon of All Trades Jan 11 '13 at 20:14
With respect, whilst it's a well-defended point, it's also the absolute opposite of what I said. I was most definitely not advocating that newbies should get more leeway. –  MadHatter Jan 11 '13 at 20:20
add comment

An item you forget to mention is that there have been numerous new SE groups generated including:

 DBA SE group
 UBUNTU SE group
 UNIX&Linux SE group
 SuperUser SE group
 StackOverflow SE group
 Security SE group
 Webmasters SE group
 Apple SE group

In the past there were only a couple of UNIX/Linux based SE there are many so I believe we are all honing in on the specific purpose of ServerFault and attempting to differentiate this group from the other SEs.

I honestly don't understand what your point is. –  wfaulk Jan 3 '13 at 18:45
Actually, we shouldn't be trying to differentiate ourselves. SE employees have stated multiple times that a question can be on topic on more than one site and shouldn't be migrated away just because it's also on topic somewhere else. –  MDMarra Jan 3 '13 at 18:57
@mdpc I think you are mostly proving wfaulk's point. The more SE adds groups with specific topics, the less we will have on-topic questions on SF. –  Alex Jan 3 '13 at 19:05
I'm with Alex. The more specific SE sites there are, the more likely we are to say "This belongs on <specific SE site> rather than here". On the flipside of this I don't see that even being an issue since those issues that could be migrated to a specific SE site are not if they are sysadmin-related or if they could even translate into a sysadmin scenario. I've only seen us push the power-user type questions to the other sites. –  August Jan 4 '13 at 12:47
@August The more specific SE sites there are, the more likely we are to say "This belongs on <specific SE site> rather than here" - Please read my comment two above. This is the wrong attitude. If someone asks how to do something on their server in bash, it's on topic here, and U&L. That doesn't mean that it should be migrated to U&L if it is also on topic here. If something is asked and is currently on topic, it should not be migrated just because there is a more specific site. –  MDMarra Jan 4 '13 at 16:03
@MDMArra, It can be on topic on many SE sites...but, just like this question… I can bet my pocket change that if there is a SE sites specific to the question, it WILL be closed as off-topic. –  Alex Jan 4 '13 at 16:31
@Alex That question is off-topic for SF, because it's about a Raspberry Pi, which is not a piece of server equipment that is used in a professional capacity. It has nothing to do with it being more on topic elsewhere. If someone asked a question about apache on a real server, we wouldn't migrate it to Unix and Linux just because they were running it on Red Hat, for example. –  MDMarra Jan 4 '13 at 17:11
@MDMarra, I'm sorry but his question has nothing to do with the platform being used. The question solely relates to vsftpd and umask settings. And the question is absolutely legit from that point of view. On top of that, assuming a Raspberry Pi cannot be used in a professional environment is wrong. We are now testing them as out-of-band console access for our firewalls, switches and routers. Does it mean I would be off-topic as well? –  Alex Jan 4 '13 at 17:29
@Alex It appears that 5 users as well as at least one moderator agree that that specific question is not appropriate for the scope of server fault. It's also getting off of the point a bit. That question aside, if something is on topic here and also on topic at a more specific site, it shouldn't be moved. –  MDMarra Jan 4 '13 at 17:40
@MDMarra, agree to disagree...I think this is a perfect example of what the OP of this meta question was trying to describe and the answer we are commenting being precisely the reason why the question I was referring to was closed. –  Alex Jan 4 '13 at 17:44
@Alex Considering the fact that the founders of Stack Exchange, as well as many of the community management employees have been very vocal about this, there's not really anything to disagree about. If there's a database question that's asked on SF and is on topic, we should not migrate it to DBA. If there's a Solaris question that's on topic here and also at Unix, we don't migrate it. It's really cut and dry. Things can be on topic on more than one site. Things shouldn't be shuffled around just because there's a site whose name more closely matches the question. –  MDMarra Jan 4 '13 at 17:52
@MDMarra I think you misunderstood me. I am not advocating that a question be migrated simply because there is a specific site for the topic. I'm only saying that as more and more specific sites are brought up on SE, the more that is GOING to happen. I thought I made that clear if you read the rest of my comment... –  August Jan 4 '13 at 17:54
@Alex: You have to consider broken windows‌​. The vast majority of, for example RaspberryPis (there are many other examples) are not and never will be professional server systems. If we leave them open then others will use them to justify their asking similar questions and thus SF's mission become diluted. –  Iain Jan 4 '13 at 22:14
@alex The other thing to remember is where might a question get a good answer. Rather than looking to gerrymander the on-topic/off-topic sections of a site's FAQ, a much better thing to consider is where a question will get highest quality answers. For example, questions about a "home server" will generally get better answers on Super User because there are more experienced in that scenario there. There are a lot of assumptions about doing things "the right way" that we make as pros on this site, whereas on superuser they may be more open to "good enough for home use". –  RobM Jan 23 '13 at 22:47
add comment

I agree wfaulk. I had a question re. tcptraceroute closed by overzealous mods seemingly because they thought tracert did the same thing or that it was "shopping question". (How can I find out what troubleshooting tool I need without it being considered a "shopping question"?!) As far as I know, at least one of the votes to close was from someone who incorrectly answered my question and has by virtue of this demonstrated poor technical knowledge in networking. A closed question is discredited and won't get the attention of people who may actually possess the technical knowledge I visit SE to obtain.

I think a few people on SE need to remind themselves of the real purpose of this site.

No moderators were involved in the closing of your question - it was members of the community. –  Iain Jan 23 '13 at 7:02
I wasn't involved in closing the question, but it is a shopping question. And closing questions like that for the various reasons discussed in here over the years has been part of "the real purpose of this site" for a very long time. –  RobM Jan 23 '13 at 8:21
You asked for a tool to do a job. That's a shopping question. Had you described a problem and asked for solutions your question would still be open. The difference is is very significant and makes the question either on or off topic. –  John Gardeniers Jan 23 '13 at 8:29
Apologies Iain - semantics - I didn't mean moderators per se. RobM - and so do you see my point? You cannot simply ban all questions inherently requiring the mention of tools which you need to achieve a solution! Say someone using Linux wants to find out how to add resilience to their storage iscsi mount using multipathing but doesn't know how to do it - is this banned? Because the answer requires mention of something like multipathd... I can see the need to protect against hugely general questions like "what's the best Linux distro" but this is different. –  James Jan 23 '13 at 8:29
OK John, I've re-worded it. But the fact is that the solution requires a particular tool...! Or I'll get something like "you need the Windows equiv of tcptraceroute for Linux"... This censorship just doesn't seem helpful - it's removing an intrinsic element of the question and helps explain to people what I'm trying to achieve (which others can then reuse in the future). –  James Jan 23 '13 at 8:42
add comment

You must log in to answer this question.

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