What's the communities view on questions of a theoretical nature, I.E. things/situations that would never happen in real life but are still quite interesting. Of course, this would only reffer to things that were still within the FAQ/Rules (I.E. have a possible answer based facts and are actual questions, rather than discussions). They might even contain some information relevant to real life things.

An example I was thinking about earlier;

What if my datacenter had open racks, and anyone could just wander in and plug their laptop into an Ethernet cable behind our firewall. Is there any way to prevent Ethernet connections to the network from unknown MAC addresses without having to firewall each server.

Obviously IRL the real problem would be having such an un-secure datacenter, but in this theoretical situation what's the best course of action?*

So yeah, are questions about unrealistic/impossible things acceptable?

*Don't answer this question here, I actually had a think about it and have probably figured out a way to do it, I just wanted to give an example. So yes, the example theoretical question is in itself a theoretical theoretical question. My brain hurts

  • That particular theoretical question might get some traction on security.se. Searching for ways to implement defense in depth, and considering the potential of a physical network breech seems like it might be on-topic to me.
    – Zoredache
    Commented Feb 3, 2012 at 17:50
  • @Zoredache You completely missed his point. He was posing a hypothetical question in the context he is trying to address.
    – Publiccert
    Commented Feb 3, 2012 at 18:02
  • @Publiccert, I think I was pointing out that a better example might be useful.
    – Zoredache
    Commented Feb 3, 2012 at 18:14
  • The example is still valid and hardly necessitates a downvote.
    – Publiccert
    Commented Feb 3, 2012 at 18:16
  • @Publiccert, downvoting on meta sites usually means someone disagrees with the point, not that they think think the question bad.
    – Zoredache
    Commented Feb 3, 2012 at 18:20
  • Fair enough. Pretty new to meta but I can't say I'm a fan of downvoting for an opinion. Don't agree? Leave it be. But that's just me.
    – Publiccert
    Commented Feb 3, 2012 at 18:21
  • @Publiccert Its a quick and easy way of seeing if people agree with a post or not. It might seem a bit harsh at first but most people are aware of that here so they (I) don't take it personally
    – user80776
    Commented Feb 3, 2012 at 18:44
  • FWIW, that's actually a pretty good question. We've got a half-rack in a secure datacenter for a government contract and this is exactly the sort of thing we've had to implement. Commented Feb 3, 2012 at 20:13
  • Also, people seem to forget that meta has no effect on overall reputation.
    – Tablemaker
    Commented Feb 3, 2012 at 22:04

4 Answers 4


I think if it's not an actual problem, just a brain teaser or artificially constructed situation, it has no place on SF. This would be better suited for some kind of forum (which SF by definition is not).

  • 2
    And, ask in chat instead! I'm sure chat users would love to hash out contrived situations. I hear technical-minded folk like a good puzzle. :P
    – Aarthi
    Commented Feb 3, 2012 at 18:09

I think "theoretical questions" is another way of saying "debate/discussion" and those have no place here - Not only is it against the FAQ but I think the site is poorly structured(*) to deal with "so what happens if I... ok well how about if I then..." type discussion.

By the way, I'd say your example of a theoretical question could turn out to be a real world example, if a field engineer is working in a server room, by the way, so its arguably a valid question. ;-)

(*) nothing wrong with this of course, as the site isn't supposed to be for those kinds of posts.


It depends on how theoretical your question is: As long as it can be translated to an actual practical problem it's probably fine IMHO.
Specifically looking at your example, strike the "What if" bit and it's a situation I've encountered many times. It can be translated to:

My company's datacenter is a mess: Our equipment is in 4-post telco racks and the door to the room has to be left open for cooling or people start hanging turkeys in the racks for Thanksgiving dinner.
Because of this anyone could just wander in and plug their laptop into an Ethernet cable behind our firewall - Obviously not a great situation. Is there any way to prevent Ethernet connections to the network from unknown MAC addresses without having to firewall each server?

This question is ikely to solicit a bunch of comments to the effect of "Sweet Something of Somewhere, buy an air conditioner and secure your freakin' room!", and also likely to get some useful answers about using managed switches, MAC address filtering, and making sure you admin-down all the unused ports (or assign them into a dead vLAN).

  • 2
    +1 If the question could be worded so as the difference between theoretical and actual problem fades then it should be fine. If it's whimsical thought experiment then there's no "correct" answer as such.
    – Chris S
    Commented Feb 3, 2012 at 19:38
  • 2
    The correct answer is "Keep the cat in the box forever. At the end of the universe heat death pretty much ensures the cesium atom will have decayed and Kitty will have snuffed it." (This sort of thinking is why Schrödinger stopped inviting me to his dinner parties.)
    – voretaq7
    Commented Feb 3, 2012 at 19:43

At the moment, open-ended questions seem to get shutdown pretty quick. Would I like to see them? Very much so. It would be interesting if we used a standard tag, i.e., 'fantasy', or 'hypothetically'. I can't see it hurting, as long as it focuses on facts and not opinion. You starting getting into opinions and you'll see flame wars sprouting up all over those tagged thread.

  • 3
    a) those tags you have suggested are worthless (blog.stackoverflow.com/2010/08/the-death-of-meta-tags) b) give some example of questions that you feel would be on-topic, could have a useful answer, but have been closed.
    – Zoredache
    Commented Feb 3, 2012 at 18:10
  • a) That's your opinion. See how that works? b) The OP already gave a perfectly good example. While not perfectly suited to Serverfault, it's a good example.
    – Publiccert
    Commented Feb 3, 2012 at 18:14
  • 5
    No, that is the option of the majority of mod, and stackexchange owners. The owner didn't give a good example. I think it would have gotten good answers as written on security.stackexchange. Voretaq7, suggested a minor edit that would result in it probably not getting closed, and getting useful answers here.
    – Zoredache
    Commented Feb 3, 2012 at 18:17

You must log in to answer this question.