25

I have noticed what seems to be a trend of general confusion about what should be done when pure IP networking questions are posted to Stack Overflow. I have collected some very recent examples over the last few days:

To summarize:

Some users and moderators dislike theoretical IP networking questions on Stack Overflow and close them:

Sometimes theoretical IP networking questions remain on Stack Overflow, as suggested by Gilles:

Sometimes networking implementation questions remain on Stack Overflow (instead of migrating to Server Fault)

Sometimes theoretical IP networking questions get punted to Server Fault:

Sometimes DNS and other networking services questions get punted to something other than Server Fault, because of another popular keyword in their text

  • One domain, multiple IP addresses (Posted to Stack Overflow, migrated to Webmasters because he was talking about websites, but the the real issue was intelligent DNS responses, not HTML or httpd)

In this example the OP asked questions about how to use research-grade techniques, but he got punted to a forum that cannot help.

To moderators and question-close voters:

My bottom line suggestions for improvement are:

  • If the subject is IP networking, please do not merely close the question as off-topic; place it in what you think is the right forum.
  • Could someone address IP networking in the Stack Overflow FAQ to reduce ambiguity?

I write this because IP networking is my career, and like programmers need Stack Overflow, networking specialists need some consistent place to get a good answer... And the field is wide open... It is not just generic questions about IP networking or how do I deploy XYZ service... If it was well-done, it could conceivably expand into what is currently dominated by non-Stack Exchange sites right now... like Cisco NSP, Juniper NSP, and perhaps smaller network vendors which don't have a useful place for users to gather and collaborate.

I'm pretty sold on the value of the Stack Exchange network, but it can only be as good as the consistency of moderation for a given subject. With great respect to the moderators on Stack Overflow, Server Fault, and Super User, the consistency of moderation for networking questions seems like it could be improved.

If I am correct, what is a good path to get there? If I've misunderstood something, please set me right. Thanks to all the Stack Exchange staff and volunteer moderators who have built a fantastic discussion framework for programmers.

BOUNTY GUIDELINES

This bounty will be awarded to the person who can demonstrate that the following have been done:

  1. Stack Overflow's FAQ has been updated with exact guidelines regarding IP Networking questions. I even have some suggested guidelines:

    • Any IP networking question involving host-level theoretical or practical TCP or UDP issues belongs on Stack Overflow
    • ALL other IP networking questions belong on Server Fault.
  2. The IP networking questions listed above as closed should be either reopened, or migrated to the appropriate site.

Just because Gilles has a good answer to my original post does not mean I would not award the bounty to another person who makes the changes above.

4
  • 4
    Have you seen Server Fault? Bunch of networking questions there... That said, some of your examples are just... really poorly asked. If you ask a question on the wrong site, and don't include enough information for users not familiar with your topic to pick the right site, it's hard to expect good results.
    – Shog9
    Commented Apr 19, 2011 at 21:44
  • @Shog9, I brain-farted above, and I corrected the original post... yes he should have gotten punted to ServerFault. Users can help moderate and ask clarifying questions... part of the reason I'm doing this is because I want to be sure I'm suggesting the right thing when I flag the question on SO. Commented Apr 19, 2011 at 21:49
  • 2
    You're barking up the wrong tree here. It's not up to the devs to give a home to these questions, it's up to the community. Even if these particular questions are reopened or migrated, this doesn't mean they will get good answers or that future questions will get to live. If you feel strongly about this, you'll not only need to gain some kind of consensus on Meta, but also to watch the site and refer would-be closers to some Meta consensus. And arrange for the questions to get good answers, by answering yourself if necessary. Commented Apr 22, 2011 at 22:06
  • @Gilles, that is why I'm saying two things need to happen... and that includes modifying the SO FAQ with whatever the right direction is for networking questions. So far, I disagree that this is barking up the wrong tree. Commented Apr 22, 2011 at 22:39

2 Answers 2

11

There's no single Stack Exchange site about networking (IP or otherwise), and some topics fall through the cracks.

  • Development questions (how to use sockets or other APIs, how to implement a certain protocol, …) belong on Stack Overflow.
  • Questions about networking-related hardware and software can be appropriate for several sites, depending on the type of equipment and on your perspective:
    • Questions about professional equipment (e.g. Cisco routers), or about complex server or multi-client setups, belong on Server Fault. Basically, if you're a pro, ask on Server Fault.
    • Questions about home equipment or simple client setup belong on Super User. Basically, if you're a user, ask on Super User.
    • If your question targets a specific platform, you can ask on a platform-specific site:
  • Questions that are mostly interesting to webmasters can be asked on Webmasters
  • Questions specifically about security can be asked on Security

Questions that are purely about networking protocols fall through the gaps — none of these sites welcomes them.

Going through your list, here's my opinion:

Given what you state about your interests, I'd say you would fit on Server Fault. Even if it's dominated with Windows and Unix questions, there are networking professionals as well. There was a prior proposal to create a specific site for network administrators, which was closed as a duplicate of SF.

If you like, you could propose a site about networking on Area 51. But that would have to be about networking in general, not specifically about network administration. However, considering that only a small subset of networking topics are not covered by an existing site already, I don't think that's a good idea. And a site specifically about networking protocols feels too specialized.

Officially, questions about computer science belong on Stack Overflow (except for research-level theoretical science, which has its own site). Network protocols are applied computer science, so I'd say that they're on-topic because Jeff says so — but the community doesn't welcome them.

10
  • @Gilles, thanks... your answer gets at the source of my confusion... I was seeing what you are calling computer science questions, but to me these are questions that looked pure IP networking; so my assumption has been that they should be punted to a non-SO site. If this is the community direction, distinguishing between theoretical networking questions (appropriate for SO) and professional implementation issues (appropriate for SF) clarifies a lot for me. Commented Apr 19, 2011 at 22:30
  • @Mike: Et what? This question is asking for a software tool, it has nothing to do with CS theory. None of the questions you mention are remotely appropriate for CSTheory. Some of them would be ok on a hypothetical site about all CS. (Note that I'm a proponent of such a site, but as I mention in my answer the official position is that it's a subset of SO.) Commented Apr 19, 2011 at 22:47
  • @Gilles, uggg... my brain fart again. I was confusing two posts... this is the question that should go to CS Theory Commented Apr 19, 2011 at 22:57
  • @Mike: definitely not TCS either. What makes you think it's on-topic there? Asking about a tool is very much applied CS, and the tool isn't anything research-related, any more than asking about screwdrivers would be research-level physics, even if an applied physicist might need an oddball screwdriver on a piece of equipment. Commented Apr 19, 2011 at 23:08
  • @Gilles, he said he is doing his graduation thesis... projects like this sound very much research-related. Speaking as someone who operates networks all day long, there is no production tool that measures bandwidth from one host... all rely on some form of two-sided feedback. AFAIK, his needs are for something that isn't available to the general public off github or sourceforge. Commented Apr 19, 2011 at 23:32
  • @Mike: If he wants an existing tool, that's SU. If he was asking about a tool he was writing, that would be SO. If he was asking about applicable methodology, that would be applied CS: not CSTheory.SE material, but IMO not SO material either. Commented Apr 19, 2011 at 23:36
  • @Gilles, all he knows is that he wants a passive-bandwidth measurement tool... I have no idea whether he would write it... AFAIK, people in a research list are happy to share their experimental source code with others, and that's what I think will help him. Commented Apr 19, 2011 at 23:41
  • @Gilles, can you or someone else explain why What is a VALID subnet is deleted, if theoretical IP networking questions are allowed on StackOverflow? Commented Apr 21, 2011 at 11:07
  • @Mike: one more that fell through the cracks. Unfortunately, there are enough people who don't want them on SO or SU or SF to close them, and there may be enough people who think they do belong to close an Area 51 proposal as well (but this hasn't been tried, the networking proposal was specifically about network admin). Commented Apr 21, 2011 at 11:14
  • @Gilles, thank you for clarifying... I will add it to my list of examples above. Commented Apr 21, 2011 at 11:17
3

I am going to argue that the answer is the new Network Engineering site.

You must log in to answer this question.

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