8
$\begingroup$

The Stack Exchange team is in the final stages of evaluating a subject for a proposed Stack Exchange site. I'd like to get your feedback and thoughts on this proposal:

Area 51 Site Proposal — Operating Systems

I'm not generally opposed to creating a more-specialized site when it would attract a new audience whose subject is simply spread across too many sites to attract a comprehensive audience. But I wasn't sure if "Operating Systems" fits that criteria.

If you can look past the first dozen-or-so "best choices to write an OS" questions, are the subjects being discussed generally on topic for this community?

We don't have to nit pick specific questions. Even if there is a *bit* of an overlap, we are generally looking at the broader scope of the site as a whole. Basically, we are trying to determine if this proposal will attract a new community, or are we just duplicating or splitting of pieces of this (or other) sites.

So between CS and CSTheory (and Stack Overflow for those low-level systems questions), does this proposal add anything to the subject space of Stack Exchange? Or would you consider the subjects being discussed a welcomed part of this site?

$\endgroup$
  • 1
    $\begingroup$ thx for the courtesy of asking! we discussed this in chat quite a bit. can you guys compile some stats on new sites and whether they are getting out of beta & se mgt pov on that? that seems like a bigger issue for stackexchange that some sites pass the area51 beta hurdle but then get "stuck" in beta for a long time (such as this one, cs.se!) or possibly fizzle & are cancelled later. but personally, think that any proposal that gets the reqd signups (which is actually quite difficult) ought to continue on its opening based on its members (hard) earning it...! $\endgroup$ – vzn May 21 '14 at 23:05
  • 4
    $\begingroup$ I will probably add a more elaborate answer when I find the time, but I'll say now that there has been quite some confusion/concern in chat about the number of CS-subsites that have been sprouting. In particular, I'd say OS should be covered by SO+CS (and, reading the questions, SU+U&L) but is of course the focus on neither site. $\endgroup$ – Raphael May 21 '14 at 23:26
  • 1
    $\begingroup$ I very much doubt that the internet at large will understand the difference between the proposed operating systems site and the existing Super User site. As such, I imagine that the OS people will spend half their lives migrating questions to Super User. $\endgroup$ – David Richerby May 24 '14 at 13:16
9
$\begingroup$

I am deeply ambivalent. I love cs.se, but we haven't been able to attract the critical mass of OS experts required to make a vibrant community, and we only rarely get good OS questions. (Look at the tag, of the 17 questions with a score of 5 or above, 16 were asked in 2012, and the one that was asked in 2013 is closed as off-topic.)

Every question that would be appropriate for the proposed os.se would be on-topic for one (or more) of:

But...

There is no critical mass community of operating systems experts on any of those four sites. Partly this is because any specific OS question is likely to be off topic for at least one of those sites. Questions that include code or about the internals of a specific OS, would be unwelcome here. Questions about kernel internals would probably be off-topic in SU. Questions about OS in general, or about non-unixes (e.g., Windows, OSX, BeOS, VMS, z/OS, VxWorks, AmigaOS, ...) would be off-topic in U&L.

SO probably has the appropriate critical mass, but there the good OS questions get lost in the noise, good questions about OS and distributed system theory will probably get closed as off topic, and, well, SO is so large and diverse that it is difficult for any subcommunity to form and grow there. SO doesn't provide the technical facilities required to build a smaller subcommunity (e.g., portals for specific groups of tags, each with their own user rankings, moderators, community review-queues, chatrooms, meta, and norms about what makes a good/bad question.) SE doesn't provide the technical facilities required to allow a subcommunity to work across different sites (no cross posting or tag sharing.) The only option for building such a community is to propose a completely separate new site on Area51.

The problem (and the opportunity) are large. It's not just Operating Systems (although that's the only one that's close to achieving full commitment at the moment.) There are many, many, many, many, many, many, many, many, many, (oh, and also many, many, many, and many) Area51 proposals that overlap with CS.SE (machine vision overlaps with cs and stats, computer architecture overlaps with cs and electronics, compilers is another example that overlaps between cs and so, quantum-computing overlaps between cs and physics). Clearly there is demand for building cross communities that isn't being met in any other way than through Area 51 proposals.

TL;DR

Summary:

I wish that cs.se could attract all the OS experts that are committing to the Area 51 OS proposal. (But we haven't). I wish that SE would acknowledge the need to allow cross posting and/or other ways of growing subcommunities. (But they won't). So a few weeks ago I sadly committed to the Area 51 OS proposal and, should it go to Beta, will make every effort to help it succeed.

Summary of the summary: I'm ambivalent.

$\endgroup$
  • 2
    $\begingroup$ 1) If there is no critical mass on either site, how can we expect a critical mass on an os.SE? 2) FWIW, you can open and moderate/maintain a chatroom about any topic, afaik. That may actually be a good idea; let the core users of the other sites know where to sent people and that might work out as a condensation point for a community that works on several SE sites. 3) The many proposals illustrate a problem. I'd argue that it's not only a platform but also a people problem (laziness, mainly), but that's a discussion for another day. $\endgroup$ – Raphael May 22 '14 at 7:52
  • $\begingroup$ @Raphael: (1) I agree. The only hope is that there is a critical mass on SO that hasn't been able to find each other, but will find each other at the new site. (Which I agree is unlikely.) (2) Good idea, I'll try it. (3) A platform that only works if its users aren't lazy (and angry, greedy, prideful, lascivious, envious and gluttonous) is unlikely to succeed with the users available on this Earth. $\endgroup$ – Wandering Logic May 22 '14 at 10:33
  • 1
    $\begingroup$ disagree with R.; but it is not clear if he is talking about critical mass of cs.se (which is still in beta) or operating systems within cs.se and other sites. his comment assumes that users come from other sites and that if the subcommunity is not significant on an existing site where it is ontopic, then it would not be significant on a new area51 group. but that is frequently revealed not to be the case. ie area51 is exactly a "condensation point" for minority groups, much superior to se chat rooms (which frankly are not very active in general! even far less than area51!) $\endgroup$ – vzn May 22 '14 at 19:16
  • 1
    $\begingroup$ My prediction is that if OS.SE launches, it'll have the same problem as Embedded Systems (low initial activity), but not the same content quality (so far ES has pretty good stuff). The OS experts who are committing to Area 51 are for the most part not really CS.SE's audience, they're programmers. $\endgroup$ – Gilles 'SO- stop being evil' May 24 '14 at 7:07
  • $\begingroup$ think you have a great point re existing linux/unix sites. it seems like a LOT or majority of OS questions might tend to be linux/unix oriented and there is not much of an audience outside of that... maybe a better title is "OS internals"? dunno $\endgroup$ – vzn May 24 '14 at 18:25
  • $\begingroup$ @Gilles "The OS experts who are committing to Area 51 are for the most part not really CS.SE's audience, they're programmers." Isn't that argument enough that CS.SE doesn't really overlap with the purpose of OS.SE, and that trying to shoehorn the topic into CS.SE is the wrong move? $\endgroup$ – Adam Maras May 29 '14 at 20:17
  • $\begingroup$ @AdamMaras There is no shoehorning: the topic is already on-topic here, whether it has an audience or not. The only argument for going forward with OS.SE is that there's an audience that's looking for a place. If that audience isn't going there, what's the point of creating another place for them not to go to? $\endgroup$ – Gilles 'SO- stop being evil' May 29 '14 at 21:33
  • 1
    $\begingroup$ @Gilles quoth the FAQ (under "Don't ask about..."): "Programming questions, even if they are homework for a class in a computer science curriculum (try Stack Overflow)" and "How a particular piece of software or hardware works (this site is about computer science)" ...and you've just eliminated most of the content that would be provided by the current makeup of OS.SE proposal committed users. $\endgroup$ – Adam Maras May 29 '14 at 22:04
  • 2
    $\begingroup$ @AdamMaras The content that this eliminates is content that's well at home on SO. $\endgroup$ – Gilles 'SO- stop being evil' May 29 '14 at 22:05
  • $\begingroup$ @Gilles it would be, were it not drowned out. See the comments in my answer pertaining to both points. $\endgroup$ – Adam Maras May 29 '14 at 22:05
5
$\begingroup$

Background: I'm a former computer scientist (not in an OS-related field), current developer working on an in-house OS, and I've been around Stack Exchange quite a bit (including being a moderator of the Computer Science SE site).

I do not feel any need for an Operating Systems site. When it comes to OS design, engineering questions go on Stack Overflow, and scientific questions go on Computer Science. Most questions are clearly on one side of the fence, and for those that are borderline, a site can be picked.

There's a whole continuum of questions that are related to OSes to various degrees: security, concurrency, debugging, linking, device drivers, memory management, … All of these topics could go on either the OS site, or the existing programming and CS sites.

This is in contrast with other topics for which I do feel a need for a separate site, such as cryptography (because it overlaps between computer science, math, programming and security engineering) or machine learning (computer science, programming, statistics) or embedded systems (programming, electronics, and what for want of a better word I'll call embedded system administration). In these cases, it's not just the discipline that overlaps but very often individual questions.

Looking at the example questions on Area 51, I see:

  • a majority of programming questions would work on SO;
  • a few more theoretical questions which would work on CS;
  • a couple of questions that are about existing OSes (always Unix variants), which I suspect would be off-topic on OS.SE, e.g. “What gives better performace for syslog, logging through sockets or kernel calls?”, or “Why does /proc/kcore on Linux show as having such a huge size? How does it actually exist on disk?” which a comment states is intended to be off-topic.

OS.SE wouldn't drive traffic away from SO, it's just a drop in the ocean and I'd expect most people who are active on SO and knowledgeable about the topic to keep not noticing the existence of other SE sites.

OS.SE might drive potential traffic away from CS.SE. CS.SE is doing well on theoretic topics (unsurprisingly, given that a significant part of the founding community came via Theoretical Computer Science SE), but lacking on most applied topics. I don't think OS.SE would attract many researchers in OS-related fields, however: if the site launches, considering the example questions and the cross-site participation of the committers, I expect it to be populated mostly by professional developers and hobbyists. Nonetheless, if OS.SE does make it through, it will be hindrance for CS.SE to grow in the direction of applied topics in the OS domain.

I committed to the OS proposal because if it exists, it interests me. However I think that the best course of action would be to close that proposal as a duplicate of primarily SO and secondarily CS.SE.

Rather than attempt to build a new community, we should spend effort fostering the existing community of CS.SE. If you're interested in the theory of operating systems, please join us!. Start asking questions and browse our existing questions (but a word of warning, these are mostly undergraduate-level).

$\endgroup$
  • $\begingroup$ kind of self serving as cs.se mod... "If you're interested in the theory of operating systems, please join us!." that sentence would make more sense posted anywhere outside of this site.... what can we do to attract these users to the site as you propose other than "preaching to the 'insider' choir?" but ofc plz divulge your affiliation. $\endgroup$ – vzn May 24 '14 at 18:27
  • 3
    $\begingroup$ @vzn, but we all know that Gilles is a cs.se mod. Among other things, there is a diamond right after his name, and that's what it means. But the mods don't "get" anything out of these sites except for a lot of work to do with little/no thanks, so he's not self serving in any way. Well okay, they "get" a little diamond after their name. Still not self serving. $\endgroup$ – Wandering Logic May 24 '14 at 21:37
  • 3
    $\begingroup$ @vzn I was mostly in a frame of mind as an OS professional when I wrote this point. That I'm a CS.SE mod is indicated at the bottom of my post, but I've added it to the opening paragraph for historical reference. As for “self serving” — this post is primarily self-serving as an OS professional, because that's the perspective I mostly wrote it from. $\endgroup$ – Gilles 'SO- stop being evil' May 24 '14 at 22:59
  • $\begingroup$ @vzn Regarding promotion, I do try to foster more applied questions when they arise — and I asked a few during the private beta — but I have no promotion channels to people who are potentially interested in SE. In particular I or my colleagues can't ask questions that are too close to our work because we can't just advertise to the world (including our competitors) what we're working on! $\endgroup$ – Gilles 'SO- stop being evil' May 24 '14 at 23:02
  • $\begingroup$ the mods of existing sites all seem to oppose new sites that have similar or overlapping features, there are many cases of that. my suggestion is that those who oppose the creation of the group go to area51 and help convince/inform them via comments etc that they are welcome here. also, a compilation of "top questions relating to OSes" on the cs.se site could be helpful toward that cause... $\endgroup$ – vzn May 25 '14 at 4:40
3
$\begingroup$

I think Wandering Logic brought up several interesting points that should to be studied and discussed independent of this discussion. E.g. it might worth creating and experimenting with a view of questions on SO and CS.SE that gives the feeling of an individual site (way more than tag-sets), where posted questions are posted automatically (based on content analysis and used tags) to SO or CS.SE, displayed reputation is based on posts having OS tag, etc.

Now going back to the topic of this discussion:

I am for using the existing sites as much as possible, so normally I would be opposed to another CS site. However, it has been over a year now and it looks like that there is an agreement that we haven't attracted the necessary group of OS experts for a healthy OS Q&A. I think we all can agree that bring in OS experts is a good thing. If CS.SE can do it, very nice, if not and OS.SE can do so, it is still nice.

The danger is that OS.SE will not build such a community for OS and will not bring in new OS researchers but will become a location for moving OS questions from various existing sites. A good test might be checking once by one the people who have committed to OS.SE who are OS experts. If the proposal is not lead by OS experts by others then there is a good chance that it will be the later case not the former one. It would not be a good thing. If the proposal is lead by OS experts then it has a chance of attracting more OS experts.

My suggestion would be to give CS.SE say 3 more months to see if it can reach the required mass for a healthy OS Q&A. If it fails, then I think it is reasonable to try other ways of building a community for OS. OS.SE can be one. So give it a chance to build an OS community within a limited time frame (say 6 months or 1 year). State it clearly that the site needs to attract significant number of new OS experts, not just be a place for existing SE users to put their OS questions. If it fails, then they cannot complain about its closure. The questions can be migrated to suitable existing sites. The benefit in this case can be attracting some new OS experts to these existing sites.

$\endgroup$
  • $\begingroup$ We (cs.se) may have about 3 months to attract OS researchers and experts in any case. The OS proposal is only about half way through commitment, and currently getting approx 1 new commit a day. Even if SE staff decide to let the proposal move forward it won't go into private beta until it is fully committed (200 people). $\endgroup$ – Wandering Logic May 24 '14 at 21:48
  • 1
    $\begingroup$ Three months? I don't think we can attract more people in three months than we have in two years -- without a concrete plan of action. I have not seen one, so I don't think activity on OS (or any other applied topic) will suddenly spike. $\endgroup$ – Raphael May 27 '14 at 21:08
3
$\begingroup$

I'm going to go ahead and throw my hat into this discussion, as an avid SO user and a supporter of the OS.SE proposal.

First of all, I've been a programmer for a pretty long time. I know my way around most languages and paradigms, and I've worked professionally in several different contexts. That being said, I (like many people in the software development field) have virtually no formal training when it comes to what would be considered undergraduate-level computer science topics.

The net effect of this is that, to me, CS.SE looks scary and CSTheory.SE is basically unintelligible. If I was someone who came to either of those sites cold and looked through a few pages worth of questions, I would deduce that neither of these sites would be the right place to post an OS development question.

With that being said, the remaining SE site on which to post such a question would obviously be SO...

...but there the good OS questions get lost in the noise, good questions about OS and distributed system theory will probably get closed as off topic, and, well, SO is so large and diverse that it is difficult for any subcommunity to form and grow there.

Wandering Logic hit the nail on the head on that one.

SO is a great community if you're writing a .NET, Java, C++, Python, or web app. It's a great community when you're working with preexisting frameworks, protocols, and formats. It's even a great community when you're trying to do something unique by branching off of the already-known components. The problem is that SO attracts few subject matter experts on topics that aren't the most popular amongst the software development community—which is okay, because there's often an equally low number of people using the more rare languages and frameworks, and those people tend to gravitate toward each other and subject-matter specific sites for such questions because the "critical mass" isn't there to serve them on SO. That is the gap between SO and the proposed OS.SE.

What I see as the next most important question becomes "can we build a community of OS developers around a new site?"

The danger is that OS.SE will not build such a community for OS and will not bring in new OS researchers but will become a location for moving OS questions from various existing sites.

Yes, but first consider that OS developers already comprise a community that exists today. I cite the consistent edit activity on OSDev's wiki, consistent discussion activity on OSDev's forums, and a subreddit with a decent number of followers for such a niche area. The reason the wiki and the forums are so active is that they're comprised of a group of people that are doing more or less the same things over and over again—which is why Stack Overflow became such a success among programmers.

The question in my mind is not whether OS.SE can build a community from the ground up, it's whether we can attract the community that already exists and help it grow further. I believe it can. We just have to bootstrap and market it correctly.

$\endgroup$
  • 2
    $\begingroup$ That CSTheory.SE is unintelligible is unsurprising, it's highly specialized. Why does CS.SE scare you? This is definitely something we need to address. This isn't about OSes, and I definitely don't want to fragment CS.SE into one site per domain of applied CS. CS.SE welcomes anyone who has questions about computer science, theoretical and applied. As the moderator who's working in industry, I've paid special attention to see that applied CS questions were not rejected by our more theoretical majority. Have I failed? What are we doing wrong? $\endgroup$ – Gilles 'SO- stop being evil' May 29 '14 at 22:11
  • $\begingroup$ @Gilles you've done nothing wrong. It's just that CS (formally speaking) is outside both my wheelhouse and my current domain of knowledge. Of the top 36 tags on CS.SE, I only know the broad definition of eight of them, I consider myself to have "entry-level" knowledge of four of them, and the rest are basically gibberish to me. I compare this to SO, where of the top 36 I know what literally all of them are, I've used 26 of them professionally, and I'm an expert in at least four. $\endgroup$ – Adam Maras May 29 '14 at 22:20
  • $\begingroup$ @Gilles don't get me wrong, I recognize the value of CS.SE to those who seek it out or come across it in their travels. But the site just doesn't have the more technical "boots-on-the-ground" tone that SO does, or that OS.SE would (in my opinion, based on the committed users). $\endgroup$ – Adam Maras May 29 '14 at 22:24
  • 1
    $\begingroup$ Adam, are you familiar with the tag system and the way you can use it to filter your view of CS.SE or SO? For instance, take a look at cs.stackexchange.com/questions/tagged/operating-systems and stackoverflow.com/questions/tagged/operating-system. Have you considered bookmarking them and making them your main entrance to CS.SE/SO? Also, are you familiar with how you can "favorite" a specific tag? meta.stackoverflow.com/questions/251970/… These features are obscure and well-hidden, but also very useful. $\endgroup$ – D.W. May 31 '14 at 0:05
  • $\begingroup$ Oh, and see also filters (Tag Sets): blog.stackoverflow.com/2011/04/improved-tag-sets $\endgroup$ – D.W. May 31 '14 at 20:12
  • $\begingroup$ @D.W. I'm familiar with all of the above. They're great features, no doubt, but I think they're orthogonal to the issues I've raised above. $\endgroup$ – Adam Maras Jun 2 '14 at 0:16
  • $\begingroup$ OK, fair enough, Adam! I realize they don't really solve the issues you mentioned. I just it might be a partial step towards mitigating some of them (e.g., the statement that "CS.SE looks scary"), though I realize it has huge limitations and doesn't really address the fundamental issue of it looking scary to newcomers who might be expert in OS-related topics. $\endgroup$ – D.W. Jun 2 '14 at 5:16
0
$\begingroup$

My motivation behind proposing this website started prior to taking an operating systems course. I noticed there people did in fact talk about multithreading, semaphores and other OS-related questions, which was great. However, when I started taking the course and ran into specific questions, it wasn't as convenient as asking programming related questions on SO, or electronics related questions on EE.SE. Instead i'd always debate whether I should ask on SO, or CS.SE, or SU.SE. Just like I've been debating recently whether I should post this machine learning and computer vision algorithm questions on SO.SE or Stats.SE or DSP.SE. I'd either get an answer back when it's too late, or just have it unanswered.

Here's one example of a question I asked over half a year ago.

In later part of the same class, we were suppose to implement a distributed system along a set of computers on a network using an algorithm like Lamport or Maekawa's for synchronization. This was more troubling in the sense of implementing as oppose to dealing with concepts.

The motivation didn't come from setting up page for the mass OS users/developers, but mostly for making it more accessible to the OS experts, when these OS users/developers need help. That way they don't have to jump on several sites and search for OS related problems to solve.

I believe it's worth giving it shot to see how the beta would go. Real world and industrial problems do exist, and people and students could use assistance in these fields. Hoping experts will be equally motivated as they are in SO, SE.EE and other active websites.

$\endgroup$
  • 4
    $\begingroup$ I suggest you spend some serious time thinking about whether you want a focus on building OS's and a community of OS experts, vs a focus on answering questions that students might have when taking an OS class and undergraduate-level questions. If you allow the latter, you might find that you are inundated with exercise problems from OS classes, driving away expert implementors and the others who might be interested in the former. (You could take a look at CS.SE vs CSTheory.SE, for instance.) $\endgroup$ – D.W. May 23 '14 at 22:04
  • $\begingroup$ The purpose of this proposal is as suggested in the area51, for those interested in Operating Systems in General, from gaining knowledge and better understanding of Operating Systems and Real-Time Operating Systems Concepts to Implementation. It's intended for both building OS and dealing with existing concepts. I don't understand the purpose of choosing one or the other. $\endgroup$ – Iancovici May 24 '14 at 1:31
  • $\begingroup$ Yeah, that doesn't clarify for me whether undergraduate-level questions from an undergrad OS course would be welcome or not. (If you try to have it both ways, I suspect it'll probably default to a site that has mostly questions from students taking an undergrad-level OS course and who want help with their course, given how university education works and given the number of students out there. Take a look at the operating-systems tag here for a preview of what you can expect without a clearer focus.) But anyway, it's your site proposal, you can take whatever path you think is wisest. $\endgroup$ – D.W. May 24 '14 at 1:38
  • 3
    $\begingroup$ CS.SE does lack people to answer serious OS questions — but I don't see OS.SE attracting scientists either. Rather than try to build yet another community, it would be more efficient to recruit people to ask and answer on CS.SE. If you know someone who can answer your event counter question, rather than hope that they come to OS.SE, invite them here! $\endgroup$ – Gilles 'SO- stop being evil' May 24 '14 at 7:10
  • $\begingroup$ What does a small site give you that you don't get from a chatroom about OS with domain experts that can point you to the best existing site (which can bring their whole community to bear on your problem)? $\endgroup$ – Raphael May 27 '14 at 21:10

You must log in to answer this question.

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