This is probably skewed. We did our best to make sure we were getting a wide swath of respondents by also running the survey on StackOverflow, but even there, our choice of which tags to advertise the survey with probably biased the results. I'd take the OS distribution not as an indication of who in general developers for which OS, but rather just as information on which percentage of respondents were on which OS. I.e., it tells you about the survey, not about developers in general.

Stackoverflow itself seemed pretty Windows-centric last I looked at it. Makes sense considering it was launched by two prominent Windows developers.

For some time, when Apple had its goofy NDA on the iPhone SDK when it was first released, Stackoverflow was probably the best site to get CocoaTouch related help. It's still a pretty decent resource for iPhone development.

I'm not as convinced this is terribly skewed away from the normal. Outside the Bay Area, the large organizations that hire thousands of programmers primarily use Windows for development. For example, go to any Defense Contractor (Lockheed, Northrop, etc) in the country and sample the computers they use for dev work. Each developer probably has a Windows box they do everything on. There might be some Linux boxes mixed in, but those will probably be headless. Or the developer runs Linux in a VM on top of Windows.

While in the Bay Area, we might think these are skewed because everywhere we go we see Mac laptops, in reality, the employees at big corporations are going to outnumber us.

I suspect that the survey is skewed in other directions when it comes to large corporations, though. I would expect to see much more Perforce if the big not-primarily-computer-companies were included, because for many years Perforce was pretty much the only game in town for large, multi-million-line codebases. I still can't imagine Lockheed running their sourcebase off Subversion or VSS; even when I interned at a mid-sized (~100 person) government contractor, we used Perforce.

I think the survey is representative of the geographically-distributed, Joel-on-Software reading, micro-ISV crowd, and not much else. Micro-ISVs often develop on Windows (because that's what their founders are used to) and use Subversion (because it's fairly easy to setup and generally adequate for 1-10 devs). I suspect you'd get very different numbers at YCombinator or Google or Lockheed, though.

Another mega-corp data point: 15,000 Engineers plus everyone else uses desktop/laptop windows XP. Linux is used on the supercomputing clusters and on test equipment in the factory. IT controls the desktop domains, while the engineers control the clusters and test equipment.

I can confirm this (working at mega-corp) and can also add that a lot of specialized and embedded tools are primarily hosted on Windows. Even high-end CAD and modelling tools are primarily on Windows where I've worked.

High-end CAD tools used to be run almost entirely on Unix systems (Solaris, SGI, HP/UX, etc.). Back in the 1990s Windows systems couldn't handle these tools the way Unix workstations could, but once that changed there was no reason to spend far more money on a workstation. Perhaps if Linux had become popular a few years sooner the transition in high-end CAD would've been from proprietary Unix to Linux. See for instance http://www.ptc.com/partners/hardware/current/support.htm where support for various *nix are being phased out by Parametric Technology, Linux included.

Yeah, I remember those days well, we had Unix boxes for CAD and 386s with Win3 for office work side-by-side. I think you're right - the transition to Linux would have been an obvious choice if the timing had been better.

The article said: "We asked subscribers to our developer newsletter (788 respondents) and professional developers via StackOverflow (169 respondents)"

I imagine that the FogCreek developer newsletter is going to skew reasonably heavily toward Windows developers.

Worked for three different (non-startup) companies in the last 2 years, never left Windows XP. I'm thinking it's normal for the developer industry.

I asked why at my first job interview. Stated reason was that as our end users use XP (standard at their company), we should use XP. Makes sense.

I guess the cost of mac vs pc explains why macs aren't more popular in regular companies.

As for Linux, well, I think it's just about it being easier to buy PCs with windows than with Linux preinstalled.

If your time is cheap and your IT department is free, Windows on a blowout-priced Dell is a bargain.

If there's one thing that's killing corporations it's this narrow focus on acquisition costs and a near total disregard for support costs. People moan about Apple being "too expensive" all the time, but the reality is most computers are too cheap.

I hope one of the major vendors develops some alternative to Windows. HP is shovelling billions into Microsoft's pockets with nothing to show for it.

In my experience, with modern versions of windows, support is a wash for technical users.

I used a windows 2k3/2k8 machine every day for 3.5 years at the same company developing software, and never had a support issue with it. On the other hand, I solely use Macs at home (and, since leaving the aforementioned job to work on my own company), and have also never had a support issue.

However, my grandmother has recently gotten ahold of a vista laptop and a MacBook laptop. She has really shown a greater affinity to the Mac.

Here at HP we're using our own equipment on a fairly large installed base so we actually have a reasonably good feel for relatie support costs.

Also, we have an internally supported and externally available standard Linux environment (LinuxCOE) which engineers (like myself), who run multiple OSes in their cube, can use.

Speaking for myself only here, I would say that the alternatives are available but we can't really ignore what the market demands.

Windows is also a "bargain" when it comes to IT costs. Windows has an excellent track record of strong enterprise support, whereas I have heard from an IT dept that there are all sorts of headaches (and compatibility-breaking changes) to Mac operating systems.

IT departments are also notoriously afraid of what they don’t already know.

Care to explain why a huge swath of corporate IT departments are still running XP and IE6 (or 7) then?

Oddly enough, HP is the one of the major vendors that most appears to be actually planning an alternative to windows. They've already said they plan on bringing WebOS to PCs.

Never underestimate a company's reluctance to change. Usually, the bigger they are (or the more clueless the decision-makers are), the greater the reluctance.

Actually, if you have thousands or tens or thousands of desktops and hundreds of applications something as simple as upgrading Windows (or even Office) becomes quite a serious undertaking - so unless there is a really good reason for doing so it's understandable that organization don't upgrade just for the sake of the latest shiny bit of tech.

This is largely a symptom of Windows itself, not a symptom of needing to upgrade something. It'd be much simpler to push updates to all users if you use a self-configured apt repository than it is to push updates to Windows clients. Not so nightmarey at all to upgrade systems with good package managers, actually.

When I was a sysadmin, I could push my updates via Group Policy. It's not that bad. (I'm now in development so I don't know the Microsoft current recommended method).

The cost of an upgrade is the daunting part IMO. We need to buy a bazillion CALs (licenses) if we want to switch to Windows 2008 server-side, for example.

I don't do Windows (not necessarily by ideology, I just don't do it anywhere I'd learn about IT issues), so this is an honest question: Assuming a decent IT policy that closes off the worst of the security issues in other ways, and which won't change even if you upgrade, and assuming you're not a Windows development shop, what is the big business advantage of moving your company from XP to Windows 7?

1) System stability. If your video driver dies on XP, hello crash. On Win7 you see your screen go blank for about a 1/4 of a second and then redraws itself, back in business.

2) If you work remotely, RDP is way better. I've even streamed video over RDP and forgot that I was streaming it from a remotely located computer.

3) If using SSDs Win7 will have better perf and longer lifetime.

4) Virtual folders.

But I would say that if your machines are completely locked down, for example all XP machines are not on the internet, just some closed internal network, then I think XP might be doable. Once you touch the internet though, all bets off.

We've had to wait longer for cheap computers from places like Dell if ordered with XP than with win7. If you're in a hurry and just need a basic desktop for someone, it's something to consider. At least that's how we ended up with a mix of 2k, XP, and win7.

IMO, Laptops. Supporting XP on recent laptops is an exercise in frustration due to drivers, and in my situation is compounded by the lack of standardization on a small number of models.

At my day job we're making the move from XP to Windows 7 because the large number of cumulative patches to XP have, over time, caused all of the machines to run really really slow. Our ICT department has a really cute name for this OS upgrade project: "CPR" (for "Client Platform Refresh").

My guess is that a lot of that 70% represents the OS these people use professionally at work. I'd be more interested in seeing the split between OS used professionally and OS used personally.

That would be the corporate world. Not only web startups need programmers ;)

In my neck-of-the-wood it is probably higher, more like 85-90% Microsoft. Programmers use Windows in the Enterprise, Banking, Insurance, old media, etc. Startups and Web-dev are on other platforms, but they are in the minority.

I've used Windows at work for my last two gigs. At home I use OSX and Linux (both desktop and server). My wife and kids use Windows (what can I say? She's the boss).