They are not. That would mean that more than 10% of the people sitting on chairs in companies are doing software development. That would fail to take into account jobs that don't have a chair, every PC outside an office and even non technical jobs in sw development companies.
Let's check some facts.
http://www.bls.gov/ooh/computer-and-information-technology/s...
Number of jobs: 1,114,000
http://data.bls.gov/timeseries/CES0000000001
All employees, end of 2014: 140,402,000
It's less than 1%, still much more than I would bet on. However I think that this figure is lower globally and Windows is sold globally. Hence, much less than 1% of PC users are software developers (Windows, Mac, Linux, anything) and having Windows the larger share its stats are close to the global figure.
Did you find any data relevant to that point?
I tend to think that it's true. There are more developers on Windows. Way more.
OSX has 26.2, Linux 21.7.
They sum to 100.1. There should be some inconsistent rounding somewhere.
If the figures are representative of reality it's true that Windows is less popular among developers than it is among the general public.
If 1% of the desktop users are developers, that's the same as 50% of the Linux users, and a really big share of the Linux users are developers (really, I expected it to be more than 50% what, clearly isn't true). That's also circa 12% of the Mac users, and I bet a biggest share of them are developers than the Windows ones.
That seems unlikely.
The fact that there are a number of programs without real non-Windows equivalents (games, emulators, various commercial packages) probably does suggest a lot of people doing Windows development.
It does certainly not mean that there are no Windows developer. That would just be repeating the same flawed argument, but twisted on the direction where it's the most flawed.
The people here on HN live in a bubble.
People are using Macs because it gives them a preconfigured and stable Linux like environment, almost 100% compatible with the server. Windows tend to have different toolings and different quirks. It doubles the effort. In my experience it's OK if you're using Java, which mostly hides the OS and it's obviously a must if you go with .NET. Basically only large companies use those two technologies, and not all of them. Everybody else is on scripting languages, JS, PHP, Python, Ruby. Btw, Java is almost always deployed to Linux too.
So, it could be a bubble but it's a large one.
Your overall point stands on its own but are you really trying to argue that 1 in 10 PCs are software development workstations?
Also, you should not expect that having a 90% market share overall will translate into 90% market share among any particular subset of users.
Personally I'm very happy with the hardware options available to me on the Mac, but there's a whole world of hardware options over there in PC land I'll never take advantage of. Conversely Linux graphics card driver support is still woeful, but that's really not Microsoft's problem and if Linux didn't have the PC platform to run on it's hardware costs would be much higher. What would be the alternative, Sun-style workstations? It's not that Microsoft did this on purpose, but it's still a fact.
However network effects would have given to another company the position that MicroSoft enjoyed for some 30 years. It could have been a slower growth or a faster downfall, but somebody would have been there. Not Apple because they are interested only in selling their own hardware. Probably some company out of nowhere, not necessarily on a x86 chip. Then everybody would have started building drivers for the OS of that company and Linux and MacOS would be running on the preferred chip of that OS.
Maybe in an alternate reality Visicalc built their own OS to run their spreasheet on a large screen (by the time standards) 8086 computer and replace MSDOS and Microsoft as the OS and sw company of choice. We'd be talking about VisiOffice and Vindows now :-)
We could have gone the slightly-less-compatible route with CP/M-86. Network effects would make the industry gravitate towards a standard platform, but not necessarily a company.
I remember tons of GUI software running on proprietary Unixes on vastly different architectures, all written aiming POSIX compatibility and an X graphics environment.
Yes, CP/M-86 could have played to role of MSDOS but people really needed to buy a box with a floppy disk to install from or a preinstalled OS on their new PC. MS did that right.
All that is not very relevant. Had Microsoft never existed, it'd have been an ecosystem too different to make any valid prediction. We'll have to eventually restore the universe and run a different path to see.
Just like it happens with any standard, we ended up with lots of #ifdef, because many APIs might had the same name, but not the same semantics.
And like in UNIX like OSes, all the cool goodies weren't in the POSIX APIs anyway.
It is quite sad that despite some of the cool GUIs like on NeXT and Sun OpenWindows, the best that they could settle on as a standard was Motif.
Well, we did have Richard M. Stallman, Emacs author, not only create an OS (GNU), but also extend Emacs to the point that it might as well be an OS. More pointedly, Miguel De Icaza created Gnumeric (a fantastic Excel clone) before he started the GNU Network Object Model Environment (GNOME).
Corel attempted something similar to what you were saying, as well; Their Wordperfect/Office/Draw empire was leveraged against Microsoft's Embrace/Extend/Extinguish in order to promote their (debian-based) linux distribution.
As for free drivers, the only one lagging behind is Nouveau for Nvidia users. AMD's free/open driver has made some serious performance improvements in the past few months, and Intel's driver (or just plain modesetting) has performance parity with their windows driver.
I remember 2008, and fighting with AMD/ATI drivers, and occasional reboots on my NVidia systems that led to X not working anymore. I've had exactly one instance of that in the past 4 years (under Arch). Can we stop spreading 2008-era misinformation in 2016?
Because Microsoft has the largest market share and porting drivers from Windows to Unix systems is basically a full rewrite, which doesn't happen?
My personal experience with OpenGL is that all Valve games run faster on it than on DirectX (both on Windows, but I heard Linux + OpenGL also has better perf than Windows + DirectX), but that's about the extent of it.
So how is Vulkan support on Windows these days?
Windows is still 90% of the PC market. If only 10% of those users are developers, that means there are more Windows developers than all macOS users combined. There are a lot of us out here, and I think we're starting to get fed up with being treated like we don't exist, just because of the shame that gets lumped on us from a small cohort of other developers.