The FANGs employ a lot of engineers to work on open source — and so do many other companies, from IBM to Walmart.
That's enormous corporate subsidy of open source development. The opportunities today to get paid for writing free software are amazing compared to twenty years ago.
Corporate open source is often an attempt to take or crush competition rather than give.
When's the last time you've heard anything credible about Linux on the Desktop? IBM doesn't care about it. Google doesn't care about it.
Android is a zombie operating system that keeps Apple from having a viable competitor. Companies will block bot access via http and then hire a PR agency to congratulate them for providing an API that lets you access 10% of what you can do with the web site with much more complex code. (eg. this is the authentication process)
Read between the lines and you'll see that Google is pushing Tensorflow because they'd rather you upload your data to their cloud to run on their processors and they'd rather you not take advantage of the latest NVIDIA drivers because they hate the idea that you'd buy NVIDIA chips or own your own computer for that matter.
It also is not fair to say that open source development is "subsidized", if anything it is the exact opposite. Open source software creates billions and billions of dollars of value that is not being captured by the people who create it. Open source software subsidizes IBM, Google, ...
Why do people do it?
It might seem irrational, it certainly is from a dollar and cents perspective, but I think a lot of people reach the point where they realize that working at a place like Jane Street makes you a Negative Net Productivity Programmer no matter how hard or how smart you work and that if you want to be anything more than an 0.1x developer you have to get out from under the thumb of the "cult of product" and the scum-sucking lying rat bastards who sell enterprise software.
Thus you have people like Wes who should be billionaires (or at least tenured) in terms of the value they give to society, and then you have "useful idiots" like Sheryl Steinberg who are allegedly intelligent, but would have run the first moment they heard anything about George Soros but no... She should have been fired for that, but no, we'll just have to wait another 10 years for Facebook to become the next Yahoo and get bought by some phone company like Frontier.
I think the absolute ridiculousness of the parent's comment really shines through when you consider just how massive Google and IBM's contributions are to the Linux kernel.
He has a few good points in there. The thing about the desktop is one of them. Sure, Google and IBM have contributed to the kernel, but that's because they use the Linux kernel for things very important to their businesses, and it's to their advantage to mainstream their changes instead of maintaining entirely separate kernel trees. However, they don't contribute to the desktop at all, nor does almost anyone else. For Google, it isn't important to their business strategy because they're interested in Android, which isn't a desktop OS and doesn't use the desktop parts of the system, only the kernel. IBM uses Linux on servers/mainframes, so again they don't care about the desktop. However, at both, I wouldn't be surprised to find that many, many developers do use desktop Linux (maybe in a VM) to get their work done, so it would actually help these companies, and many more, if they contributed to desktop Linux as well, but they don't, because their management doesn't see that part.
Looking over the long arc too I am not impressed with the development of Linux.
I started using Linux in 1993 and back then it was way better than Windows. Bill Gates donated money to my uni to try to kill off the Unix culture, but the only people who would use Win NT 4 to do scientific work was the one guy who loved Windows and me, who would vnc from Windows to Linux so I didn't have to fight for one of the few machines that were running Linux so grad students could get their work done.
On single-processor systems Linux was OK up to 2.4 but 2.4 did not really work on SMP machines. I would discover the strangest kernel bugs and the more I looked at it I realized these couldn't be fixed so I'd write patches that would cause it to barrel on despite corrupted data structures and wait for 2.6 to be ready.
The problems w/ 2.4 were covered up and dismissed much like the way Microsofties will tell you that all of those Windows NT 4 crashes were the fault of your hardware.
2.6 got the RCU stuff from IBM which meant Linux actually worked right on SMP; that was a real contribution both technically and in terms of legal protection from Novell.
Since then the major itches people seem to be scratching Linux have been:
"ext4 doesn't corrupt data often enough, we need to invent a new filesystem that corrupts data more"
"xpoll and ypoll don't work right so let's add a new zpoll that will work incorrectly in a different way"
I was a Linux enthusiast back in the day, I would compile the new kernel each time it came out, Alan Cox would bitch me out for filing bug reports that gcc would segfault when I compiled the kernel on my overclocked machine.
What I've seen is that Windows had advanced by leaps and bounds since Windows 95 and Linux has been stagnant. It's a boring and reliable operating system for servers, which is a good way, but it is not an operating system driven by enthusiasts anymore.
This seems a little over-the-top, but I do have to agree that in many ways, Linux just isn't driven by enthusiasts any more. It just doesn't have the excitement it had back in the late 90s and early 00s that I remember, and the desktop environments in particular have really stagnated, which I mostly blame on GNOME.
Of course, but those opportunities within those companies are few and far between comparatively. Engineers that get to work on OSS full time probably represent less than a fraction of a percent of their engineering workforce. I think Wes was talking more about places where more/most developers work on OSS full time, which was what I was talking about. The vast majority of code produced by FANG is closed and proprietary.
That's enormous corporate subsidy of open source development. The opportunities today to get paid for writing free software are amazing compared to twenty years ago.