Fragmentation is a red herring
I disagree that fragmentation is really "the problem", if problem there is. And ironically, the article's text sort of makes that point for me.
Let's consider some of its arguments:
"We have many excellent Linux desktop distros, which means none of them can gain enough market share to make any real dent in the overall market."
OK, sure, but then...if you add up the market share of *all* of them, it's still not much.
It's also worth noting there have been points in the history of Linux on the desktop where 'fragmentation' appeared much less significant. For a few years after Ubuntu emerged, love it or hate it, it was clearly and by a long chalk both the most popular and the most buzzy desktop Linux. That lasted for quite some time. But even still, its market share never got very high, and Canonical never made any money out of it.
"None of the major Linux distributors – Canonical, Red Hat, and SUSE – really care about the Linux desktop. Sure, they have them. They're also major desktop influencers. But their cash comes from servers, containers, the cloud, and the Internet of Things (IoT). The desktop? Please. We should just be glad they spend as many resources as they do on them."
This is true, and it's a good point, but *it has nothing to do with fragmentation*, does it? It would be true even if there were only one commercial Linux distributor. They still wouldn't be able to make much money off the end-user desktop.
And in the end, that's the real problem: money. Exactly as you say, nobody has ever figured out the magic way to make money off desktop Linux (with the possible exception of ChromeOS; I dunno how the financials on that look, honestly, not sure if Google breaks it out). At this point, given how many people have tried, that probably means it isn't going to happen. And without a large and self-replenishing pile of money, shifting a behemoth like Windows is just not realistically going to happen.
There are, rough guess (I'm too lazy to go look up the commit logs), probably less than 50 people in the world who are paid to work, full-time, on a *nix desktop environment (again excepting ChromeOS). That's *all* of them put together. So is fragmentation really the issue? Even if you got rid of all the desktops but one and put them all to work on that, 50 full-time devs is still not a large number. How many full-time devs do you think there are working on the Windows or macOS shells? It's a hell of a lot more than that.
Ditto for the F/OSS graphics stack, only if anything it's probably worse. Ditto audio. Ditto bluetooth. Ditto, let's be honest here, basically any part of the OS that doesn't make Red Hat, Google, Facebook or (a bit less significantly, since they have much less in the way of resources) SUSE or Canonical some money.
Given how few resources we (I'm the Fedora QA lead) have for this stuff, honestly, it's pretty amazing - and a tribute to the folks involved and the community volunteers who help - that Linux on the desktop works as well as it does. If you gave Microsoft or Apple the resources we have they probably wouldn't cut another release for a decade.
That's the real issue, not fragmentation. Fragmentation is, if anything, more of a consequence. If you think about it, when it comes to the areas where Linux actually is a serious, money-generating player - in the cloud, on-prem in the enterprise, Android, maybe ChromeOS - there is *much less* effective fragmentation in those areas. Where there's money to be made, a dominant player or a handful of them somehow emerge naturally. There's one serious player in Linux phones - Google with Android. ChromeOS is also all Google. There's a handful of serious players in the enterprise and the cloud - Red Hat, SUSE, Canonical, plus Debian for non-commercial cases. There aren't dozens and dozens of serious enterprise distros. If there was real money in the Linux desktop, the same pattern would likely emerge.