to the kinds of development that UX needs.
In the commercial world, there is a hierarchy whose basic job is to say "no" to everyone's pet idea. To refuse to adopt an initiative proposed by someone, and instead to allocate their resources, against their will, to the *single* direction that the team has been ordered to take. Good or bad. Because even if bad, a single bad direction properly executed by a sizable team with enough labor to complete it well is better than a thousand bad directions each executed by a single individual or a small handful of individuals who lack the resources to complete it, yet chuck it out there alongside all of the other 999 incomplete bad directions.
But the whole *point* of OSS *is exactly* that if you don't like what everyone else is doing, you can do your own thing. That is the basic philosophy. And that's why Linux UX never improves in the free and open space. Because there is nobody with the authority so say, "No, the product will *not* include that, and you *will* dedicate all of your labor to what it has been decided *will* be included."
So the bazaar happens. But the problem with the bazaar as opposed to the cathedral is that the bazaar is only a single story high. You can't build seriously tall stuff without an organized, managed collective of labor. Surge, you get lots of interesting stuff. But very little of it, if any of it, is epic. It's all the size that one single bazaar shopkeeper can build, to man their own little shop.
The Linux kernel avoided this problem because of the cult of personality (not meant in a bad way, but in the technical sense) surrounding Linus. People defer to him. He decides what's in and out, and he does a reasonable amount of labor allocation even if in an interesting, socially backhanded way that's not common. But it works—he is "in charge" enough in everyone's minds that there ends up being one kernel, with leadership.
Nobody similar has emerged in Linux userspace, and it would seem that Linus-like people are a rare enough phenomenon that it's unlikely that one will emerge at any point before the question is irrelevant. The pent-up demand just isn't there now for good Linux UX, like it was for a sound kernel and high-capability OS that didn't cost a fortune, as it was during the late '80s/early '90s boom. The social mechanics just aren't there to generate it.
The Linux desktop as a really sound piece of tech and UX engineering... will never happen. That era has passed, and the problems have been solved—by other platforms. And Android is a very good counterexample. There *was* enough emerging demand for a mobile operating system that wasn't iOS but that offered the same capabilities, and voila—Android. When there is enough demand, there is space for one shopkeeper at the bazaar to emerge as a champion for the needs of others, and to accumulate sufficient influence by acclamation that a cathedral structure can emerge organically.
The bazaar is merely an incubator of ideas. The cathedrals are the epic and actually useful accomplishments. It takes demand and allegiance-pledging at the bazaar from many attendees to lead in the end to a cathedral. This means that the bazaar has to be big, and that the shopkeeper in question has to have an idea that many, many are not just interested in, but willing to work toward—enough to sacrifice their own autonomy and submit to leadership. This just doesn't exist for desktop Linux any longer. It got close during the height of Windows dominance, but there was never quite enough demand to make it happen organically. And now the time has passed. The desktop Linux people are running little shops at the bazaar that don't get a lot of foot traffic, and nobody is seeking them out. They are the kings of very tiny, forgotten kingdoms without enough labor resources or wealth to even maintain their castles any longer—and as a result, there is nothing but infighting, strange hacks to maintain castles on the cheap, and lots of started-but-never-to-be-finished foundations of castles for historians to pick through (or, more likely, forget).
I predict that Linux will continue to be a significant part of whatever new "booms" in technology happen, so long as Linus is significantly involved in kernel development. But the window for desktop Linux has just plain passed.