There were a bunch of these during the netbook boom and they were grandparent-proof. Sadly, all of my favorites (Moblin/Meego, Joli OS and xpud) are long dead.
Netbook boom happened and then people actually started trying to use them.
That is why the netbook boom died. Because while they promised to make cheap Linux devices that anybody could use none of them actually delivered on it.
In fact the whole thing was a fiasco.
The classic Linux problem can be described as "9 clicks to shit". Meaning that a Linux desktop looks, on the initial viewing, as something that is actually usable and modern, but once you start clicking around things start going bad. Going bad quickly.
A major problem cause of the problem is "linux is about choice" mentality.
Another problem is that programmers specialize at being good at programming, but things like documentation and UI testing are their own disciplines that are separate and distinct in a lot of meaningful ways. So, in attempt to make up for this, they leave configuration and details up to user choice. It is effectively pawning off the last stages of development onto the end user, who are generally most ill-equipped to make technical decisions on software they are not familiar with.
Instead of presenting something that "just works" users are presented with a myriad of choices and options that they have to make decisions about before ever actually being able to use the software.
Having lots of choices in applications, widget libraries, desktops, ways to install software, init systems, etc etc... results in very significant complexity as it all has to work together.
And complexity breeds bugs.
So each and every user experience ends up being its own unique things. There is no "standard configuration" no "supported configuration" and no "documented configuration".
The end result is that each user is forced to find ways to make the desktop work for their specific use case. Essentially finding a magic combination of things that isn't broken for what they specifically want. The desktop will still be full of buggy behavior, but just not in the way that that particular user cares about.
This creates a extraordinary of frustration and friction when time comes to doing the things that the desktop is actually intended for... which is getting actual work done.
This isn't a problem for a certain class of highly technical users that love technical minutia and configuring things. Knowing how to carve a working system out of a OS riddled with historical artifacts and highly technical choices is a badge of honor for many people.
But this isn't how most people want to do things.
Netbook boom tried to fix this in certain ways, but all that ended up happening is that each little corporation tried to do their own thing and then abandoned it when they realized there was no money to be made and they really lacked the resources and expertise to make it work.
Can't you say most of these things about Windows or even any other useful desktop OSes?
If you want a computer that does useful things and isn't just a locked down social media consumption machine, you end up with a lot of that baggage you've mentioned. Sure MacOS and Windows fare better (insert trillion dollar companies backing it spiel here...) but they are no smooth ride if you are doing anything more than consuming stuff or using a web browser.
I agree, it feels like linux is on a perpetual "under construction" flow, except that lately development has been picking up speed (which is a good thing), but results in an even "bumpier" ride.
And you can't really retreat to a more stable distro (Ubuntu LTS, numbered debian release etc) because you often get stuck with "outdated" software, missing hardware support and missing features (might be essential to your use case). Chasing the bleeding edge ends up being pretty essential in desktop linux but as the name entails it ends in lots of pain for the user. Nerds, hackers, developers and maybe even some power users can definitely have fun and thrive in that environment but the casual consumer.. yikes :(
> Another problem is that programmers specialize at being good at programming, but things like documentation and UI testing are their own disciplines that are separate and distinct in a lot of meaningful ways.
Even having decent UIs is a problem for many Linux apps and many are often deeply unpleasant to use day-to-day. This is one thing that has started to slowly get better in the last decade or so, but I can always tell what applications were designed by programmers and which ones have had at least some UI work done on them.
Also, accepting feedback from users on UI improvements often gets either ignored or de-prioritized in favor of adding new features. It's very frustrating to see an otherwise really fantastic application with a lot of neat functionality hamstrung by a bad UI.
The only desktop to have real meaningful large-ish/professional usability testing is Gnome. And that is only a couple times.
The first time was financed by Sun Microsystems in 2001 for Gnome 1.x and the result was Gnome 2.
The second time was financed by Novell around 2005 or so for their attempt to compete with Microsoft with Novell Linux Desktop. Unfortunately for them the real beneficiary of the improvements that results from their work was Canonical's Ubuntu.
Since then there have been numerous smaller/informal/ad-hoc attempts for both KDE and Gnome.
The results of all of this, of course, is that many "Linux users" believe that Gnome is the result of a conspiracy between IBM, Redhat, and maybe even Microsoft to "destroy the Linux desktop".
So that is fun.
The real success story, from what I can tell, is Blender.
They successfully revamped their user interface without a huge budget. Although it was still financed somewhat by some EU initiative, IIRC. They accomplished this by getting developers sat down next in a big room to actual 3D artists working together to produce a animated feature.
By physically placing users next to devs and having them work together, likely with a great deal of humility and openness, they managed to transform their UI into something that was actually decent.
This is probably a model that can be duplicated by other open source developers, although finding the right type of technical users not experienced in using said software willing to participate is going to always be a major challenge.
