Re: Of course we all know the permanent fix
Many (myself included) have dabbled occasionally and given up, remembering the last attempt. I deployed a Pi with emulation software because I could get an image, but only because of that.
Before it, my previous memory was a disastrous attempt to deploy a box that would simply load and repeat a video. I had to pull packages by command line to roll my own. Atop the "4,000 text editors" type conundrums, to get a simple video playing on a bare-metal fresh install required 21 packages, a moment seared in to my memory, all of which as a newcomer I had to find and choose whilst navigating a new command structure and methodologies.
How to guides (better nowadays) were either missing important caveats or kept touching on smug holier-than-thou condescension. By the time you'd figured it was missing something, you were 2/3 through with no way to u-turn, noting no other guides seemed to follow the same route. The next guide would smugly chastise anyone who chose the XXXX that the first guide had you install, and idiots that thought the right way was the order the first guide had used.
After a day or two of wrangling, I finally got my video playing...to realise there was no sound. A cursory glance showed a similar process would be needed to build my own sound card support. I binned the box and abandoned the project.
Linux maintains an issue with infighting, and whilst it may be better now, my past experiences sour any desire to try again. If I do get "it" built and deployed, I'm facing either: updating it which often seems insurmountably complex (finding a way to strip back layers and redeploy); or the likelihood of a project being abandoned and having to rebuild with a new approach of supporting projects around the new component and go through the whole mess again.
I raise these issues and someone says "it's not like that now" or "did you know you can use SqueakyTrimCheeseSwitch variant that does all this" which is *another* fork in the vein of XKCD 927 by one person who will stop updating in 9mths.
Reality is, I (and many others) just want stuff to "work" within reason. I don't want to compile my own drivers in an enterprise environment in the same way I don't want to build a kiln to forge my own blades for my hedge trimmer so I can cut my hedge, although it would be useful so I can also build a pottery wheel to throw and fire a cat-bowl to feed my cats. I'd rather not learn cartography so I can plan my drive to my next business meeting and understand the intricacies of road construction and tyre compounds in order to get there. Some people do, I'm happy for them. But in the modern world I don't have time to "keep an up to date comprehension of the botanist impact of the rainfall and soil composition, and the home life of those who pick the beans and their continued battle against marmosets, in the Sumatra along with the geopolitical impact on the trade routes and the financial management of a global FMCG organisation in order to have a morning coffee" for every task I undertake.
Until Linux can reliably and predictably match the ongoing stability and relatively straightforward deployment of Apple or Microsoft, it's going to always be "iTs tHE tImE TO sWAp To LinUX" shouted loudest by people who have time and interest, or those who refer to everyone else as "sheeple"