I think about this problem often. I think it's caused by economies of scale and is an escapee of Pandora's Box that we're not going to be able to put back. Instead, we need to think carefully about what we build and how we build it - educate consumers, regulate to price in obsolescence and recycling costs, and place greater value on long-term quality as a society - to prevent this from growing even worse.
I build industrial automation equipment, so I'm in part to blame for the problem, however, I'm also in a position to advocate for easier repair processes and more sensible assembly stackups. It's simply a fact that the six figures in initial equipment capital divided by, hopefully, a similar number of unit sales, plus the few dozen seconds of operator hands-on time per unit, always going to produce an end product which costs less than hours of diagnosis and maintenance by a local, high-skill repairman. The math simply doesn't add permit any other arrangement, no matter how standardized your tooling is, how open your repair manual is, or how efficient your aftermarket parts ecosystem is.
Repairing and maintaining something for more money than it cost to produce it puts those activities firmly in the realm of luxury goods. The better we get at producing useful things cheaply, the more we have to multiply the value of human labor, and the less it makes sense to repair cheap things with this expensive labor.
One interesting option today is that you can get an electronic watch and get access to all the internals as well:
https://open-smartwatch.github.io/
However, its operation depends on electrons and bits instead of springs and escapements, and you can see those bits moving about in PlatformIO code instead of under a magnifier.