Young children are not famously good decision makers, and it's easy to imagine them making very limiting choices that they would later come to regret.
Plunging young children into an anarchical learning environment with nothing but YouTube to guide them won't end well. But if you add guidance that encourages their strengths and tolerates their weaknesses, I speculate that net learning will exceed that of today's school system.
All sorts of approaches are possible with 5 students per teacher, but not 10, and with 10, but not with 20—let alone 25 or more.
Which means that a wide roll-out would require one or both of: more spending on training; more spending on salaries to increase the quality of the candidate pool.
A top-quintile teacher may be way more effective than a middle-quintile teacher, to the point that entire education frameworks and strategies are viable for them that aren't for the middle quintile, but the trouble is only 20% of your teachers are top-quintile. Ditto admin. You'd be amazed at how mediocre admin can completely ruin a good plan.
I wouldn't be so quick to attribute those two events, which happen around the same time, as causal. I know a few teachers of young children and the things they love about their jobs is just that curiosity you're talking about. Unless you think it just takes a few years for the system to really beat them down, it's not exactly synchronous.
When thinking back on sitting through all the boring classes on subjects I wasn't interested in feels like it was a lot more of a me "problem". I find most of that stuff interesting now or at least appreciate that I had the exposure to it. It seems to me that I was much more interested in play and/or social development at the time.
That really depends on the environment a child grew up in. When I see parents who delegate their child rearing to an iPad with Cocomelon on 24/7 I'm quite sure that kid won't benefit from unstructured self-paced anything.