It's most definitely going to be mandatory for L3, and there is the question as to whether L3 will even be allowed to travel at freeway speeds. I also doubt L3 will ever be allowed on cities streets.
The problem with driver monitoring is you can only go so far with it. It's a great way at detecting someone falling asleep, and curbing cell phone usage. In fact I think Europe is going to make it mandatory not just for partial autonomous system, but for manually driven vehicles as well.
With partial autonomation there is only so far driver monitoring can go. if I had a Cadillac Supercruise vehicle I would almost certainly stare off into the void in front of me without really paying much attention at all.
I think L2 is mostly just a rigged game against human drivers. Where the car does the driving, but we take the fall. The manufacture of the L2 system doesn't have to take responsibility, and this is why they're mostly garbage.
The existing FSD on cities streets is light years ahead of any of the L2 systems on the market including Tesla's own NoA which is kinda hilarious since a person with the FSD beta will have a system that works great on cities streets, but is still plagued with the same NoA issues that have been reported people using NoA like myself). Now that likely won't be the case for much longer, and I look forwards to seeing NoA leap in capability.
My fear is that FSD will be so good that it kills itself off because it doesn't have anywhere to go. It can't leap from L2 to L4 because L2 gets so good that drivers get complacent, and edge cases kills a few people. The media erupts in anger, and the whole thing is shut down.
Or the FSD beta gets stuck in a limited release for a really long time (years) so it simply doesn't have the mass involvement of hundreds of thousands of vehicles. But, this is the only way to avoid bad publicity because the odds of a string of deaths happening is much lower.
Even if it does somehow achieve FSD at non-geofenced, but still L4 (due to weather limitations) system it might never be allowed to turn on due to the lack of sensor redundancy.
So to me the destination of FSD isn't anything less than the creation of the destination itself. Or in other words the end game for FSD is not to do autonomous driving, but to create the possibility of FSD.
It's basically the sacrifice needed to determine what exactly the public will be okay with. It's forcing the public to realize this is here, and its their move.
