Barring some radical breakthrough, progress usually follow an asymptote. Waymo has reached their asymptote after perfecting their autonomous driving for over 10 years, and is still not confident enough to launch it even in their spherical-cow geofence.
The real world is full of edge cases - cameras fogging up, proximity sensors confused by ice buildup, thick fog limiting visibility, sun low on the horizon blinding the cameras, badly marked construction sites, black ice, line markers under snow, potholes, slush, contradictory traffic signs, deep puddles, worn out ruts, suicidal wildlife, road debris, etc. etc.
The full self driving product that people expect isn’t a beta product that’s only available in a special region. To be accessible as running water, decades away is probably accurate estimation.
In the mean time we can all drive around in our self driving cars while we wait for them to become a reality? It doesn't make any sense. Either they exist, albeit in a limited form, or they don't. Does it get you from a to b in most normal daily use cases? If yes, then it counts.
I would apply the 90/90 rule here [0]. It looks ready, but I don't think their safety is up to scratch yet and I'm almost certain the cars aren't safe in adverse conditions.
The real world is full of edge cases - cameras fogging up, proximity sensors confused by ice buildup, thick fog limiting visibility, sun low on the horizon blinding the cameras, badly marked construction sites, black ice, line markers under snow, potholes, slush, contradictory traffic signs, deep puddles, worn out ruts, suicidal wildlife, road debris, etc. etc.