Well, I was basing my assessment of ‘not-ready-for-prime-time’ on more than just the complaints from users on this particular thread. The lack of consistent reliability of FSD has been well documented elsewhere in these forums and in information available from other sources.
I am willing to discount the complaints to some extent due to the persistent anti-Tesla bias in much of the media and Web, where anything that goes wrong on a Tesla is unfairly magnified and over-reported. But that does not entirely explain the volume and nature of the problems reported.
Even putting aside city driving element, I consider Tesla’s own assessment of FSD capabilities to be an admission that it is simply not ready: “Autopilot and Full Self-Driving Capability are intended for use with a fully attentive driver, who has their hands on the wheel and is prepared to take over at any moment” (from Tesla's website:
Autopilot and Full Self-Driving Capability). In my view, this level of capability represents an undesirable and fundamentally unsafe mid-way point to actual FSD. The admonition that the driver must remain alert, and be ready to take over at any time, fails to recognize that the very nature of FSD is to render the driver less than totally alert and ready to take over. That is because the driver is no longer engaged in the moment-by-moment exercise of driving. Alertness to potential problems will tend to be less and reaction times will tend to be longer. I have real misgivings about the current version of Tesla FSD from that viewpoint.
In my view, actual ‘ready-for-prime-time’ FSD would involve (at least):
1. A rate of unreliability and inaccuracy in day-to-day driving situations much lower than currently reported. FSD that works ‘much of the time’ is not good enough. It has to work, and be at least as good as human driving, pretty much all of the time.
2. Point #1 would also have to extend beyond day-to-day situations and be true also of manoeuvres to respond to sudden emergencies. Since the driver is almost inevitably not going to be as fully engaged as they would be if they were actually driving, FSD has to be at least as good as an actually-engaged human in avoiding accidents due to a suddenly arising situation.
3. An predictive ability, based on continuous assessment of the nature of the road ahead, traffic reports, weather, road conditions, light levels, and other inputs of when it will be better for the driver to take over. The driver, who (again) will almost always not be fully engaged, should be given significant notice ahead of time and not be expected to take over in a split-second. I would not expect FSD to able to be appropriate in all situations, but FSD should be able to recognize ahead of time when it is not.