> your historical observations are irrelevant for FSD beta. It's a new system
1. I find it very unlikely that Tesla completely scrapped the previous system to start from scratch for the FSD beta.
2. If they had, that would make me even more suspicious of the new system because that would mean no one would understand its weaknesses yet.
Either way, being most interested in where the previous generation system and other semiautonomous systems have historically failed makes sense.
> it works amazing without markings on the road at all.
I have yet to see any videos of the new system handling poorly marked high speed roads one way or another, but the crash where it tried to veer in to oncoming traffic doesn't provide confidence. All the unmarked/poorly marked roads I've seen so far have been low speed residential/urban, and even there performance has been unpredictable at best. Let's see it on a highway where night time rain has made the markings hard to see. Let's see it when snow covers the road and your only hing or lane
The state of the art is such that it can simultaneously be amazing and terrifying to think that normal people are allowed to "test" it wherever they feel like trying on public roads. What it does with the limited inputs it has is incredible, but at the same time it has dangerous failure modes that do not seem to be predictable even to those who use the system regularly.
---
Don't get me wrong, there is nothing I'd love more than never having to manually handle a traffic jam, or "go home car, I'm drunk", but the way this is being released will kill people.