[hintklb]

Tesla is not competition to Waymo.

There are 10 other companies that are currently testing without a driver. Those are competition.

Tesla so far is a gimmick of self-driving with a safety driver that takes over once in a while. That's where Waymo was more than 5 years ago.

[qwerpy]

The recent version of FSD in my Tesla is pretty amazing. Press "Start FSD" when in my driveway, and 20 minutes later it arrives at my destination and parks, without any input from me the entire time. I was skeptical too about FSD for a while but I'm starting to believe. These days I pretty much only disengage it when I'm impatient that it's being too polite. Unsupervised isn't far off!

[Mawr]

You're describing the 99% problem—getting 99% there is relatively easy, but once you do, you find out that you have the remaining 99% of work left to do.

This illusion is partially why Musk has been promising self-driving to be available at the end of $current_year+1 for years.

It's cool your car can drive itself. Now do it again, but fall asleep at the wheel. What's that? You're not willing to? That's exactly the core of the issue—it's not sufficient for the car to be able to drive itself most of the time, it must do so safely every single time, no matter what.

[njarboe]

I hope so. Similarly I only disengage when I want to go faster or forget to set the destination.

Too bad the owners of FSD can't decide that they want unsupervised. In an earlier or different world that would be possible.

[hintklb]

the issue is that your Tesla cannot be "too far off" as you seem to indicate. it needs to be right absolutely every single time.

One successful ride is easy. 10 successful rides is still easy.

100k successful rides without a serious incident? Way more difficult.

And that's why those anecdotal reports over a couple ride mean absolutely nothing

[nutjob2]

One successful ride on one route is not dispositive