10 comments

[ 0.22 ms ] story [ 21.1 ms ] thread
I had a discussion with a colleague today. He claimed Tesla had Full Self Driving, and had for years, and they were the first. That's the message some folks believe.

I look forward to telling him about this.

> humans generally have a crash, whether they are at fault or not, every 700,000 miles. Tesla has 7 in probably ~300,000 miles

This is the important part.

It seems like they should add lidar or radar.

What is the argument for deliberately impoverishing the Tesla sensory input?

I really don’t envy the supervisor’s job. Sitting there bored to death for weeks and waiting accident to happen. And when it happens you’re too bored and too tired to engage in timely manner.

On the other hand… the not paid version of cruise control continuously fails in my two years old model Y. Realistically looking it’s to early to fantasize about robitaxis when simple phantom braking problem is not solved yet.

If Tesla's robotaxis develop a reputation for accidents, they'll create an unpredictable traffic bubble around them.

Some people will slow down to minimize the fatality of an impact and to increase reaction time (similar to people slowing down around a marked cop car). Others will speed up to ensure they don't get stuck behind or around one.

That happens with other unsafe vehicles (e.g. a truck that doesn't have its load well secured). But it makes me wonder what will happen if Tesla trains on the data of erratic driving created by its presence.

> Unlike other companies reporting to NHTSA, Tesla abuses the right to redact data reported through the system. The automaker redacts the “narrative” for each reported crash, preventing the public from knowing how the crashes happened and who is responsible.

This part seems pretty bad

Flagged in under 15 minutes, seems the fever has still not broken
But how many miles driven? Severity? Is it worse or better than a human driver?

With just a total number it's hard to reason about what it means.