We know it's not a vehicle design issue since Teslas do very well in crash tests. Also not a sampling bias issue: "Tesla’s Model Y SUV had fatal accident rate of 10.6, more than double the average for SUVs, which was 4.8" -- those drivers are not demographically any different from other SUV drivers. - ThreadSky

fchollet.bsky.social • 102 days ago

We know it's not a vehicle design issue since Teslas do very well in crash tests. Also not a sampling bias issue: "Tesla’s Model Y SUV had fatal accident rate of 10.6, more than double the average for SUVs, which was 4.8" -- those drivers are not demographically any different from other SUV drivers.

Reposted from Alex Thiery

OK, but one indeed needs to be a bit careful with these numbers; the demographics of people driving Tesla vehicles are quite different from those of the general population

Comments

fchollet.bsky.social•102 days ago

The only plausible causal factor is FSD. If you've used it, it seems obvious. It's not actually autonomous but it causes drivers to stop paying attention.

fchollet.bsky.social•102 days ago

Since most Tesla drivers do not heavily rely on FSD, this implies that for those who do the fatal accident rate must be many times higher than average.

dkhundley.com•102 days ago

I have a Model 3 and have twice gotten 1 month trials for FSD, including right now. To be clear, I absolutely love my car, but if I wasn’t paying attention, FSD would have gotten me into 2 very bad accidents, potentially fatal.

gertfromwell.bsky.social•102 days ago

Even without FSD, the level 2 ADAS is called Autopilot, of course a lot of people will overindex on the “Auto” part and lower the attention threshold

chosunone.bsky.social•102 days ago

Would be a great thing to apply Pearl causal analysis to!

yannn.bsky.social•102 days ago

other hypothesis : tesla drivers behavior ?

fela.xyz•102 days ago

But if their numbers are to be believed there are less accidents with FSD (not necessarily causal, lots of highway driving, ... but goes in diff direction as your theory). So maybe less accidents but more serious/high speed ones? Other theories can too much acceleration, misleading numbers, ...

dmewes.com•101 days ago

Why would you conclude that it's FSD and not the stock Autopilot? Many many more Tesla drivers used Autopilot than FSD, since it comes for free with any Tesla. It's also a lot less capable in handling edge cases than FSD.

lukstafi.bsky.social•102 days ago

So how to square this with Tesla's published statistics: FSD users have half the accidents of non FSD users? Until a few months ago -- certainly during the reporting here -- FSD was only granted to people with good safety score. That would cut inferring that FSD is good, but still a defeater.

ratsbane.bsky.social•102 days ago

What about acceleration? I think I remember seeing that the AWD death rates were much higher than RWD - and AWD is quicker.

drivenbyevs.bsky.social•102 days ago

Go look up how many Teslas have FSD. Last time I looked it was less than 20%.

gertfromwell.bsky.social•102 days ago

> So how to square this with Tesla's published statistics: FSD users have half the accidents of non FSD users?

They lie, obviously. Tesla’s entire narrative and valuation is built on expectations of robotaxis and more than a just a car manufacturer multiples

natecochrane.bsky.social•102 days ago

The Volvo Effect.

Comments

Posting Rules

Reply