Tesla has the highest fatal accident rate of all auto brands, study finds

How much data does Tesla have on the details of crashes? Probably depends on whether enough electronics survived to phone home.

It's possible to dig the airbag controller out of the wreckage and read out the last 30 seconds or so. Airbag controllers have a short but nonvolatile memory and usually survive crashes. That gets you speed, acceleration in several axes, plus steering, brake, and power inputs, and detailed info about what the airbag system did.[1] Those were originally created to tune the airbag algorithm, and, over the years, false airbag deployments have dropped almost to zero.

That's the basic info needed to analyze fatal crashes. Speed at collision? Speed 10 secs before collision? Accelerator and brake inputs? Maneuvering (side accel) before crash? That, plus the crash scene, tells most of what you need to know.

Law enforcement will sometimes read out those units, when it's not clear what happened. It's not done routinely.

[1] https://www.nhtsa.gov/sites/nhtsa.gov/files/fmvss/EDR_QAs_11...

"Fun" fact... Tesla doesn't count fatalities in their accident stats!

Nor does it count accidents where there was no airbag deployment because as you point out, modern airbag systems use a wide variety of parameters, not just "if speed > x and collision = true; deploy".

So you can hit someone obliquely at 30mph, and due to factors, airbags don't deploy, and Tesla says "great, not an accident".

Or you can be in such a serious collision or similar where the airbags CAN'T deploy, and Tesla? "Not an accident".