|
|
|
|
|
by daryas
2521 days ago
|
|
> They don't want engineers testing code out on real people's lives. When Tesla pushes and runs code in the shadow mode, it is run in parallel to the "production" version of the software. The actual behavior of a car, even with the test code in, still relies on decisions made by the "production" version of the software[1]. The purpose of such testing is to compare decisions made by the updated code vs those made by the old version, and use this info for fine-tuning the next gen algorithms. [1] https://youtu.be/Ucp0TTmvqOE?t=6679 |
|
They've used this capability to introduce regressions in the safety of autopilot before.
When a stinky old "traditional automaker" wants to update a safety critical component, they spend months developing and testing before slowly rolling out to dealers. They want to change as little as humanly possible to limit their liability if something goes wrong.
Now some of you think this attitude is a bad thing, and that safety critical systems should be something you can update on a whim like you update your web server and keep iterating on.
But knowing developers like I know developers, being one myself, slow-as-molasses processes for safety critical code is a feature, not a bug for me.