Now that FSD Beta has, or nearly has, merged with production releases for a release to the whole fleet...
...what will Tesla's release rollout and testing strategy look like?
I've seen some people claim online that now the beta test program will just be over: only production releases that go to the whole fleet, big bang. I disagree.
First off: What does Tesla have to work on next?
A broad release of FSD for city streets, or even V11 single stack, is hardly the end! Areas of future iteration:
- Parking lot and garages navigation
- Transitioning from driving to auto-parking in a spot (public parking and garages)
- General improvement to continue moving FSDb from "mostly works" to "always works" (there is years of work in this bucket)
- Any potential future hardware changes (new camera positions, higher resolution, any new sensors)
- Snow and inclement weather
What are the benefits of a incremental release and testing program?
It's all about risk mitigation to enable a faster release cycle to get feedback faster and in smaller more targeted chunks. They've done all the work to be able to do gradual rollouts at scale (initial internal testing > employee testing > "OG" public beta testers > broad public beta testers > full production fleet).
My main argument is it doesn't make any sense for them to stop doing this now. There's only upside! Imagine they have a new feature or capability. Say, slow speed parking lot navigation that's not reliant on maps (in my experience FSDb is terrible in parking lots, so hoping this gets attention next year!). They've got a choice. Internally test a bunch and then boldly ship it to the whole fleet. Or after internal testing ship it to a smaller public test group to reduce risk from bugs. And to let people who want to be on the cutting edge and have practice with FSDb test it first. That seems like obviously the better path.
The might rebrand it from ("Request FSD Beta" no longer quite makes sense) but I think a test program is going to stick around in some shape.
...what will Tesla's release rollout and testing strategy look like?
I've seen some people claim online that now the beta test program will just be over: only production releases that go to the whole fleet, big bang. I disagree.
First off: What does Tesla have to work on next?
A broad release of FSD for city streets, or even V11 single stack, is hardly the end! Areas of future iteration:
- Parking lot and garages navigation
- Transitioning from driving to auto-parking in a spot (public parking and garages)
- General improvement to continue moving FSDb from "mostly works" to "always works" (there is years of work in this bucket)
- Any potential future hardware changes (new camera positions, higher resolution, any new sensors)
- Snow and inclement weather
What are the benefits of a incremental release and testing program?
It's all about risk mitigation to enable a faster release cycle to get feedback faster and in smaller more targeted chunks. They've done all the work to be able to do gradual rollouts at scale (initial internal testing > employee testing > "OG" public beta testers > broad public beta testers > full production fleet).
My main argument is it doesn't make any sense for them to stop doing this now. There's only upside! Imagine they have a new feature or capability. Say, slow speed parking lot navigation that's not reliant on maps (in my experience FSDb is terrible in parking lots, so hoping this gets attention next year!). They've got a choice. Internally test a bunch and then boldly ship it to the whole fleet. Or after internal testing ship it to a smaller public test group to reduce risk from bugs. And to let people who want to be on the cutting edge and have practice with FSDb test it first. That seems like obviously the better path.
The might rebrand it from ("Request FSD Beta" no longer quite makes sense) but I think a test program is going to stick around in some shape.