I'm waiting for Tesla to correctly write their manual as a signal that they have the institutional rigor to write safe software.
I have been a car guy since I was born and one of the things I love is driving the car, I have no interest is FSD and all the update they have put out for FSD have done nothing but mess up the UI. It is unfortunate that you cannot decline updates that are related to this if you do not want FSD.
These comments are accurate for me as well. I'm a long time car guy, but also a long time tech guy and I love the idea of self driving cars and the progress that has been made. I watched and have followed Waymo cars when driving around Mt. View for years.
However- FSD as it is currently designed is a non-starter for me as a software guy. I've done software for 35 years, including a small amount of 'AI' neural network based machine learning, and there is not a chance in hell that I'm trusting Tesla software to get this right- when my life is on the line.
FSD 'Beta' is classic Demo software. Works just well enough to demonstrate, is not even remotely close to shippable. And yet here we are, shipping the Demo to people and using them and unsuspecting other drivers on the road as guinea pigs.
If the other Tesla software on the car were flawless, or they could show they can write a manual, I'd cut them some slack. But the car software is glitchy as hell. Almost every time I drive it does something weird. The FSD 'visualization' now that it abusively wastes a full third of my screen for their FSD advertisement- is terrifying. It shows stuff that is not there, it misses stuff that it should see. I'm telling you as a software dev, this is nothing more than Demo software, not even Beta.
Call me back when your software can make automatic windshield wipers and automatic high beams that aren't pathetic. Or when your so-called Automatic Emergency Braking can stop before pounding into a
Fire Truck. What I see with Tesla software is inexperienced software devs who do the easy parts first, and then stall out when they hit the hard parts. They don't seem to finish any of their software, and they don't have very good QA or testing, or don't care about bugs. Friends of mine have had the main display go black
while driving. How is that acceptable? Nothing is ever finished, and it's always a little buggy- yeah that's a good recipe for mission critical software.
I could be talked into paying $10K for it, if it could actually do low speed, non life threating driving without any attention on my part. If we keep the speed below 30 mph, I'm not going to die, and it'd be a great boon to be able to get time back in low speed scenarios. I see people making arguments that we will never get there, but I think it's feasible if we slow things down and lower the risks using low speed driving. Do fully automatic stop-and-go traffic with zero attention from me, and you've got something work $10K.
I hate to miss out on the self driving progress, but Tesla has not demonstrated any ability to properly write software, so at least for me, there is zero chance I'd pay $10K for it, and I'd never be first in line to use it even if it were subscription. It'll be in that deathly spot for years, where it's just good enough to lull you into a false sense of security that it works, and then screws up or insists that it's your job with 0.25 seconds before impact. See the Uber self driving experience for how humans react to systems that work 98% of the time.