So, I’m going to be “that guy’ and disagree with most of this list.

Having spent a lot of time with Porsche Club folks, I can confirm that some (not all) are insufferable about their cars and the brand.

I’m glad you asked.

For full transparency, I have - in the past - owned not one but two Honda Elements, so make of that what you will. At least I never stooped to a Pontiac Aztek.

Tesla Model S. When it was new it was certainly “something else”. It’s not objectively a bad vehicle; but all it really gives you over the competition is more range (and even those claims are questionable when put to the test). For a near $100k “luxury car” it is surprisingly not a luxurious place to be. It never was.…

Overpriced, maybe.  Overrated, no.  Virtually all their cars were objectively great driving cars in their time, and most continue to be so.

I never thought they were that special. I’ve always been a BMW guy with an RX-8 and an FR-S mixed in over the years. But, since picking up a (thoroughly pre-enjoyed) 981 Boxster S late last year, I am quite smitten. They really are excellent cars.

Yeah, the owners are insufferable, but they really are worth it.

Lol.

We’re more concerned with the guy juicing his stock prices and his fervent fanbase with objective lies, while selling them a product that *legally* contains none of the abilities that its users are ready to ascribe to it.

It’s like handing a child a smartphone.

I’m not explaining that analogy any more than that,…

I’m not sure where you got “level 2 conditions” out of my post.

“A crucial criteria of Level 3 systems, I believe at least, is that the system must be able to get out of active traffic lanes when requesting the human to drive, freeing the human from having to remain constantly vigilant and ready to take over with no warnings.”

Sure, I’ll rebut an armchair engineer.

Difference between level 2 and 3 isn’t how much trust. For level 2, you are the driver. For level 3, the system is driving and must notify you to take over with enough time for you to respond. You may be correct that data gathered from level 2 may be required to achieve level 3, it is a company’s choice whether to put…

No one is actually doing data training in production, that requires processing power only available on server farms. The issue is that you cannot just record normal drivers, because then you cannot train out all the possible wrongs, as no human driver may make them. For example, say there is a new situation never seen…

There is so much misinformation here, including from Jason. Please read the actual SAE document instead of guessing what the levels mean: https://www.sae.org/standards/content/j3016_202104/

There really, really does not need to be any data from the past, and in theory could be detrimental.

1. you don’t need or want 8k you want 2k at 120 fps, at 30 fps traveling at 85 mph you would be blind for every meter (since 85 mph is ~38 meters a second so divided by 30 for the FPS means that in between each frame ~1.25 meters have passed). 2k is more than enough resolution and much more viable at faster refresh…

I’m no expert in machine learning, but who the hell does data training in a production environment? Collect the damn driving data and do it offline. Besides, I have to share the road with those tech stunt bros, and I never signed up for that. Go train your robots off my lawn!

And better still - navigate that Boston traffic in a raging snowstorm.

All these not-quite-robocar manufacturers keep harping the wrong point about self-driving: they act as if the percentage of miles that a driver covers in a week that can be self-driven matters. (“Look! 80% of the miles you drive can be done by the car! We’re nearly there to Level 5!”) That line of logic is utter…