Eh... there's a lot of hindsight factored into this point of view.
There's a lot to be said for being able to hire a personal trainer and manage nutrition very early in one's career to accelerate development and maximize earning potential. Tatis got all those benefits at someone else's expense.
He sold equity in his potential earnings to an investor. If he gets injured or flames out early in his career, that investor has to eat the loss. He signs a $340M deal and everybody wins.
I'm a Tatis Jr fan, but I have very little sympathy for him here. He has an AAV of around $24M. Between his agent and BLA, they probably take around $2M each, leaving him with around $20M before taxes per year for the next ten years. Cry me a river.
Marihuana was the primary spelling until a concerted effort in the early 20th century to change the spelling to make it seem more "ethnic" and undesirable within God-fearing 'Murican culture.
Eh... O'Dowd got to to play his gotcha card, but like so many "gotchas" it's a surface-level critique that ignores nuance and important facts. It ignores the "why" question.
Musk's 2019 statement alludes to the fact that Waymo's solution will not work outside of a geo-fenced area. Their system will only work in meticulously mapped areas.
Re: the current statement - Tesla's robotaxi will use geo-fencing for two purposes:
Keep vehicles inside of approved test area.
Avoid intersections and interchanges that are more likely to cause issues with the system.
The important question and distinction is why each of these systems use geo-fencing. Waymo uses it because their system literally cannot work outside of the geo-fenced areas. Tesla is using it to limit the operation of the vehicles to an approved test region and to maximize the system's opportunity for success within the test region.
The doc went into this a bit. One of the biggest expenses for a submersible is the support vessel. A lighter submersible can get by with a smaller (read cheaper) support vessel and lighter duty support equipment (e.g. cranes).
Here's one guy who went to a remote road to test. The Tesla failed against a carboard box with reflective material on it, but stopped at all speeds up to 55 mph for a pedestrian in the middle of nowhere. And this is a year-old version of the software.
And this guy's soundtrack is horrible, but lots of real world pedestrian interactions with FSD where it almost looks like FSD... anticipates?
Literally thousands of videos of FSD in action on YouTube showing it stopping for pedestrians of all sizes.
The problem with that video isn't that it is crappy quality, but that the quality is good enough to see that there is an alert notification on the screen, but bad enough to not be able to read the notification. It just so happens that the notification on screen looks exactly like the one that comes up if you press the accelerator to go faster than FSD/autopilot is currently set at.