George Hotz confirms Ego is Dooming his company and the Driverless Industry

4 min readMay 25, 2021

George Hotz, CEO of comma.ai the open-source poor-man’s mini-Tesla, recently appeared in a Disruptive Investing YouTube interview which can be found here — https://www.youtube.com/watch?v=4PwgZX9G9Xo

In the interview Hotz says some important, smart, reckless, and ridiculous things

· Using Waymo as an example he confirms why most AV makers will drive themselves into the ground vs make major autonomous system development paradigm shifts. Ego. Of course, the problem here is he thinks that paradigm shift is to do what Tesla does. Which itself is untenable from a time, cost and safety POV. (More in my first article below). Hotz is failing himself. Often making ridiculously transparent excuses. On one hand he states he doesn’t need or want more customers. But then says if he had them and more money, he could add sensors on the sides to cover turns. He also says it’s a data issue but doesn’t need more right now. It appears that is because of the same chicken and egg problem. He can’t sell more so he can hire more to make more progress. (I myself have seen this ego problem over and over. When I show AV makers why relying on the real-world is untenable and the gaming-based simulation technology they use is inadequate and explain the technology difference between that and what aerospace/DoD use, I then his a brick wall. No one wants to tell their superiors they need to make a paradigm shift. No one wants to tell their boss they were this far off.)

· Hotz likes to focus on his system staying in the lanes and ignore objects. He goes so far with this ridiculous cop-out that he actually says the whole thing is a psychology problem not one of physics. That the systems have to learn to drive like a person and predict object movement. And that is hard to do. While all of that is true you still have to detect them properly. I think he plays this card because he knows his single mono-camera — poor OEM radar sensor system is a mess like Tesla’s. (This is why Hotz stays with highways. He needs to avoid stationary and crossing objects. Tesla doesn’t get this.)

· Hotz, like Elon Musk, says he wants to ditch radar. Why? Because of the poor fidelity. This shows Hotz is lying or ignorant of the technology now available in this domain but NOT in the vehicles he rigs. Hotz used the example that the radar will pick up all the manholes and bridges. And that they cannot do any better, especially in the vertical plane. None of this is true. The poor radars Tesla and most AV makers and OEMs use have this issue because they use too few transmitters and receivers with wide beam patterns and do not want to spend the money to do any better. The design results in a coverage that is too broad. That results in not being able to discern object lateral or vertical position well especially at distance. That in turn means you cannot tell if the object is a car in the lane or a bridge pylon next to it. To avoid the false breaking the object is ignored. Hence the Tesla stationary and crossing object crashes, injuries, and deaths. With The 4D imaging radar from Arbe likely solves all of this. But using it would mean the OEMs adding it or comma.ai adding it. Neither is going to happen.

As I have written before I think Hotz is incredibly smart, usually in a micro context, but inept in the macro. Likely due to his ego. He would rather lie about technology than admit he was wrong all along and he is stuck with the poor sensor systems in the vehicles he modifies. He is going to end up in the same place as Elon Musk, and most of the autonomous vehicle makers. Bankrupt, unsuccessful, and having to deal with the knowledge that they injured or killed people as needless human Guinea pigs.

George Hotz confirms Ego is Dooming his company and the Driverless Industry

Written by Michael DeKort

Responses (1)