Meet Dan O’Dowd, Elon Musk’s least favourite billionaire

Self-driving cars are all the rage at the moment. Waymo is spreading the movement internationally and Tesla wants to get in the game too. But on a smaller scale the latter company is promoting Full Self-Driving (FSD) for Elon Musk’s vehicles, and that’s got one coding veteran’s back up.

Dan O’Dowd founded Green Hills Software in 1982. It has earned him over $1 billion in profits, selling secure software to the US Military and commercial clients. His philosophy is simple – build code that is secure from the ground up and make sure it works. Which is why he has an issue with Tesla’s self-driving system and the deaths it has apparently led to.

O’Dowd’s an interesting chap. While he’s made serious money writing code for US Air Force bombers, the F-35 and NASA control systems, he still has some California wildness about him. There’s the skeleton of a Tyrannosaurus Rex dubbed Samson in the corporate headquarters, for example.

Five years ago O’Dowd set up the Dawn Project, to track and record Tesla’s progress – or lack thereof – in promoting self-driving cars. The organisation created tests to question Elon’s claims and bought Super Bowl adverts to question Elon’s assertions that his cars are safe for public roads. The Dawn Project says 60 deaths and 3,384 accidents have been caused by buggy code and wants the regulators to step in before more crashes happen.

Is this an anti-Tesla/Musk thing or do you actually drive them yourself? Many Californians do.

I have nine in total, across the family. Just not using FSD.

So what’s your issue with Tesla’s FSD system?

They have had years to make this safe and we have dozens of videos showing how the driver monitoring system does not work. We publish them; they don’t solve the problem. It’s amazing.

Years go by, and they don’t solve these problems. The new version is just pathetic, and the latest version, the cut down version that works on the old hardware Model 3 cars, and it’s a disaster area. It is running into barriers, it’s crossing the yellow line in the road. It’s been doing that for years, it hasn’t solved that.

What differentiates companies like Waymo from Tesla’s second self-driving systems?

Firstly, Waymo is very careful. They have multiple redundancies. They have five Light Detection and Ranging (LiDAR) systems and they have 19 cameras.

Second, they design appropriately and carefully, and they test it for a long time with professional drivers, not with civilians, not with ordinary people driving around beta software. [Tesla] does their beta software testing on the road with real cars with real people who aren’t professional drivers. If it does something dumb, [the cars] don’t know what to do. They weren’t trained, they didn’t go to defensive driving school. 60 people have died from involvement in these self-driving accidents.

There have been reports that Tesla cars turn off self-driving when an accident is inevitable. In your opinion, is there any truth to this?

Yes, that’s absolutely true. I mean, it doesn’t know what to do. It says, “I don’t know what to do. If I turn left, that’s not going to work, if I go right, it’s not going to work”. You’re in an inevitable crash situation where it screwed up. It got itself in a really big mess, and then it turns to the human driver and says “You take over.”

And you have one second to figure out what to do, which is not enough. And they smash into something, and then they say, “Oh, on page 93 of the manual, it says you have to be ready to take over at a moment’s notice in an instant. And then they say, “Not our fault; it’s the driver’s fault.”

And this is why you launched the Dawn Project?

We continue to do what we’re doing. Continue to publish about self-driving car screwups and how they should be better developed, designed, and regulated.

Throwing more money at something that’s not working just means you lose more money. You have to change fundamentally what you’re doing when you’re doing it wrong, and do it differently and better. We figured that out 30 and 40 years ago, and we built up the technology base of all these things being done right instead of wrong, and we never did it on a three-month schedule. We do it on planning schedules of many, many, many years.

So are you totally against self-driving cars?

Waymo is proving that it can be done. They aren’t perfect but, according to independent analysis, they are reducing traffic accidents by 80%. So they are in fact safer than a human driver, whereas Tesla is not.

Despite Elon Musk saying it completely, unambiguously, and clearly, that Tesla is safer than a human driver, it is absolutely not true. We have tested it. We’ve given it the eye test. We also gave it a California driving test. It failed completely four times. It is the worst driver on the road. Driving, it can’t drive ethically, it can’t park, it’s just insanely bad.

Can Tesla fix this?

This team can never make it right. I don’t think it’s ever going to be good.

They could fire the team and start over, but then that’s probably a 10-year effort. Waymo has proven it can be done. And in China, there are three companies that have full self-driving cars now. I have a friend who lives there, works there, and he says, “Oh yeah, I can get a self-driving car anytime I want”.

I’m sure somebody else will do it along the way, and the technology will become standard. But it won’t be Tesla. They are complete idiots. They’re not just second place, they’re no place. They’re not even in the rankings. It’s a complete joke.

More by Iain Thomson

• Don’t panic, says Katie Moussouris: AI security isn’t replacing humans, it’s proving the need for them
• View from the Valley: Are we facing an AI subprime debt disaster?
• SpaceX IPO makes Musk the first trillionaire, thanks to your pension funds