Let's just say I've talked with people who have worked with him on comma.ai and... I heard his cars would be driving into everything if it were just his code running them without all the safety features the folks around him added.

geohot isn't dumb, but he's also extremely good at making himself look like a lone genius, all the while he heavily relies on those around him. This has been a pattern his entire career. He started out pretty clueless and relying on others; these days he's pretty good but still relies on others, and at no point did he learn to humble himself a bit and credit everyone else.

I personally experienced this when we gave a presentation on Sony's PS3 ECDSA screwup, and a week later he posted the root keys obtained using our method. I had to email him to ask him to give us credit since he literally just implemented what we described in our talk, and made no mention of it. Then Sony sued us all (my name was on the lawsuit next to his, even though I had nothing to do with his idea of posting the keys outright).

The reality is "lone geniuses" don't exist. I'm pretty good at what I do (and I also have 15-hour coding marathon live streams on YouTube, FWIW), but to accomplish something great you have to work with others, whether you give them credit or not. The major projects I've worked on aren't "my" projects (even those which I lead/led); they've all been a team effort.

Thank you for this. Reminder that comma.ai famously cancelled their first product[1] after getting a very reasonable[2] letter with questions from a governmental safety agency. It generated quite a discussion at the time[3] but seems to have largely been forgotten since then. Given their history of playing fast and lose I’d rather not have any cars driven by their code anywhere near me on the road.

[1] https://techcrunch.com/2016/10/28/comma-ai-cancels-the-comma...

[2] https://news.ycombinator.com/item?id=12872399

[3] https://news.ycombinator.com/item?id=12815948

I'm probably one of the people you've talked to. For all his flaws, and there are many, I would give a lot for any team I've been part of since to have someone even half as good as him.

It is very evident in his videos that he is trying to show off. There is a lot of code snippets that he copies and lots of pasting code which is kind of boring to watch. In fact I really like to watch your videos as it deals with a lot of low level stuff.

I remember when Covid was new and he was messing around protein folding.

Yes. I have a watched a lot of his videos which I usually keep in the background. He dabbles into a lot of stuff from FPGAs in embedded to COQ (Formal language theorem prover) but there is not much depth in his videos. In a few weeks he would jump to something different again.

Kudos for those safety features, smart people don’t hire people dumber than them.

And it’s great that you found out that PS3 uses XKCD’s random generator for nonces (or maybe XKCD stole the random generator function from Sony, Sony should sue them).

Elon likes to take his teams ideas as well and present it as his: the gigacasting was a documented example where the Italian company had the idea years before Elon started talking about it, and they met Tesla at an auto show. Elon presented as he was looking at his Tesla matchbox and thought that it would be the best to scale it up.

Probably this similarity between Elon and George was the reason why they couldn’t work together.

I still love the idea that he (and his team) wants self driving cars to drive as an average human driver as possible, just with non-stop focus, while if you look at Tesla, they care about accident rates. The problem with Tesla’s approach is that on the videos nobody knows if the human driver should take it over or not, and I think that’s what’s really dangerous, and not the optimal way to get to full self driving. At the same time Tesla has orders of magnitudes more money.

I think he's a bit out of his depth with the self-driving startup and bigger projects he's undertaken, but when it comes to CTFs and hacking competition, he's pretty damn good. Idk if he still does, but for a while he competed until tomcr00se and he'd always be at the top of the leaderboards for nearly every competition. I don't think that's just luck or plagiarism.

You can see him in top10 of US [0] during 2013-2015 when he was most active, that's not a trivial feat.

[0] https://ctftime.org/stats/2014/US

He's usually on leaderboard (top 100) in Advent of Code. I don't know if he was this past year.

He placed on the leaderboards on a few days, but did not make top 100 overall.

And, of course, solving programming questions quickly is not a direct qualification for everything he says he is.

For example (not exhaustive):

https://adventofcode.com/2021/leaderboard/day/4

https://adventofcode.com/2021/leaderboard/day/5

https://adventofcode.com/2021/leaderboard/day/6

That’s not a problem with Tesla’s “cone AI”, it’s a problem with their self driving stack’s lack of a “do not drive into solid objects” feature.

It doesn’t matter what a static object is, you don’t drive into it unless you’re dang sure it fits in the class of “things that are safe to collide with.”

In comma.ai the solution is to just not run into things / go around things how a human would do and learn to peioritize what things to go around and where to be extra careful.

The important part of the thinking is at that point knowing whether a thing is a cone or not is just a technical debt in the code base, as it shouldn't be acted upon. At the same time those cones are perfect test to see if the software can learn to go around things or not, so that extra code hinders development of the final software.