Odd Lots Search Engine Presents: Are you a good driver?
50 snips
Apr 8, 2026 
Don Burnette, self-driving car engineer and Kodiak Robotics founder, joins Sebastian Thrun, Stanford roboticist and early Google autonomous driving pioneer. They trace the long road to driverless cars, from DARPA desert chaos to Google’s secret project. The conversation digs into teaching cars human instincts, fierce internal rivalries, Uber’s pressure, trade secret drama, and whether robot drivers are actually safer.
AI Snips
Chapters
Books
Transcript
Episode notes
Comfort In Driving Depends On Context Not Physics
- Human driving comfort depends on context, not just physics, which makes teaching a robot to drive smoothly much harder than computing forces.
- Don Burnette says the same lateral acceleration feels fine on a highway on-ramp but insane in a residential cul-de-sac.
The Larry 1K Gave Google A Concrete Mission
- Google's team used the Larry 1K as a video game style gauntlet of ten brutal California routes, repeating each until the car finished without a takeover.
- They expected two years, finished in a little over one, and celebrated each completed route with cheap Corbell champagne.
The Real Product Question Was Assist Or Replace
- Google debated whether autonomy should assist human drivers or fully replace them, a fork that split Waymo's path from Tesla's.
- Sebastian Thrun argues robotaxi fleets could use cars far more efficiently than privately owned cars parked 96% of the time.
