Odd Lots · the Podbrain notes ·
4 min read

Search Engine Presents: Are you a good driver?

This episode features Tracy Alloway and Joe Weisenthal presenting Search Engine's investigation into autonomous vehicles. The story includes interviews with Alex Davies, author of Driven The Race to Create the Autonomous Car, Sebastian Thrun (former Google X...

Odd Lots Odd Lots
Subscribe to Notes Upgrade
Odd Lots episode thumbnail: Search Engine Presents: Are you a good driver?
Odd Lots
Key Takeaways
  1. 01

    Google's secret self-driving car project began in 2009 with just 11 engineers reporting directly to Larry Page

  2. 02

    Waymo claims its cars are 80% safer in serious crashes and 90% safer in injury-causing crashes than human drivers

  3. 03

    Anthony Lewandowski stole 14,000 technical files from Google before starting a competing company that Uber bought for $700 million

  4. 04

    Uber's safety drivers had to intervene once every 13 miles while Waymo's intervened once every 5,600 miles before the fatal Arizona crash

  5. 05

    4.8 million Americans drive for a living, making it one of the most common jobs threatened by autonomous vehicle technology

  6. 06

    Waymo has driven over 200 million real-world miles without causing a fatal crash, though statistical confidence requires 300 million miles

Get the latest ideas from Odd Lots.

Plus the best new takeaways from other top podcasts — read in minutes, not hours.

or

By continuing, you agree to Podbrain's Terms and Privacy Policy.

These notes may contain occasional inaccuracies. Learn how Podbrain notes are made

This episode features Tracy Alloway and Joe Weisenthal presenting Search Engine's investigation into autonomous vehicles. The story includes interviews with Alex Davies, author of Driven The Race to Create the Autonomous Car, Sebastian Thrun (former Google X leader), Chris Urmson (Aurora CEO), Don Burnette (Kodiak AI CEO), and Timothy B. Lee (Understanding AI newsletter author).

The narrative traces the 15-year journey from DARPA's failed desert robot races to Waymo's current robotaxi operations across 10 American cities. It explores the engineering breakthroughs, corporate espionage, fatal accidents, and safety data behind claims that robot drivers are significantly safer than humans.

The episode examines how a small Google team secretly developed self-driving technology on California roads, leading to today's multi-billion dollar autonomous vehicle industry that threatens 4.8 million American driving jobs.

DARPA's Desert Challenge Sparks the Self-Driving Revolution

In 2004, DARPA offered $1 million for any team that could build a self-driving car to navigate 150 miles of desert terrain, aiming to replace soldiers in dangerous convoy missions.

The first challenge was 'an utter hysterical disaster' - Chris Urmson's Carnegie Mellon Sandstorm got stuck on a berm and melted its tires, while Anthony Lewandowski's Ghost Rider motorcycle toppled immediately.

Sebastian Thrun observed from the sidelines and realized 'all the teams treated this like a hardware problem' when it was actually 'a software problem' requiring artificial intelligence.

The 2005 race with a $2 million prize saw multiple vehicles complete the course, with Thrun's Stanford team winning using machine learning to teach their car Stanley to recognize road surfaces.

Google's Secret 11-Engineer Team Tackles Public Roads

Larry Page, disguised in hat and sunglasses at the DARPA challenge, recruited Sebastian Thrun in 2009 to build a self-driving car 'that can drive anywhere in the world.'

The team's first challenge was the 'Larry 1K' - ten 100-mile routes through California including the Bay Bridge, Lake Tahoe, and Lombard Street, requiring zero human takeovers.

Early testing revealed major problems: 'It was weaving around like a drunken sailor' on Central Expressway due to lateral control issues unnoticed in empty parking lots.

Don Burnette worked on 'nudging behavior' - teaching computers to instinctively move left when a truck passes on the right, encoding human driving intuition into algorithms.

Corporate Espionage and the Uber Rivalry

Team tensions emerged between methodical Chris Urmson and risk-taking Anthony Lewandowski, who used self-driving cars for personal Bay Bridge commutes 'probably outside the bounds of what the team actually wanted.'

Uber's 2013 entry into autonomous vehicles created an 'oh shit moment' - Don Burnette realized 'these guys are going to eat our lunch' due to Google's lack of commercial urgency.

Anthony Lewandowski downloaded 14,000 technical files before leaving Google to start a company that Uber bought for $700 million, leading to criminal charges for trade secret theft.

Uber's 2018 fatal crash in Arizona occurred when their safety driver streamed TV shows while the robot spent 5.6 seconds misclassifying pedestrian Elaine Herzberg as 'unknown object, vehicle, bicycle.'

Waymo's Safety Data and Current Operations

Waymo releases unredacted crash data showing 80% fewer airbag-deploying crashes and 90% fewer injury-causing crashes compared to human drivers over 127 million miles.

Timothy B. Lee notes that while Waymo has driven 200 million miles without causing a fatal crash, 'statistically speaking, that could still be a fluke' - 300 million miles needed for confidence.

Recent incidents include a January crash where a Waymo hit a 10-year-old in Santa Monica, reducing speed from 17 to under 6 mph - faster than typical human reaction time.

Waymo employs 'human fleet response agents' in the US and Philippines who can't remotely drive but help cars navigate confusion and call emergency services after crashes.

The Future Battle Over 4.8 Million Driving Jobs

Former Google team members now lead major autonomous companies: Chris Urmson runs Aurora trucking, Don Burnette heads Kodiak AI, and Dmitri Dolgov co-leads Waymo.

Professional drivers represent 4.8 million American jobs - 'one of the most common jobs we have' - with unions organizing resistance: 'not in my city at the expense of our jobs.'

The technology rollout varies globally: robotaxis operate in 10 US cities providing millions of rides, while China has deployed them in twice as many cities.

Consumer confidence jumps from 20% among non-riders to 76% among those who've experienced robotaxis, according to JD Power data.

Odd Lots
From Odd Lots. Get a note like this from every new episode.
Subscribe to Notes Upgrade

These notes may contain occasional inaccuracies. Learn how Podbrain notes are made

0 / 0
Link copied