Amid the avalanche of damning emails and embarrassing text messages that flowed out of the Waymo-Uber trial, what struck me the most was when ousted Uber CEO Travis Kalanick and Anthony Levandowski — Kalanick’s onetime business partner who he described in court as his “brother from another mother” — described the race to build self-driving cars as a “zero sum game.”
“We need to think through the strategy to take all the shortcuts we can,” Levandowski told Kalanick. “I just see this as a race and we need to win, second place is the first looser [sic].”
A pirate mentality has infected the autonomous vehicle space
This callous attitude toward a potentially game-changing, life-saving, and paradigm-shifting technology like autonomous vehicles is extremely troubling. A pirate mentality has infected companies in the autonomous vehicle space. “Move fast and break things” may have worked as an ethos for Uber when it was trying to overcome entrenched interests like the taxi industry, but it’s exactly the opposite philosophy you want to hear from a company that wants you to ride in its driverless cars.
Rule-breaking is nothing new in the auto industry. From the Firestone tire scandal to Dieselgate to the massive Takata airbag recall, car companies have always skirted the law in their relentless drive to gain a competitive edge. But autonomous driving is being held up as a panacea by many of these same companies, the ultimate solution to all of our automotive woes. That makes it much more fragile in terms of public perception.
The next couple years in autonomous vehicle testing are crucial. Both Waymo and Uber, as well as other companies like Lyft and GM, are expected to ramp up their experiments, deploying cars in more cities where they’ll invite an increasing number of regular people to take rides in them. The success or failure of these experiments will determine the degree to which self-driving cars will replace our traditional vehicles. Will they become our predominant mode of transportation? Or remain a niche market available only in select cities?
Meanwhile, consumer trust in autonomous vehicles is on the rise. A recent study found slight upticks in the number of people who say they would feel comfortable riding in a self-driving vehicle, though concerns remain.
Many people agree they would trust autonomous vehicles with a proven track record for safety. Almost three-quarters (71 percent) of U.S. respondents said they would be more likely to ride in an autonomous vehicle if they had an established safety record, up just slightly from 68 percent in the 2017 study.
But the quick settlement less than one week into the Uber-Waymo trial could undermine this budding confidence. Waymo sought to portray Uber as a spurned company pursuing a ruthless mission to grab the lead in the race to make money off of this technology. Kalanick’s answer to the question of what he meant by his desire for autonomous vehicle “cheat codes” — the former CEO described them as “elegant solutions to problems that haven’t already been thought of,” rather than a mercenary attempt to leap over Google’s dominant position by poaching its own talent — could be extrapolated and applied to any CEO who sees the rise of automated driving as an existential threat.
“cheat codes” and “shortcuts”
This win-at-all-costs wasn’t unique to Uber. Internal communications revealed Waymo to also be extremely insecure about its market position. “We have invested in self-driving cars, but over the last six months, we’ve stopped playing to win, and are now playing to minimize downside,” said Chris Urmson, Google’s former top self-driving car engineer who eventually left to start his own company, in a 2015 email to Larry Page and Sergey Brin. “We have a choice between being the headline or the footnote in history’s book on the next revolution in transportation. Let’s make the right choice.”
The idea that one of the companies that has already deployed autonomous cars was actively seeking “shortcuts” and “cheat codes,” or are only “playing to win,” should scare the shit out of everyone. These odious sentiments would never have come to light if not for the trial’s discovery process that brought so many of Uber’s underhanded tactics out into the open. Obviously, Uber’s new leadership is aware that there are trust issues and is actively seeking to claw its way back into the good graces of its users. But the company is still on a very aggressive timeline. “We will have autonomous cars on the road, I believe within the next 18 months,” new Uber CEO Dara Khosrowshahi said at a recent event in Davos. “And not as a test case, as a real [use] case out there.”
It wasn’t just Uber’s unprincipled drive for dominance that should leave a sour taste in the mouths of the public. Waymo’s complaint against Uber — that it stole over 100 trade secrets related to self-driving cars when it acquired Levandowski’s company Otto — was viewed by legal experts as an uphill battle. The company had yet to connect the dots on how Uber had unfairly enriched itself off Waymo’s proprietary information before the surprise settlement on Friday. If anything, the lawsuit could be seen as a warning shot to its own engineers to think twice before hanging out one’s own shingle — even though Waymo CEO John Krafcik recently told the New York Times that he didn’t foresee suing any other former employees. In its opening arguments, Uber sought to portray Google’s self-driving car projects as drained of all its smartest brains. The trial wasn’t about trade secrets, Uber argued, but about the market scarcity of expertise in self-driving cars.
Indeed, the race to build autonomous vehicles has set off a gold rush for talent: GM’s $581 million purchase of Cruise, Uber’s $680 million Otto acquisition, Ford’s $1 billion Argo Ai project, and Aptiv’s $450 million NuTonomy acquisition. This, in turn, is fueling considerable salaries for the computer scientists and engineers who have the expertise to build cars that can drive themselves. As reported by Forbes last year:
Salaries in the Bay Area, including annual bonuses and equity, currently average $295,000 a year for top self-driving car engineers, and range from $232,000 to as much as $405,000, based on data from Paysa, a Palo Alto firm that analyzes pay and job trends using an artificial intelligence-enabled data platform. The average is more than four times the California median household income of $64,500 in 2015 and over five times the U.S. average, based on Census figures.
But underlying these figures is a harsh reality in which the acquired talent fail to live up to their promises. Otto failed to meet the technical milestones outlined by Uber, which netted co-founder Lior Ron only $20,000 from the acquisition. Documents reviewed by Mark Harris at IEEE Spectrum suggest that Otto’s price tag was actually much lower than the reported $680 million figure.
It’s worth mentioning that the information these companies choose to disclose publicly is often incomplete or sometimes misleading. California, which has some of the more robust reporting requirements for companies testing autonomous vehicles on public roads, recently released its annual disengagement reports. These reports are intended to quantify the extent of these companies’ testing regiments, including the number of miles driven, frequency the vehicle disengages from autonomous mode, and the reasons why.
These reports don’t tell the whole story
These reports don’t tell the whole story, especially about companies like Waymo that has moved most of its testing to other states with less rigid regulations. Its definition of disengagement is pretty broad, as seen in GM’s failure to report an encounter with a taco truck that caused one of its safety drivers to take control of the vehicle. But these reports take on outsized significance because they are practically the only data that companies are required to disclose publicly that doesn’t end up becoming slick marketing opportunities like the vehicle safety reports issued by both Waymo and GM.
The little we do know about the self-driving tests that are underway is only based on what the companies choose to tell us through tightly controlled media tours. Local news outlets like The Incline have taken to crowdsourcing the movements of the dozens of autonomous cars that are operating in Pittsburgh. We haven’t seen any interviews with participants of Waymo’s Early Rider program in Arizona, most likely because NDAs prevent those people from speaking out. Most recently, we learned, thanks to reporting by The Information, that Waymo’s cars have trouble making left turns.
Legislation currently under consideration in the US Senate would more or less enshrine the voluntary guidelines outlined by Barack Obama and made even less burdensome under Donald Trump. (It’s unclear how something that was entirely voluntary could have been burdensome in the first place, but I digress.) Policymakers have become overly fearful of upsetting the sacred act of innovation and of asking too much of the innovators. And giant tech and car companies are taking advantage of the current regulation-averse environment to lobby for laws that wouldn’t even mandate a minimum level of safety and security.
Outwardly, these companies talk relentlessly about the life-saving potential of self-driving cars, but behind closed doors, they are plotting ways to outmaneuver their rivals — often at the expense of an unsuspecting public. It’s telling that we learned more about the behind-the-scenes machinations to build self-driving cars from the Waymo-Uber trial than we did from any publicly disclosed safety report. To be sure, there are probably a lot of good people working on this technology, many with the desire to build something that can really change things. But the industry is still in an experimental phase, and the public deserves to know more before these experiments become our reality.