On Waymos and their safety record

Open original

This is a critical reply to Waymo's safety record, arguing that its crash-rate comparisons are misleading and that the company has little reason to prioritize road users like cyclists. The author's core worry is that once self-driving fleets become essential, safety becomes a setting the company controls, not a democratic policy decision.

  • Waymo's headline comparison pits its vehicles against all drivers in the counties it operates in, ignoring that its fleet does little highway driving, which is far more dangerous, and that its cars are newer and safer than the median US vehicle.
  • Remote operators often guide Waymos, and the company decides what it discloses about how much human control is involved, which the author says cannot be independently verified.
  • Waymo drives very conservatively, which helps its safety record now but would make it slow and costly; the author argues that once cities depend on it, safety becomes a dial the company turns for profit.

Laura Michet's core objection is structural. Self-driving companies have little incentive to protect cyclists and pedestrians, since they can absorb fines and bring legal influence to bear, and she argues that letting a tech corporation become a road user is a bigger long-term risk than the benefit of automated driving. The author, who opposes corporate autonomous vehicle deployment for related reasons, starts by saying Waymo's safety claims deserve heavy skepticism. The specific target is a Futurism piece from last year, "New Data Shows Waymos Are So Safe That It's Almost Comical," which compares Waymo's crash rate to the crash rate of all drivers in the counties where Waymo operates. That comparison, the author argues, is badly apples-to-oranges. Waymos do little highway driving, which the US DoT data suggests is far safer than rural driving (rural roads run about 50% higher in fatality rate per mile). The cars are also newer, smaller, electric minivans and crossovers, compared against a median American fleet of SUVs and pickups. Waymo has published its safety papers in the same journal, and the author questions whether anyone with a real stake is positioned to check the methodology. A further problem is that Waymos are sometimes steered by remote operators, whom the company does not call drivers, and Waymo alone decides how often that happens, with no independent verification. The piece then makes a sharper point about what the numbers actually mean. Many crashes come from drunk or speeding drivers, so a sober rideshare passenger is already safer than the average road user, which inflates the apparent gain from Waymo. More important, the author argues Waymo's current cautious driving style, which makes the cars slow and prone to getting stuck, is a strategy for winning regulatory approval, not a permanent feature. If Waymo becomes a major transit provider, road safety becomes a dial that Alphabet can turn based on business goals. The author predicts the company would settle on something close to the fatality and cyclist-collision levels that policymakers already tolerate in the US, Canada, and Latin America, because safety costs money. The author's broader claim is that Waymo cannot solve a political problem with technology. Road safety depends on democratic policymaking, and a corporation moves those decisions into a private, unaccountable setting. Easy technical fixes also divert attention from proven measures like traffic calming and infrastructure changes that most American and Latin American cities have been slow to adopt. Treating Waymo as a safety solution, the author concludes, means buying safety from a company with no commitment to human life.

Ask about this article