Waymo outperforms comparable human benchmarks over 7M+ miles
Waymo claims its driverless robotaxis have a significantly lower crash rate than human drivers over more than 7 million miles in Phoenix and San Francisco, prompting both praise from riders and skepticism from those wary of company-run safety studies. Commenters probe how comparable the numbers really are, noting limits like fair‑weather operating domains, carefully mapped service areas, and the lack of independent, standardized benchmarks for autonomous systems. Beyond safety, many argue over long‑term impacts: whether self-driving fleets will reduce private car ownership and crashes, or instead worsen congestion, undermine public transit, and concentrate liability and control in a few large companies.
Safety performance & benchmarks
- Many commenters find the reported crash stats impressive and consistent with personal Waymo rides; several say Waymo drives better than most Uber/Lyft drivers.
- Others question whether comparing to “average human” drivers (incl. drunk, teen, reckless drivers) is meaningful; some want comparisons to sober, law‑abiding, or professional drivers.
- Several note the study controls somewhat for location and surface streets, but only covers SF/Phoenix/LA‑type conditions and excludes severe weather; safety in snow/ice and poorly marked roads is called “unclear.”
- Some stress that even large mileage (7M+ robo‑only, 40M+ total) is still far from what’s needed to get statistically strong fatality rates.
Trust, data, and regulation
- Skeptics see this as PR timed after Cruise’s implosion and point to incentives to cherry‑pick metrics (e.g., definitions of “crash,” “disengagement,” and remote assistance).
- There’s repeated concern that NHTSA/independent bodies need standardized, audited benchmarks; reliance on self‑reported company data is viewed as fragile.
- Debate over whether AVs must be merely safer than average humans or dramatically safer (an order of magnitude) to be socially and legally acceptable.
Legal liability & ethics
- Discussion about who goes to prison or pays when a driverless car kills or injures someone: company, executives, engineers, or no one unless there’s gross negligence.
- Some argue society already tolerates tens of thousands of annual road deaths; others insist machine failures without clear accountability feel very different.
User experience & technical approach
- Many SF/Phoenix riders report Waymo as smooth, cautious but increasingly assertive, and reliable even in rain and complex urban scenarios; Cruise is described as more failure‑prone and now politically toxic.
- Ongoing debate: Waymo’s HD‑maps + lidar/radar/cameras vs Tesla’s camera‑only, end‑to‑end AI. Waymo is seen as the only one currently delivering true driverless service; Tesla’s FSD is widely called non‑autonomous and over‑marketed.
Traffic, urban design & societal impact
- Some predict safer, coordinated AVs will drastically cut congestion and crashes; others foresee induced demand, more VMT, more gridlock, and undermined public transit.
- Strong pro‑transit voices argue that car‑centric planning remains the core problem; AVs may entrench it unless coupled with policy (congestion pricing, transit investment, speed enforcement).
Economics & scalability
- Doubts about current unit economics: operating cost per mile is likely still higher than human‑driven taxis; R&D is massive.
- Counter‑view: hardware and software costs will fall, human labor is expensive, and once scale is achieved, robotaxis could undercut human ride‑hailing and reduce private car ownership in many cities.