[JumpCrisscross]

> Obviously the "it’s safer than humans" claim is selection biased, statistically underpowered apples-to-oranges comparison with limited sample size

I haven't seen a good criticism of their methodology. If you have one, I'd be curious about their take.

On a more-direct measure, Waymos have had starkly lower fatalities and at-risk incidents than human drivers on average and, I think, near their best.