This! (Thank you for the comment). There’s a reason a 1000 random samples is adequate to reasonably estimate what’s common metrics in a population the size of USA or India (or infinitely large).
Yeah, that’s fair, but is implicit since I’m arguing against the “sample size is inadequate” POV, not the “there are distributional biases in data” POV. There are a gazillion ways to adjust for these biases (ex. propensity score matching) going beyond just user-base but also including weather type, road type, location, time of day, day of week, traffic density, pedestrian density … that can be done easily with far less than the sample size waymo has. And I bet they do these adjustments.
If the user base of "waymo riders" and "everyday drivers" does not match then you're not sampling what you think you are.