Except the stats *do* matter. It’s how we measure things that operate in the real world. The blind baby stroller benchmark might be academically interesting, but if distracted drivers are smashing kids on public roads for another decade while Waymo perfects it’s craft the net result is just more flat people.