> Yes, some of that is people who would rather not bike at all than bike safely, and there are some non-linear effects from that, but it's still a net reduction in injuries.

Doubtful, since it raises the number of miles driven in cars, which would increase the number of injuries there.