He's justifying shooting by the speed with which an assailant can shoot, but with zero data on the accuracy of the assailant when moving that fast. I'd want to know what is the estimated time to pull and shoot a gun with at least a roughly 10% chance of the officer being hit, not to mention being killed.