I've always been baffled by how much we over-rate the statistically insignificant differences that separate competitors at the top end of the distribution.