Or, we could have done a comparison with large strings and see the opposite result. Silly benchmark is silly (or should I say, specific)