But why publish this in HN, unless it's meant as an object lesson in poor experimental design.