1. The effect sizes we're talking about are tiny. A randomly selected cohort of men and women can routinely be expected to produce more women innately skilled at math, or more men innately skilled at negotiation. It's one of the more galling aspects of the debate about this stupid "manifesto", which at one point redraws a well-known chart about the overlap in ability between men and women to exaggerate the difference between the sexes: at no point do any of the advocates of the "manifesto" address effect size compared to the observed disparity in the field.

2. Very little about computer science as it is practiced in the industry is tied with any rigor to any particular skill. For the most part, software development is a standard white-collar symbol manipulation job in which productivity is defined mostly by meticulousness and generic learning and pattern matching. Attempts to break down aptitude by gender tend to imagine computer science in terms of compiler theory and algorithm design, when in reality 90% of all software development is repeated iterations of "wire this database column into this UI table".