The irony of this thinking is that if you do the whole "teams must be gender balanced" routine early on where you are getting the 5-1 or worse ratio of men to women applying for a software position you can often be the one passing over way more talent than the company with the toxic white-male dominated culture lacking diversity. If men and women are equally good at the job (and in programming there is no reason to think otherwise) and you have 1,000 men and 200 women applying and there is a uniform distribution of talent there are going to be 200 men and 40 women in the top quintile, and if you are hiring 200 overall that means you are passing over 100 top tier men for 40 second and 20 third tier women. And by trying to discriminate to correct discrimination you leave the ~66 men you would have on average hired instead if you went purely based off choosing the "best" candidates for the job in the same position as minorities historically rejected from work due to the inverse bias - they get declined from the position because of facets of their being they have no control over, not on their own merit for the job.
Avoiding bias in hiring can never be as hamfisted as "hire equal numbers of X always" without causing a ton of harm in the process. Especially when you start trying to always hire an equal number of men and women, every ethnicity, every religion, every spoken language, every sexual orientation, every country of origin, etc. There has to be a balance between being inhospitable to diversity and using discrimination to force arbitrary amounts of diversity.
Simply applying normal statistics to candidates feels wrong as we usually have to make the “random independent variable” assumption for the simple statistical analysis we do.
I fairly certain that assumption does not hold about humans. Our social environment has a direct impact on our work outcomes.
Ergo, simply picking the top best candidates for some individual measure candidates does not guarantee the best work outcomes and so, depending on the effect size, doesn’t necessarily confer any advantage.
Avoiding bias in hiring can never be as hamfisted as "hire equal numbers of X always" without causing a ton of harm in the process. Especially when you start trying to always hire an equal number of men and women, every ethnicity, every religion, every spoken language, every sexual orientation, every country of origin, etc. There has to be a balance between being inhospitable to diversity and using discrimination to force arbitrary amounts of diversity.