You can easily test whether anonymization is working: have interviewers try to guess the identity characteristics of the applicant and if they're able to distinguish between them then it's failing. If the interviewer is unable to infer it, then the anonymization is working.
> I'm not sure how to have a conversation with a person's whose standard of evidence for "genders and races are inherently inferior" is "a guy on a website said it."
Who on earth said this? This reads like a complete no sequitur. Who is saying that genders and races are inherently inferior?
I'm not sure how to have a conversation who thinks that the mere notion that not all disparities are due to bias means "genders and races are inherently inferior." Men are vastly overrepresented among murder convictions. If someone says this is not because of bias, but because men commit more murders are you going to accuse them of sexism?
You need to revisit what "culture" means, and how many people of a given race share a culture.
> "Patterns and approach" are part of the software development skills being measured.
This was about speech patterns and conversational approaches, which vary greatly across cultures, both socio-economic, national and internal.
> Did you miss the empirical evidence to the contrary linked earlier?
You posted a blog post, from a startup. It was not peer-reviewed. It was not readily duplicatable. I'm not sure how to have a conversation with a person's whose standard of evidence for "genders and races are inherently inferior" is "a guy on a website said it."