"Being good at technical interviews" is not the same thing as "being a good programmer". You can be bad at interviews but good at programming. You can be good at interviews but bad at programming. Turns out that there's a correlation between the former set and certain demographics, because of unconscious biases in the interview process.
I'd like to know their definition of technical interview because it seems to cover everything from fizzbuzz to arcane problems. I very much doubt there are good programmers that can't solve fizzbuzz.
Of course it doesn't. I talk about this in my other comment (https://news.ycombinator.com/item?id=12859833), many companies are okay with false negatives. And false negatives are okay. Sad, but okay.
However, if your metric for hiring is also correlated with favoring a particular demographic, you shouldn't be using it. False negatives are not okay when you disproportionately target one group.
"Being good at technical interviews" is not the same thing as "being a good programmer". You can be bad at interviews but good at programming. You can be good at interviews but bad at programming. Turns out that there's a correlation between the former set and certain demographics, because of unconscious biases in the interview process.