Equality of outcome might not be a good target if you assume that there are underlying factors that we don't want to control for that would cause unequal outcom

Equality of outcome might not be a good target if you assume that there are underlying factors that we don't want to control for that would cause unequal outcomes - which I think we can probably agree on, since (the relevant) minorities are under-represented in the pool of potential candidates and let's just say that's probably beyond the scope of what YCombinator can reasonably be aiming to achieve. However, let's say I build a function that selects a random item from a list of length 100. I run it 50 times and it only ever returns an item from the first 5 elements of that list. That's a pretty good reason for me to take another look at the function. It's probably broken. In the same way, whilst you're not aiming for a specific outcome, if the outcome is a long away from your priors - if you seem to be excluding a specific group of people from your company, you probably need to take another look at why that's happening and I don't think "Let's just assume we're doing a great job and that there are just no under-represented people to hire" is a reasonable explanation.