This is the core way we look at it. Our models are hopelessly simplistic, and have no chance of modeling the true complexity of human ability. It's important to stay humble and recognize this. However, what we're competing against is people (often non-technical people) making gut screening decisions. And it ends up that even relatively simple models can do a better job than most recruiters making gut calls.

And the fact that we are able to do this while being blind to background (and open to people from all sort of backgrounds) is something that I think is very positive. We totally make mistake and reject good people. But we get better over time, and we also help lots of people get jobs they might not have been able to get without us.