Thanks Alex. I think this latest feature is much more basic than that; we're not implementing a collaborative filtering algorithm for applicants. We let startup

Thanks Alex. I think this latest feature is much more basic than that; we're not implementing a collaborative filtering algorithm for applicants. We let startups rate everyone who applies for their job. We think this helps, as everytime they log in, they can see who their preferred applicants are. The "collaborative" part of this just refers to the fact that we make it easy for multiple people on their team to rate a resume. A positive side effect of this is that we get data about how good our candidates are (or how good startups think they are, which is generally what counts here). This will let us focus our very limited resources much more accurately. "Empirically proven best practices" are tough in HR, especially for small companies, because every startup has such a distinct personality. This idea of "fit" is huge, and it's something we're working to address, but we also think it's important for our process to be as transparent as possible in such murky territory.