Honestly, I think the BCS had it right. Use a bunch of hard evidence metrics and aggregate them and if you want to
include a human element then go for it. But say the committee picks their seeds and then weigh them 50/50 against the computer elements. The current system just allows these committee members way to much power.