I despised the RPI based seeding/bids, as RPI is too easy to manipulate by home/away and playing average teams with great records. I do agree though that the logic is off on losing to "good" teams (Top 50 RPI), which I think is actually showing that the formula is still too RPI dependent (in the Knight D3 example, weren't the #1 seeds all the top 4 in RPI still? I wish I had saved the data there). Now, it's darn near impossible to eliminate RPI from a computer simulation, as in the end, it's a matter of formulas, but I do like the shift away from it, and I just think it needs a bit more tweaking when it comes to games against 'top' teams.