B = confPt * (1-e_Pct) instead of B = confPt - maxPt * e_Pct => see example from paper: main.IS gets 0 P while 2 ts are correct. 4 total, 9 minimum. 2 of 9 is more fair then 0 of 9.