I agree with the idea that maybe it shouldn't weight "sub-standard" the same. Easy fix: use zaprank as a probability weight.