i feel some more non-sensical blabbering coming, read at own risk.
==========
think of the rankings as a sample of people racing? as not everyone participates in the system.
I'd have to give it more thought, but let's use strava for an example - say i'm 10th in a ranking, but i don't see any of the usual names
in front of me (or behind me) - so i really can't consider myself 10th (??) - but if there are 100 people in the ranking, than i'm in the top 10% -
this works for strava cause the segment doesn't change (although the wind and conditions do) - since there are others that aren't in the segment that
would end up behind me, the 10% strata should hold....(there may be a bias to faster people using strava - so i'm not sure if the sample is representative.)
the USAC ranking is like CXresults, in that it uses strength of field to get to a relative ranking - here is their FAQ.
http://www.usacycling.org/usa-cycling-rankings-faq.htm
they use 'best results' too - which is interesting, no weighting (pun) most current results (12 month window - not having enough events is a torpedo.)
----
here's the point.
i would postulate that there is a big data solution to finding potential dopers. Too much velocity to their improvement? Cyclical improvement? not sure of the discriminator, but there must be one.
Then having more races sanctioned, means more data, would improve their methods of identification, making it less expensive to find cheaters->lower fees.
vicious circle eh?
having significant rewards for turning in cheaters would probably help too.