Does anyone have a ballpark figure as to what extreme gammon's pr is? I would hazard to guess in the 0.1-0.2 range. I have an experimental idea of how to estimate it, anyways...For those of you who believe that it is zero, numerous counterexamples exist..weird positions that ex just does not grok..
Okay, here is my idea: Backscattering data: Have 100 players play 100 games at DMP. It does not matter how strong or weak the players are...variety helps. Then, correlate the number of moves made for each player, as well as their prs. Then aggregate the group, use the prs to calculate the expected losing rate for the aggregate, and compare this with the actual rate. Use mathematical techniques to estimate the strength of ex gammon, and calculate ex gammons pr. This will give us a good estimate...we can back propogate as need be to get as many iterations as we would like...all that we need? 100 volunteers! Any takers??

XG's error rate is very low. Your estimate of 0.1 to 0.2 sounds about right.

The error rates of Jellyfish 3.0 and Snowie 4.7 (as measured by XG) are about 1.4 and 0.8, respectively.

A much easier way to get the number you want is to play some money sessions against XG and then roll out all of XG's plays which are non-trivial. Playing a couple of hundred games should be plenty, and XG will do the number crunching for you.

Depreli Study covers all this in details here:
http://www.extremegammon.com/studies.aspx