The argument I want to advance, Bob, is really a probabilistic one [ETA this is my pompous way of saying "this looks kinda likely to me"
]:
-- in the 2010 paper Judy says that the samples she sent to Silverman for confirmation came from patients with "persistent viremia" and that it was possible to detect XMRV in them using only single round PCR without culturing ie. as far as she was concerned these were bona-fide, barn-door-obvious positives
-- Silverman took extraordinary steps to avoid contamination [see the partial retraction] and treated all samples identically but (eventually) found VP62 plasmid contamination and, amazingly, it was only in the patient samples, not the controls!
So what is more likely, that Silveman managed to selectively and exclusively contaminate patient samples or that the samples were contaminated on arrival (remember: persistent viremia == lots of testing, retesting, sample handling) ?
If you listen to the last TWiV you'll hear Racaniello and guests discussing this and it's pretty clear where they think the problem is.