I think the paper starts from the empirical existence of Born rule "weights" and attempts to explain in what sense they should be treated, decision-theoretically, as classical probabilities (since in the MWI sense, everything that might happen does happen) - but I admit I didn't grok the mathematics either.