Psillos (1999, 2011), Kitcher (1993), and Leplin (1997) have defended convergent scientific realism against the pessimistic meta-induction by arguing for the divide et impera (DEI) strategy. I argue that DEI faces a problem more serious than the pessimistic meta-induction: the problem of accretion. When empirically successful theories and principles are combined, they may no longer make successful predictions or allow for accurate calculations, or the combination otherwise may be an empirical failure. The shift from classical mechanics to the new quantum theory does not reflect the discarding of “idle wheels.” Instead, scientists had to contend with new principles that made classical calculations difficult or impossible (the Maxwell-Boltzmann equipartition theorem), and new results (the anomalous Zeeman effect) that were inconsistent with classical theorems (the Larmor theorem), and that suggested a new way of conceiving of atomic dynamics. In this shift, reference to atoms and to electrons was preserved, but the underlying causal explanations and descriptions of atoms and electrons changed. I propose that the emphasis on accurate description of causal agents as a virtue of background theory be replaced with Ruetsche’s (2011) advocacy of pragmatic, modal resourcefulness.