Ted Sider argues that semantic theorizing involves a metaphysical commitment to structure, i.e. a pre-carved, ready-made world. Because reference magnetism is the best response to Putnam’s underdetermination argument, and since magnetism requires objective structures, we should endorse the latter. I will show that reference magnetism is not the best response to Putnam’s underdetermination argument (at least not in the way that Sider understands that argument). Reference magnetism fails as a plausible conception of how our terms refer because it is an instance of meta-externalism, an altogether implausible view about reference. I will argue that, for the same reason, Sider’s alternative conception of how metaphysicians could decide to speak Ontologese, fails too.