In his conceivability argument, Chalmers assumes that all properties have their causal powers contingently and causal laws are also contingent. We argue this claim conflicts with how conceivability itself must work for the conceivability argument to be successful. If conceivability is to be an effective mechanism to determine possibility, it must work as a matter of necessity, since contingent conceivability renders conceivability fallible for an ideal reasoner and the fallible conceivability of zombies would not entail their possibility. But necessary conceivability must either be governed by necessitating causal processes or by a necessitating non-causal mechanism. We argue the latter option is untenable or mysterious; whereas, if Chalmers chooses the former and applies it only to conceivability, his solution is ad hoc, but if he accepts necessary causal powers or processes generally, the conceivability argument fails. We conclude that, as it stands, the Conceivability Argument does not establish that physicalism is false.