How does your information change when you learn that something might be the case, where the modal “might” is epistemic? On the orthodox view, a proposition is added to your information base; on the view defended here, no propositions are added to your information base but some are removed from it. I (i) argue that Stephen Yablo’s recent attempt to define this removal operation as a kind of propositional subtraction fails, (ii) offer a definition of my own in terms of the part–whole relations between the truthmakers of the propositions one accepts, and (iii) argue that a deontic analogue of this account solves a problem about permission posed long ago by David Lewis.