Possible-world semantics has been widely and successfully applied both in philosophy and in linguistics, and some of these applications are, or seem to be, largely independent of metaphysics. But not all are. Kripke claimed that the “distant planets” picture of possible worlds was a mistake. I agree. This picture is particularly pervasive and problematic when it comes to its treatment of intensional entities, such as propositions. According to the most prevalent version of possible-world semantics, a proposition is a set of possible words where it is true. What this set of possible worlds might be varies greatly to modal realists (e.g., Lewis, 1986a), actualists (e.g., Stalnaker, 2012) or deflationists (Thomasson, 2009). I take possible worlds to be abstract entities, akin to those postulated in science to help us clarify and understand the world: formal, heuristic tools, not explanatory ones. Following Perry (De Gruyter, 1989), I argue that propositions are abstract entities used to encode conditions on objects and properties: truth-conditions (for utterances) satisfaction-conditions (for desires), etc. I take my proposal to be similar to deflationary views, but I complement it with a realist position with regards to modality: modal properties and facts.