ABSTRACT I discuss three philosophical puzzles about changing the past , and offer puzzling solutions, informed by the Coombes–Lewis–Vihvelin view of ability, or CVL for short. The queries are, respectively, (1) whether changing the past can be modeled , (2) whether rational agents are able to change the past, and (3) whether, given determinism, the necessity of the past hampers the prospect for free will . A model is a complete, consistent way the actual world might be. The first puzzle is whether, supposing the possibility of time travel, one can model changing the past. I argue that all models of time travel are Ludovician , meaning they do not permit changing the past. Call this weak accidental necessity : Another question arises about the strong accidental necessity of the past, whether weak accidental necessity entails that the past is beyond human control. David Lewis famously argues against this inference, offering a model of no‐difference time travel along with a solution to the grandfather paradox : time travelers are able to change the past though they never do. CVL accepts Lewis's claims but replaces his contextualism with Kadri Vihvelin's wide‐narrow ability distinction, informed by Olivia Coombes's cognitive science model, and supplemented with my own account of warrant for judgments about wide ability. The first two puzzles give rise to the third. CVL is a compatibilist view, holding that free will—understood as requiring the ability to do otherwise—is coherent given determinism. However, assumptions about the necessity of the past are prevalent in versions of the consequence argument for incompatibilism , the denial of compatibilism. My comments about the consequence argument are complex. I begin with the no‐past objection and show that a standard reply leads to a determinism‐independent argument against the ability to do otherwise. In addition, the consequence argument rests on the assumption that a lack of wide ability logically transfers : I am unable to leave a room if the door is locked unless I am able to unlock the door. I show the inference is invalid.
Jan Campbell (Mon,) studied this question.