Time Travel and Brian's Account of Imaginative Resistance
In his paper on Imaginative Resistance, Brian Weatherson says that the impossibility theory can't be true because "there are science fiction stories, especially time travel stories, that are clearly impossible but which do not generate resistance". Since you're reading this blog, you've probably also read the recent entry on TAR where Brian discusses time travel movies. Interestingly, he begins by noticing that "some [of these movies] seemed unintelligible even on relatively generous assumptions". I agree, and I would say that these are cases of imaginative resistance: A story tells me that certain facts obtain, but I find it unintelligible how these facts could obtain. Maybe we don't get the kind of immediate phenomenal resistance experienced in paradigm examples of IR, but I don't think this has any philosophical significance. I think it is largely due to the fact that we are not clever enough: We can't be struck by an impossibility if noticing the impossibility requires careful reasoning and keeping track of exactly what happened at various earlier passages in the story.
Indeed, Brian's blog entry shows how difficult it is to write a clearly impossible time travel story: Many time travel stories (including the ones quoted as impossible in the IR paper) are possible on the assumption of branching time, or branching time with cross-branch causal connections, etc. Oddly, Brian almost makes the same point in the IR paper itself when he uses something like branching time to explain (in section 9) why the Fixing a Hole time travel story (from section 5) isn't a case that his own account wrongly classifies as a resistance case.
While I'm at it, here are some other worries about Brian's account of IR:
First, despite the (very useful) introduction of four different kinds of IR puzzles it remains unclear to me exactly what kinds of cases Brian regards as (alethic) IR cases. I would say that we have an IR case whenever the ordinary conventions of story-telling imply that something is true in a story but rational consumers of the story do not believe that it is. But these cases also include Holmes' war wound, the digital watch in Titanic, the Tower of Goldbach, certain time travel stories, etc. that Brian excludes.
Another problem: Imagine a (rather boring) story that first describes the microphysical structure of a certain region, and later says that there is a donkey in that region, even though the microphysical structure is incompatible with the presence of a donkey. Like a time travel story with carefully hidden inconsistency, such a story won't provoke immediate resistance, though it is an IR case on Brian's account. Hence Brian can't make an immediate feeling of resistance necessary for IR.
Finally, I don't think Brian has a good explanation of the cancelability of IR: For almost every case of IR, we can remove the puzzlement by adding further facts to the story. Even the claim that the fork and knife in A Quixotic Victory are really a television and an armchair would become acceptable if in the story, Don Quixote is a magician who regularly turns people into frogs and furniture into cutlery. Brian says that stories always have an implicit "that's all" clause, whose effect is presumably that anything that is not explicitly said isn't true in the story. But this is clearly false, for in the Holmes stories the earth is round and Baker Street is near Paddington station. The explanation also doesn't account for the fact that much more is allowed (without IR) in a weird science fiction story than in a more realist novel. For example, I find it very hard to believe that in Thomas Mann's Zauberberg, the dead brother of Hans Castorp really reappears at a seance, even though the story is quite explicit about that. I wouldn't find this at all puzzling in a Douglas Adams story.