It turns out that if you let everything else go – blogging, answering email, grading, class preparation, sleeping, etc – it is possible to update a paper. Here’s the latest version of my imaginative resistance paper.

bq. “Morality, Fiction and Possibility”:http://brian.weatherson.org/mfp.pdf

There are several new features.

* I got rid of the most embarrassing examples.
* I added a discussion of Derek Matravers’s solution to the alethic puzzle.
* I added a discussion of Kathleen Stock’s objection to Gendler’s objection to the impossible solution.
* I dropped the claim that imaginative resistance is a feature of _all_ stories with morally deviant content. The data doesn’t support such a strong claim, what with it being entirely composed of short stories or moral claims that are easily detachable from longer stories, so I should never have made it. I’m currently agnostic about whether this stronger claim is true, though I’m tempted to say it isn’t.
* I changed my main theory.

The last bit is probably the biggest, though actually once I saw what I’d got wrong in the original it was an easy fix. Here’s a simple version of the puzzle. Imagine we have a story in which the lower level facts (e.g. the descriptive facts) are inconsistent. For example, we have two descriptions of X’s life, and on the one he always does the just and moral thing, and on the other he tortures children for fun and profit. (I think this is possible in inconsistent time travel stories, but since I’m meant to be neutral about how straight-up inconsistency works in fiction, I have to allow the prospect that it could just be true in the fiction without any trickery.)

So we have inconsistent descriptive propositions D1 and D2, where D1 is the description of a good person and D2 of a bad person. On the original version of the theory, it would have been impossible for any moral claims about the character to be true in the story. “That guy is good and D2” is an asymmetric compound impossibility, so it couldn’t be true. And “That guy is bad and D1” is also an asymmetric compound impossibility, so it couldn’t be true. So we end up with the position that the guy is neither good nor bad in the story. But that’s implausible. What’s more plausible is that he’s *both* good and bad.

Here’s the fix. What causes problems is when we have a higher-level claim (e.g. a moral claim) and the story is such as to exclude any descriptive claims that could ground it. So if it’s true in the story that X is bad, then (in general) there should be descriptive facts in the story that ground this fact. (The parenthetical qualification is that if the story is silent on all descriptive facts about X, it doesn’t need to make the descriptive facts match the moral facts. Sheer silence is permissible.) In this case, since D1 grounds “X is good” and D2 grounds “X is bad”, X is both good and bad in the story, as required. This solves a delicate problem I had to do with the _Fixing a Hole_ example in my paper, and which I had somewhat unfortunately pressed against rival theories without having a solution to offer myself. That sorry state of affairs has passed.