It’s meant to be perfectly general, but schematic. So plug in your favourite theory about the metaphysics of prudential norms/semantics of prudentially normative statements.

If you’re a contextualist, then I think you should accept the claims I made in any context. That will mean whether someone made the right decision or the wrong decision is context sensitive.

If you’re an IRI person, then it will follow that what the right decision to make is depends on which things the person does take as live options, and which as closed options.

Both of those consequences strike me as basically right – these are hard cases, and quirky theories (including my own quirky theories) should say odd sounding things about hard cases.

How general is this meant to be? What about cases when we know more than the agent does? What about, if these are different, cases where we’re just taking more possibilities seriously than the agent is/could?

Another question: suppose we wrote into the Newcomb case that the agent, though in a position to take the relevant conditionals as fixed points, nevertheless considers various skeptical possibilities. Your argument for one-boxing doesn’t go through in that case. Might it be that what choice one ought to make depends on what possibilities the agent is thinking about?

(unscientific)