This competition entry has been selected for publication by the Forum team.
Epistemic status: This is a constructive reply to DiGiovanni’s "Resolving radical cluelessness with metanormative bracketing." I think the patch I’m proposing actually fixes a hole he left open. I don’t think it kills cluelessness as a whole, and I’ve tried to be upfront about where this is weaker than it sounds at first.
Metanormative Bracketing (MNB) is supposed to get around radical cluelessness by comparing actions across normative views instead of trying to compare consequences directly. You look for the biggest group of views where two actions become comparable, and go with whatever that group says. It works pretty well for ruling out obviously terrible stuff.
But there’s a problem DiGiovanni himself points out in an appendix. The answer you get can change depending on how finely you split up "normative views." His Elena example is the clearest: leak to the press vs. report internally. You get different verdicts depending on whether you treat "virtue ethics" as one thing, or split it into "honesty-priority virtue ethics" and "loyalty-priority virtue ethics." He notes this and says it’s for future work.
Jim Buhler flagged a related issue in the comments with his "Nuutti" example: you could just invent a group of views that happen to agree on the answer you want. DiGiovanni’s reply was basically “the group needs shared content, not just a shared verdict,” which fixes Nuutti. But it doesn’t fix the granularity problem. Even if every group shares real content, you can still slice those groups more coarsely or more finely, and that choice alone changes the result.
1. Ad hoc grouping
Fix: You should only bracket views together if they share actual content. Same moral theory, same epistemic principles, same decision theory and you pick that criterion before looking at the specific choice. That blocks Nuutti-style reverse engineering, but still lets us separate things we actually care about, like negative vs classical utilitarianism, or honesty-priority vs loyalty-priority virtue ethics.
2. Granularity
This is the tougher one. DiGiovanni already deals with uncertain credences by requiring robustness across a whole representor set instead of picking one distribution. We should do the same for uncertainty about how to individuate views.
Call it Individuation-Robust Metanormative Bracketing (IR-MNB): a verdict only counts if it holds across every legitimate way of partitioning the space of views. "Legitimate" just means it satisfies the content-sharing rule above. If a dominance claim only shows up under one particular carving and disappears under another, it doesn’t count.
Ruling out obviously immoral actions: still works.
The argument that non-immoral actions dominate immoral ones doesn’t depend on how finely you slice things. However you partition the non-clueless views, puppy-kicking still loses. So that part survives.
The neartermism conclusion: doesn’t survive.
DiGiovanni’s case for neartermism is basically: "the reasons behind longtermist views are weaker and more uncertain than neartermist reasons, so even with bigger numbers, longtermism doesn’t dominate." But that treats "longtermism" as one big blob.
He even has a footnote admitting that Clifton’s decision rule for imprecise probabilities could justify longtermist interventions. He sets it aside on empirical grounds real beliefs are too vague not on principle.
Under IR-MNB we’re not allowed to just lump things together. If Clifton’s rule is a legitimate, content-sharing view, we have to check it separately. Once you do that, the comparison DiGiovanni needed to close gets reopened. The “longtermist reasons are weak” claim isn’t robust to how we carve up the view-space.
"Won’t this just make everything incomparable?"
Probably, yes. That’s the cost. IR-MNB is a stricter standard than MNB, so it will resolve fewer cases. I think that’s the right price to pay to remove an arbitrary degree of freedom, but it does make the framework more modest.
"You haven’t shown longtermism wins."
Correct. I’m not claiming IR-MNB favors longtermism. I’m just saying it breaks the specific path DiGiovanni uses to get to neartermism. It’s a narrow result, not a reversal.
"DiGiovanni was already unsure about neartermism."
Also fair. He says he’s “not confident” MNB implies neartermism. So this isn’t toppling a core claim. It’s just tightening up an argument he was already tentative about. Still useful, but not decisive.
"Who decides what counts as a legitimate way to slice views?"
This is where I’m weakest. The content-sharing rule narrows things down, but it doesn’t give us one unique partition. So IR-MNB trades one open problem for a smaller, better-defined one.
