TB: The Best Argument Against Deontology Is About Suitcases (Bentham's Bulldog)

Précis

A walk-through of Kacper Kowalczyk's "suitcases" thought experiment as a challenge to deontology: in cases where pushing one unidentified person saves two others who are also unidentified, everyone is better off in expectation. Every deontological response to this kind of case — ex-ante permission, sophisticated choice, resolute choice, ex-post prohibition, minimally Paretian — generates either deontic cycles, Pareto violations, or perverse incentives to remain ignorant. The conclusion the author urges: "deontology, rather than consequentialism, does not take individual people seriously."

Key Takeaways

The setup

Three people are inside opaque suitcases. Pushing one (unknown which) saves the other two. From behind the veil, every individual is better off in expectation if the push is permitted.
Symmetric, low-information cases like this strip away the usual deontological cover (identity, agent-causation framing) and force the view to give a verdict.

Failure mode 1: ex-ante deontology

Permitting pushes when beneficial in expectation triggers a predictable reneging problem — a rational agent commits to acting, then on learning identities discovers the act is wrong to execute. Pre-commitment and revealed wrongness diverge.

Failure mode 2: sophisticated choice

Timing variations let the deontologist know identities before acting, but the underlying moral calculus is unchanged — small information shifts swing the verdict from required to forbidden.

Failure mode 3: resolute choice

Binding to a sequence of acts creates implausible sensitivity to distant past intentions, and undermines the deontologist's core conviction that the wrong of harming is about the harming.

Failure mode 4: ex-post deontology

Rejecting all pushes generates deontic cycles: "no push > buttons a,b,c > buttons a,b > button a > no push" — moral preferences that loop.
The "Timer Case" forces ex-post views to require actions strictly worse for everyone than inaction — a Pareto violation by the lights of the view's own beneficiaries.

Failure mode 5: minimally Paretian deontology

"Shuffle Cases" show that for every available option, the view must violate either Pareto efficiency or a deontological constraint. Both pillars cannot stand simultaneously.

The perverse-incentive corollary

Several formulations make ignorance exculpatory — the agent gains permission to push by not learning who is in which suitcase. Poking out one's own eyes before the case becomes a permissible move.

John's annotation

Blogs.org does not include a written reaction beyond the headline, but the entry is shelved alongside Bentham's Bulldog's other ethics posts John has been collecting. The argument fits John's broader pattern of "consequentialism as the default unless a specific structural reason intervenes."

Notable Quotes

"Morality, if it means anything, means doing what's best for everyone." (the argument for pushing)
"It shouldn't be that wrong acts are strict upgrades over right acts." (against ex-post deontology)
"You shouldn't have such a strong aversion to learning things!" (against ignorance-as-exculpation)
Author's conclusion: "Deontology, rather than consequentialism, does not take individual people seriously."

Why this matters / Connections

The post is a clean structural argument: deontology is forced to violate one of (Pareto, cycle-freeness, indifference-to-information) under symmetric uncertainty. It is one of the few ethics arguments that survives translation to formal decision theory unchanged. Useful as a stress test for any normative framework: does the framework keep its commitments under symmetric ignorance and identity-shuffling?

Source

The Best Argument Against Deontology Is About Suitcases — Bentham's Bulldog, Bentham's Newsletter (Substack), 2026-05-05