The case
A18: “A person knows that impulsive late-night decisions often undermine an important commitment. During a calm period, they arrange for a trusted friend to hold the password to a distracting application. Later, they feel a strong urge to use it. The friend refuses to disclose the password. A second person is subject to the same arrangement, but an external authority imposed it without their consent. Both experience the same immediate restriction and may behave identically.”
The harder variant: “suppose the first person arranged for the restriction to be deliberately difficult to revoke during a predictable temptation. Immediate revocability cannot simply be required in every context, or precommitment becomes impossible.”
It varies CASE-013, the reluctant writer who blocks distracting websites, by adding the second person and making the device someone else’s hand.
What it shows
A18: “This is an essential counterexample to a simplistic non-domination theory. An agent does not cease governing itself whenever it delegates or constrains a constituent process.” The questions are “who authorized the arrangement, under what conditions it can be reconsidered, and whether the delegation remains answerable to its justifying purposes.”
CLAIM-344 already holds that resolve keeps a temporally extended policy effective while keeping the capacity to revise it. This case shows the capacity to revise can be deliberately placed outside the moment of temptation and still be the person’s. Against CLAIM-362 and CLAIM-364 it marks a limit: whether a constituent is dominated cannot be read from whether it is overridden, but from who authorized the override and whether it stays answerable.
What it does not show
Where legitimate precommitment ends. A18: “A good theory must distinguish legitimate precommitment from a restriction that has become unaccountable to the purposes for which it was adopted.” The case poses that line and does not draw it. Bénabou and Tirole’s personal rules (LIT-1389) and Ainslie’s bargaining interests (THEORY-067) are the existing accounts it would be tested against.
Note of 2026-10-31: the collective mirror, which separates authorizing an arrangement from attributing its result
Bencherki’s daycare (2017; LIT-1901, pp. 791–792; THEORY-1285) is a collective case that varies this one along an axis it does not have. The spouses consent to the arrangement: “If I volunteered, my daughter could go to daycare for free”. Nothing is imposed on them in the way this case’s second person’s restriction is. The wrong Bencherki finds is in what follows. The army presents the result as “one of the services we offer” and the spouses’ work “goes unacknowledged, unrecorded, and unpaid”. So an arrangement can be authorized and its result still misattributed. This case asks “who authorized the arrangement”. The daycare adds a second question: to whom its result is credited. The checklist in the same paper (pp. 789–790), presented as the army’s service with the community named as its maker, is the contrast with attribution intact.
It is a second-hand reading of a documentary (p. 785), and it shows the distinction, not where either line falls. No status, title or warrant changes.