Contribution
An economist’s case that the single, consistent consumer of textbook economics is a convenient fiction for many decisions, and that a person can be treated, for self-management, the way a family or a firm is: as more than one decision-maker. The essay’s contribution is a vocabulary and a catalogue: it collects tactics of self-command, sorts habits along dimensions that matter for those tactics, and sizes the problem. It offers no mechanism for why the selves differ.
Key insight
The tactics we use on ourselves are familiar once we see them as the tactics we use on someone else whom we cannot always control: a child who scratches in her sleep, a recruit, a parolee. The “manager” is not always in charge, “especially not when management becomes a problem”. So the manager, while in command, arranges the environment, the incentives and the options that the other self will face. Mittens at night, or a friend told to ignore your later plea, are both this.
Assumptions
- No theory of the self. The essay has “no necessary intention of questioning what the ‘self’ is or offering an answer” (p. 100). The “two selves” are a manner of speaking about behaviour that alternates.
- Conscious, strategic behaviour is the focus. Schelling’s interest is “at the more conscious strategic end of the spectrum” (p. 101). The involuntary cases (sleep, panic, blushing) are used to make the voluntary ones look less mysterious.
- The size estimate assumes that about 30 million smokers want to quit, that they would otherwise smoke for another 15–20 years, and that future savings are discounted at a real rate of 4–10 percent (pp. 110–111).
Key results
The essay argues from cases. Its results are distinctions and estimates.
- Alternation, not coincidence (p. 99). “The spirit is willing but the flesh is weak” misses that the willing spirit and the weak flesh “frequently alternate rather than coincide”. The “flesh” seems “endowed with as ingenious an intelligence and command system” as the spirit.
- Against a single discount rate (pp. 99–100). Modelling the lapser as a high discounter would need a rate of “100 percent per hour or per minute”. It is unclear whether the resolver has a far horizon or “merely has such a short memory that he forgets how disagreeable it’s going to be”.
- Stark cases (pp. 103–105).
- The pain button: pain lasts 5 minutes regardless, and 25 more unless a button is hit, and the session repeats daily until a full 30 minutes is endured. Many could never endure it. A friend would disconnect the button and be thanked. The victim, at any moment, would disconnect it himself.
- The stroke device: an implant that kills only above a set severity of stroke. Schelling conjectures it “would be attractive to many of us”, and less so if the victim had to activate it.
- Taxonomy of habits (pp. 106–109). The dimensions are:
- time profile (random, cyclical, continuous);
- “time to indulgence” or onset time;
- warning time;
- portability, cost and visibility;
- the damage function (cumulative or current, threshold or plateau, decay rate);
- whether the habit is “autonomous or self-aggravating”;
- consciousness.
- Size of the problem (pp. 109–112). The would-be quitters’ cigarette spending alone gives “a minimum valuation somewhere around 100 billion per year”. Schelling suggests this matches all other consumer-ignorance and budgeting problems combined.
- Tactics (pp. 112–117):
- New Year’s resolutions raise the stakes and answer “When?”.
- Lippmann’s “plate glass window”: bright lines that defeat “salami tactics”, and “zero is a more enforceable limit” than a ration.
- Precautionary rules against precursors and cues.
- No self-deceit. Deceit is “peculiarly unavailable” against oneself.
- Finality relieves. Patients told flatly to stop smoking quit “more reliably” and “far more comfortably”.
- Parole-like monitoring.
- Breaking large tasks into small ones, and accepting deadlines.
- A three-person reward scheme: “If A lacks the fortitude to get out of bed, B has the fortitude to do without baseball because C laid down the law at an earlier time.”
- Mandatory delays that are almost always withdrawn before they expire.
- The suspense is the worst of it (p. 117). Failure is often “a desperate dash to freedom”, not from pain or hunger but “from perpetual unfinished decision”.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | People sometimes behave as if they had two selves that alternate in command | descriptive, from cases | pp. 95–98 |
| C2 | Many tactics of self-command are the tactics for managing another person, applied to one’s later self | argued from many examples; the essay’s central thesis | pp. 100–106 |
| C3 | A single high discount rate is a poor description of the lapser | weak–moderate: an argument from implausible magnitudes | pp. 99–100 |
| C4 | Bright-line rules (“zero”) are more enforceable than quantitative rations | asserted with personal examples, by analogy to arms control | pp. 113–114 |
| C5 | Deceit, common between people, is nearly unavailable against oneself | argued | p. 114 |
| C6 | Continual resisting, not the privation, is often what becomes unendurable | reported (“allegedly”), anecdotal | p. 117 |
| C7 | Consumer self-management failures are on the order of a trillion dollars in present value | an order-of-magnitude conjecture, explicitly so | pp. 109–112 |
| C8 | There is no general criterion for which self is authentic | argued from cases on both sides | pp. 98, 104 |
Concepts
- egonomics: the “art or science of ‘self-management’”, by analogy with economics as household management. Its strategic part is “consciously coping with one’s own behavior” (p. 100).
- sometime self: the self that will be in command later and that the present self manages as another.
- time to indulgence: “the time from the moment of suggestion or temptation, or of yielding or decision, to the act” (pp. 106–107).
- mandatory delay: a self-imposed waiting period before an indulgence becomes permitted (p. 116).
Connections
- Ainslie (LIT-586, NOTE-443). Ainslie supplies what Schelling declines to give: a reason the selves differ (hyperbolic discounting) and a mechanism by which a resolution binds (staking credibility on a precedent). Schelling’s own argument against a single high discount rate (C3) is consistent with Ainslie’s, since a hyperbolic curve makes impatience depend on nearness rather than being a constant rate. Ainslie cites Schelling’s Strategy of Conflict (1960) for “limited warfare” among motivational states, not this essay.
- Frijda (LIT-540) and THEORY-056. Schelling’s examples of using one state against another (alcohol “to be braver in an hour”, declining a drink to keep the cigarette at bay) fit Frijda’s picture of control by competing states. But most of Schelling’s tactics act on the environment and the options, not on concurrent motives. They are commitment devices, which neither Frijda nor THEORY-056 discusses.
Bearing on the record
- THEORY-029. Schelling poses the ownership question in its starkest practical form: when an earlier self asks you to disregard a later plea, “which is the authentic ‘I’?” (p. 104). He gives no criterion. His hardest cases are those in which the earlier self’s standard is itself in doubt (dieting for vanity, p. 98). That is the record’s point that a higher-order or earlier attitude cannot settle ownership by its position alone. Here the position is in time, not in order. Schelling neither argues nor cites this, and the bearing is the record’s.
- Personhood. The stroke-device and advance-instruction cases are about whose wishes govern across a change in the person. They connect to the record’s personal-identity readings (LIT-446, LIT-447) only by subject. Schelling makes no identity claim.
- Self-determination theory (LIT-558). The three-person reward and penalty scheme is self-imposed controlled regulation, and Schelling reports that it can work (p. 116). SDT predicts such regulation is less stable and less well-being-promoting than autonomous regulation. The essay does not compare them.
- No ML instruction. Nothing here belongs in the anthology.
Limitations
- Anecdote and introspection. No data are presented except the smoking prevalence trend and market estimates. The tactics are offered as familiar, not tested.
- The size estimate is avowedly rough: “it doesn’t matter whether you come out with one-third of a trillion or three trillion”.
- No mechanism. The essay explains neither why the selves differ nor when a tactic will work.
- A magazine essay. It is not the AER lecture the brief named, and it does not reach the questions of policy and rational-choice theory that the lecture’s title announces.
Open questions
- Does the 1984 lecture (LIT-573) add a theoretical claim about rational choice that this essay lacks? Only a reading can say.
- Which of the habit dimensions (onset time, warning time, self-aggravation) predict which tactic succeeds? The essay proposes the dimensions and tests none of them.