What is a Ulysses contract?

Another name for a commitment device, after the scene in the Odyssey where Ulysses has his crew tie him to the mast before the ship reaches the Sirens. He decides while sane what he will not be allowed to do while tempted. A modern Ulysses contract works the same way, deciding tonight what tomorrow's weaker version of you cannot renegotiate.

Do commitment contracts actually work?

The best evidence says yes, for a meaningful subset of people. Royer, Stehr and Sydnor (2015) found that gym-goers who put their own money on the line kept exercising after a completion incentive ended, with effects still detectable a year later. The Nature megastudy of 61,293 gym members (2021) is the honest caveat: 45% of its 54 interventions raised weekly visits while they ran, and only 8% left a measurable effect afterwards.

What is the difference between an incentive and a commitment device?

An incentive pays you to show up; a commitment device makes not showing up cost you. The distinction matters because paid incentives tend to fade within about three months of the money stopping (Mantzari et al.), while a commitment you fund yourself has produced effects still measurable a year on. The money is not the mechanism. It forces you through the weeks it takes for the habit to hold on its own.

Are app blockers commitment devices?

Yes, when they are armed in advance and hard to undo in the moment. A peer-reviewed study of one sec, which only adds a short pause before a chosen app opens, found people opened it 57% less often over six weeks. The friction interrupts the automatic reach. A blocker tied to a verified goal goes further, because the way back to your apps is doing the thing you said you would do.

Commitment devices: why motivation fails and what works

Motivation fails because it runs out before a habit forms, and because skipping costs nothing once it does. A commitment device fixes that second half: you set the cost of quitting in advance, while you still mean it. The evidence says stakes chosen ahead of time beat reminders, plans and streaks, and the effect can outlast the stakes themselves.

Every plan you have ever abandoned was made by a version of you that meant it. That is worth sitting with for a second, because it kills the usual explanation. You did not lack sincerity in January, and you did not suddenly become a different person in February. What changed is that the deciding and the doing happened at different times, and the person doing had nothing holding them to the person who decided.

Motivation is real, but it is fuel, not structure. It starts things. It does not hold them in place, because it is a feeling, and feelings respond to sleep, weather, work and lunch. The habit research puts the median time for a new behaviour to feel automatic at 66 days, which means every plan has weeks of unmotivated middle where nothing is pushing and nothing is pulling. If skipping costs nothing during that stretch, skipping wins often enough to end it.

A commitment device is the tool built for exactly that gap.

What is a commitment device?

It is an arrangement you make in advance that changes the price of quitting, agreed while you still want the goal. The oldest description is still the clearest one. Ulysses wanted to hear the Sirens without steering into the rocks, so he had his crew tie him to the mast and ordered them to ignore whatever he said next. He made the decision while he could think, and removed the option to unmake it while he couldn’t. Behavioural economists still call these Ulysses contracts.

The modern versions are less dramatic. A deposit contract puts your own money at stake, returned only if you do what you said. A public deadline puts your word at stake. An app blocker puts your evening at stake. The shape is always the same: present-you sets a real consequence that future-you cannot quietly talk their way out of.

The key word is quietly. Nobody abandons a fitness plan out loud, in one decision. It happens in private, one negotiated Tuesday at a time, and a commitment device works by making that private negotiation impossible or expensive.

Do they actually work?

Better than almost anything else that has been tested, with honest caveats.

The strongest single result comes from a large field experiment with employees of a Fortune 500 company [1]. They were paid to use the gym for a month, and some were then offered a self-funded commitment contract: their own money, forfeited if they stopped. The incentive-only group faded once the payments ended, which is what paid incentives do; a meta-analysis found health-behaviour effects generally gone within about three months of the money stopping [2]. The commitment-contract group kept going, with effects still detectable a year after the incentive ended. A follow-up study confirmed the pattern [3] and added the interesting part: people who chose commitment mostly understood their own weakness and were right to.

The honest caveat comes from the Nature megastudy [4], which tested 54 behavioural interventions on 61,293 gym members at once. About 45% raised weekly gym visits while they ran; only 8% left a measurable trace once the four weeks were over. Commitment and stakes were among the more reliable levers, but reliable means it works for a meaningful subset of people, not for everyone. Anyone selling a mechanism that works for everyone is selling.

Why do stakes beat plans, reminders and streaks?

Because the alternatives all depend on you caring at the moment of decision, which is the moment the whole problem lives in.

Planning is the clearest example. A large field study assigned real gym members a when-and-where planning prompt, the kind pop psychology swears by, and found no meaningful difference in attendance against controls [5]. The plan existed. The Tuesday didn’t care.

Streaks fail in their own way: they punish the wrong event. Break one and the loss has already happened, so day two of a broken streak is cheaper than day one was, which is backwards. Reminders are easier still, because a notification can be swiped away by the same thumb that opens Instagram.

What moves behaviour is a cost that exists at the moment of the decision, not before it and not after. Even tiny costs work surprisingly well. one sec, an app that merely adds a pause before a chosen app opens, cut how often people opened it by 57% over six weeks in a peer-reviewed study of 280 people [6]. Encouraging people to switch on Apple’s own soft Screen Time limits measurably reduced use over six weeks, with no money involved at all [7]. Friction at decision time beats intention before it.

What makes a good commitment device?

Four properties, drawn from where the studies agree.

Set in advance. The whole trick is that the strong version of you binds the weak one. Anything you can configure in the moment of temptation is decoration.

Verified, not self-reported. A contract you can lie to is a diary. The consequence has to key off something you cannot renegotiate: a measurement, a third party, a sensor.

Immediate. The cost has to land the day of the miss, not at a weigh-in three weeks away. Distance discounts consequences the same way it discounts goals.

Survivable. The point is to make skipping expensive, not to make one bad week ruinous. Good devices plan rest before it is needed and treat the first miss as a make-up, not a collapse. A consequence you would never actually accept is one you will disarm.

Where does Dare fit?

Dare is an accountability app that locks your distracting apps when you miss a fitness goal, verified automatically by Apple Health. It is a commitment device by the definition above: you choose the goal and the apps in advance, verification comes from the sensor rather than from your own report, the consequence lands the same day, and it is survivable, because the way back to your apps is simply doing the thing you said you would do. It is not money and it is not shame, which we think is the right trade: losing your evening scroll is expensive enough to change a decision at 9pm, and cheap enough that you will let it stay armed. It is available for iPhone; get it at idare.app.

The tools that do this differ less in quality than in what they put at stake: money, the apps on your phone, or someone finding out. Pick the one you would actually feel losing.

Whatever tool you use, the order of operations is the point. Decide while you mean it. Make quitting cost something. Let the boring middle be carried by the structure instead of the mood, because the mood was always going to leave first.

Sources

Royer, H., Stehr, M. & Sydnor, J. (2015). American Economic Journal: Applied Economics, 7(3).
Mantzari, E. et al. (2015). Preventive Medicine, 75.
Carrera, M., Royer, H., Stehr, M., Sydnor, J. & Taubinsky, D. (2022). Review of Economic Studies, 89(3).
Milkman, K. L. et al. (2021). Nature, 600. Megastudy: 54 interventions, 61,293 gym members.
NBER working paper on planning prompts and gym attendance (null result).
Grüning, D. J., Riedel, F. & Lorenz-Spreen, P. (2023). PNAS, 120(8). 280 participants over six weeks.
Hoong, R. (2021). European Economic Review, 140. Screen Time limits adopted voluntarily, no monetary incentive.