In the Odyssey, Odysseus wants to hear the Sirens sing. He also knows that any sailor who hears them steers straight into the rocks. So he does something strange for a hero: he plans for his own failure. He has his crew plug their ears with beeswax, tie him to the mast, and — this is the crucial part — ignore every order he gives until the island is behind them. He gets the song. He keeps the ship.
It's the oldest self-control trick in Western literature, and it contains an idea most habit advice still hasn't caught up to. Odysseus doesn't try to be stronger than the Sirens. He doesn't rehearse affirmations or summon discipline. He simply arranges the present so that his future self — the one who will beg to be untied — has no vote.
Behavioral economists call this a commitment device, and it may be the most underused tool in habit formation. Not because it's exotic, but because it asks you to admit something uncomfortable: the person you'll be at 6 p.m. is not on your team.
Your Future Self Is Not on Your Team
The underlying problem has a name: present bias. Humans discount future rewards steeply and inconsistently — a pattern researchers describe with hyperbolic discounting. The practical upshot is that your preferences reverse as temptation gets closer. At breakfast, tonight's run is obviously worth it; the payoff and the effort are both comfortably distant, so you weigh them sanely. At 6 p.m., the effort is immediate and vivid while the payoff is still abstract, and the same brain, doing the same math, reaches the opposite conclusion.
The economist Thomas Schelling — better known for Cold War game theory — spent years arguing that this is not a character flaw but a bargaining problem. In his essays on what he called self-command, he described the self as something closer to two negotiators who occupy the same body at different hours, each with genuinely different interests. The morning self writes checks; the evening self declines to cash them.
Most habit advice tries to make these two selves agree — to motivate, inspire, or shame the evening self into cooperation. A commitment device skips the negotiation entirely. It changes what the evening self is able to do, or what defecting will cost, before the evening self shows up.
What a Commitment Device Actually Is
A commitment device is any arrangement you make now, voluntarily, that restricts your future options or raises the price of abandoning your plan. The definition has two load-bearing words. Voluntarily — you choose the constraint, which separates it from coercion. And now — the whole mechanism depends on acting while you're in the calm, far-sighted state, before the preference reversal hits.
Researchers distinguish two flavors. Hard commitments carry real, external consequences: money you forfeit, access you lose, arrangements you can't quietly undo. Freezing your credit card in a block of ice is the folk version; a deposit contract — money you get back only if you follow through — is the lab version. Soft commitments carry psychological consequences instead: telling a friend your plan, scheduling a workout with a partner who'll be standing there if you don't show. Softer, but far from toothless — the discomfort of breaking your word to a specific person is a genuine price.
You already use commitment devices without naming them. Buying a small ice cream instead of the tub. Booking the nonrefundable class. Leaving your phone in another room isn't just friction reduction — it's present-you disarming future-you.
What the Research Shows
The evidence here is unusually concrete. In a well-known pair of studies published in Psychological Science in 2002, Dan Ariely and Klaus Wertenbroch looked at deadlines — a commitment device so ordinary we forget it is one. Students given evenly spaced, externally imposed deadlines for their papers performed best. Students allowed to set their own binding deadlines did something revealing: most chose to bind themselves rather than leave everything until the end, and they outperformed students with a single final deadline. But they also spaced their self-imposed deadlines suboptimally — evidence that people know they need constraints, yet under-commit when left to calibrate them alone.
In a field experiment in the Philippines, economists Xavier Giné, Dean Karlan, and Jonathan Zinman offered smokers a savings account with one rule: deposit money for six months, then take a nicotine test. Pass, and the money came back. Fail, and it went to charity. Smokers offered the account were meaningfully more likely to pass the test six months later — and, notably, still more likely to pass a surprise test six months after that. The commitment had outlived the contract.
And in one of the most successful applications of behavioral science to real life, Richard Thaler and Shlomo Benartzi's Save More Tomorrow program asked employees to commit — today — to saving a slice of their future raises. Because the sacrifice was scheduled for later, the far-sighted self signed willingly, and participants' savings rates roughly quadrupled over the following years. Same people, same income, different bargaining position.
Why Stakes Beat Rewards
There's a reason deposit contracts keep showing up in this literature. Loss aversion — the finding at the heart of Kahneman and Tversky's prospect theory — says losses loom larger than equivalent gains. Getting $50 for exercising is nice; losing $50 you already posted stings roughly twice as hard. The same dollars, framed as something to protect rather than something to earn, do more motivational work.
A large 2015 smoking-cessation study led by Scott Halpern in the New England Journal of Medicine captured both the power and the catch: deposit-based programs beat pure reward programs among people willing to sign up — but far fewer people were willing to put their own money at risk in the first place. Which is the honest fine print on this whole idea. Commitment devices work best on the people brave enough to use them, and the steeper the bind, the fewer volunteers it attracts.
The Art of the Gentle Bind
That fine print points toward design principles rather than maximal punishment. A commitment device fails in two ways: too loose, and the evening self shrugs it off; too tight, and one bad week makes you abandon the device along with the habit.
A few rules of thumb from the research. Bind the decision, not the effort. Automate the trigger — the standing appointment, the auto-transfer, the pre-booked class — so the choice happens once, upstream, in your clear-headed state. Right-size the stake. It should sting enough to change the 6 p.m. calculation, not so much that a genuine emergency turns into financial or social catastrophe. Shorten the horizon. Commit for two weeks, not a year; renew from the far-sighted state each time, the way Ariely's students benefited from many near deadlines rather than one distant cliff. And keep the committed action small — bind yourself to showing up, not to heroics, because a device that enforces an oversized habit just enforces failure.
You can build one this week without spending anything: text a friend the three days you'll walk and ask them to check; schedule the thing with another person attached; move the app, the snack, the remote somewhere that future-you would have to work to reach. The mechanism isn't the money. It's making the promise expensive to break.
Making the Promise Easy to Keep
Odysseus needed rope because the Sirens were loud and the mast was all he had. Most of us need something quieter: a structure that remembers what we decided when we were thinking clearly, and holds it steady when we're not. That's the idea Cadence is built around. You set the habit small — small enough that your future self has little to object to — and Cadence becomes the mast: the plan made in your far-sighted moment, showing up on schedule in your near-sighted one, keeping the streak visible so that breaking your word to yourself has a cost you can see. If you've been losing negotiations with your 6 p.m. self, stop negotiating. Make the deal once, and let the structure hold it — you can start at cadence.lumenlabs.works.