Two Ways to Beat a Bad Moment Before It Happens

What each one is actually for

Behavior Design

Environment design treats a behavior as mostly a function of what's convenient in a given moment rather than a function of character or willpower. A common demonstration: hospital cafeterias that move healthier food to eye level and put candy in a less convenient spot see people choose more of the healthier option, not because anyone became more disciplined, but because the friction to reach one option went up relative to the other. Applied at home, this looks like putting your phone charger in another room overnight instead of on the nightstand — not banning phone use, just adding the small friction of getting up and walking to reach it, which is enough to prevent the mindless late-night scroll for most people, most nights, even though it does nothing to physically stop you if you're determined.

Ulysses Contracts

A Ulysses Contract goes a step further: instead of adjusting friction, it removes the option entirely, or attaches a cost severe enough to function the same way, and it does this while you're in a clear-headed state making the decision for a future self you don't fully trust in the heat of the moment. The classic modern version: giving a friend a sum of money with instructions to donate it to a cause you find genuinely repugnant if you don't hit a specific goal by a specific date — the money isn't the point, the fact that you can't easily undo the arrangement in a moment of weak resolve is the point. A gambler who has a family member hold their paycheck, a writer who has an app lock their internet access for four hours with no admin override, a person who physically leaves their credit card at home before going somewhere they know they overspend — all of these work by binding a choice now so that a later, weaker version of you literally cannot un-bind it in the moment it would matter most.

What you get out of pairing them

The specific thing a person gets from having both tools rather than just one is knowing which tool a given temptation actually calls for, because using the wrong one either wastes effort or fails outright. Reach for a Ulysses Contract on something that friction alone would have handled — say, moving the cookies to a high shelf instead of a formal no-cookies pact with a friend — and you've spent social capital and rigidity on a problem that a small rearrangement would have solved. Reach for friction alone on something that actually needs a Ulysses Contract — a gambling urge, a compulsive spending pattern, a substance the person is already dependent on — and the friction gets overcome anyway, because a strong enough pull will absorb the extra three steps and drive to the store at 11pm regardless. Knowing the difference means you stop underreacting to serious pulls and stop overreacting to mild ones.

What Behavior Design rests on

• Behavioral activation precedes motivation randomised trials
  Taking action before motivation appears, rather than waiting for motivation to arise first, itself generates the motivation and improved mood that waiting would delay.
• Default effect shapes choice randomised trials
  People disproportionately stick with whatever option is set as the default, so redesigning defaults changes downstream behavior without requiring willpower.
• Friction determines behavior plausible mechanism
  Reducing the physical or digital friction required to perform a desired action increases how often it occurs independent of motivation level.

What Ulysses Contracts rests on

• Implementation intention closes intention action gap randomised trials
  Pre-committing to a specific situational cue paired with a specific response lets the behavior fire automatically when the cue occurs, without depending on in-the-moment willpower.
• Removing future optionality protects present intent randomised trials
  Binding a future choice through an external cost or barrier prevents a later moment of weaker willpower from overriding a decision made while thinking clearly.
• Temporal discounting erodes willpower randomised trials
  People systematically undervalue future consequences relative to immediate temptation, which is exactly the bias a binding precommitment is designed to neutralize.

Where they actually meet

Both practices make the same fundamental move: they take a decision that would otherwise be made by a compromised, in-the-moment version of you and relocate it to a clear-headed version of you making it in advance. Environment design does this softly, adjusting the landscape so the clear-headed choice remains the easy one to fall into by default. The precommitment device does it hard, removing the option so thoroughly that the in-the-moment self has no vote at all, only the ability to execute or not execute a decision that was already finalized earlier.

The principle underneath both

The version of you who has to actually resist temptation in the moment is not the version of you equipped to make good decisions about temptation — the decision-making has to happen earlier, by a version of you that isn't currently under the influence of the craving, the fatigue, or the immediate reward. Both practices are ways of pre-loading a decision made by a clearer self so that the compromised self, later, has less work to do or no choice to make at all.

What that makes the real goal

If the earlier, clearer self is where good decisions about temptation actually get made, the goal becomes: use that self, right now, while you have it, to shape what options exist for the later self, rather than trusting the later self to reason its way to the same conclusion under worse conditions. The size of the intervention — soft friction or hard removal — should scale to how much you distrust the later self's judgment in that specific moment, which is a question worth answering honestly rather than defaulting to either extreme.

Other practices that serve the same goal

Implementation intentions, the well-studied 'if X, then I will Y' planning format from Peter Gollwitzer's research, sit between these two in strength: they don't remove an option or add physical friction, but they do pre-load a specific response to a specific cue so the later self has less deciding to do, and this specific technique carries one of the larger evidence bases in behavior-change research. On the harder end, formal commitment devices used in financial behavior research — automatic payroll deductions into retirement accounts, illiquid savings products with early-withdrawal penalties — are essentially Ulysses Contracts studied at scale, and they show measurably larger effects on savings behavior than education or willpower-based interventions alone.

Using Environment Design to Handle Everything a Contract Doesn't Need To

Before setting up a formal precommitment device for something, ask whether friction alone would actually solve it, because most everyday behaviors will yield to a much lighter touch and a Ulysses Contract is expensive in a way that's easy to underestimate — it costs trust, flexibility, and social capital if it involves another person, and it's hard to calibrate once it's set. Someone trying to stop late-night snacking doesn't need to give a friend money contingent on it; they need the pantry rearranged so the fruit is at eye level and the chips are on a high shelf behind something else, which raises the friction just enough that the 11pm urge, which is usually mild rather than overwhelming, doesn't clear the bar. Reserve the harder tool for behaviors you've actually tested against friction and watched friction fail — if rearranging the kitchen genuinely didn't change the snacking, that's real evidence the pull is stronger than environment design can handle, and that's when a heavier, harder-to-escape commitment earns its cost.

Using a Contract for the Moments Friction Won't Survive

Environment design assumes the person encountering the friction is still capable of being deterred by it, which isn't true of every moment — genuine cravings, real emotional flooding, and established compulsive patterns can absorb an enormous amount of friction and still win, because the person in that state isn't weighing costs and benefits the way their calmer self would. For those specific situations, identify the exact moment the environment-design approach has already failed — you've tried the friction and the behavior still happens — and build a Ulysses Contract around that specific trigger rather than the behavior in general. A person whose environment-designed spending limits, cards left home, one browser extension blocking checkout pages, still get defeated during a specific emotional state, say late at night after a hard day, might set up a genuine external lock for exactly that window — an app that makes online purchases impossible between 10pm and 7am with no self-administered override, held instead by someone else or a service that can't be reasoned with in the moment the way a person's own willpower can be.

The Night the Chip Bag Won Anyway

You moved the chips to the top shelf behind the cereal boxes weeks ago and it's mostly worked — except tonight, after a genuinely bad day, you find yourself standing on a chair getting them down anyway, and you notice, with some clarity even as you're doing it, that the friction did nothing tonight because tonight isn't really about chips. That's useful information, delivered in real time: this isn't a friction problem on nights like this one, it's a different kind of pull that friction was never going to hold against. Rather than moving the chips even higher, which would just be more of a strategy that already failed under these specific conditions, you notice the pattern — it's always after a bad day, always around this time — and the actual fix isn't architectural, it's a Ulysses Contract aimed at the trigger itself: an agreement with a partner that on nights you say the specific phrase 'today was bad,' they're the one who decides whether anything gets ordered or bought, not you, because you've both agreed in advance that you-on-a-bad-night isn't the one who should be making that call.

When this is worth reaching for

Environment design is the right first move for almost any everyday, low-to-moderate pull — food choices, screen time, minor spending, procrastination on routine tasks — because it's cheap, reversible, and doesn't require involving anyone else or accepting a real cost if you fail. Move to a Ulysses Contract specifically once you have real evidence, not just a hunch, that friction alone has been tried and hasn't held — a repeated pattern where the environment-designed barrier gets overcome anyway, especially around a specific trigger like stress, a particular time of day, or a particular emotional state.

Where this does not apply

Don't reach for a Ulysses Contract as a first response to something you haven't actually tested against simpler friction yet — it's tempting to go straight to a dramatic commitment because it feels more serious, but most behaviors never needed that level of intervention and the cost of a failed or overly rigid contract, especially one involving another person's trust, can outlast the benefit. And don't use either tool at all for something that isn't really a temptation problem — a task that isn't happening because of an actual skills gap, unclear instructions, or a genuine external obstacle won't respond to friction reduction or precommitment, because neither one addresses 'I don't know how' or 'I don't actually have the resources,' only 'I know how and could, but won't, in the moment.'

What the evidence does and does not support

The friction and defaults research behind environment design has strong experimental support in behavioral economics — nudge studies on food placement, opt-out versus opt-in enrollment, and default settings consistently show meaningful effects, often larger than people expect from such a small intervention. Commitment devices have their own separate and in some ways stronger evidence base, particularly in financial behavior, where randomized studies on illiquid savings accounts and payroll-deduction defaults show durable effects on actual saved amounts. What hasn't been tested is the specific triage this page proposes — using failure of the lighter intervention as the signal to escalate to the heavier one — which is a reasonable inference from how both literatures independently describe their own limits, not a finding from a study that pitted them against each other.