Where the Bridge Burns You

We started this course with a conquistador setting fire to his own escape route, and we have spent six lessons watching that same reckless-looking move win standoffs, deter invasions, and hold a nuclear peace together with nothing but the shared promise of mutual ruin. Burning the bridge behind you turned out to be a superpower.

This final lesson does two things. First, it brings the whole idea home — literally home, to your kitchen, your bank account, and your phone — because the exact trick Cortés used against an army, you can use against a much sneakier opponent: yourself, tomorrow. Second, it tells the truth about the failures. Burning a bridge is sometimes genius and sometimes just arson, and the difference is not always obvious from where you’re standing with the matches.

Committing against your own future self

Here is the strangest opponent you will ever try to out-strategise: the version of you that exists at 11 p.m. with a tub of ice cream, or on payday with an online shopping cart, or three drinks into a night when someone dangles the car keys. That future self shares your body but not your resolve. You, right now, calm and full of good intentions, know exactly what that person will do — and you disapprove.

A commitment device is any arrangement you set up in advance that removes your future self’s ability to defect from a plan you currently endorse. It is Cortés’s burned ships pointed inward. You destroy your own future option to cave, precisely because you don’t trust the person who will be holding that option.

The archetype is three thousand years old. In Homer’s Odyssey, Ulysses wants to hear the Sirens’ song — the sound so beautiful that every sailor who hears it steers his ship onto the rocks to get closer. Ulysses doesn’t want to be immune; he wants to listen and survive. So he engineers it. He orders his crew to plug their own ears with wax, lash him to the mast, and — this is the crucial clause — to ignore his later orders and tie the ropes tighter if he begs to be released. And beg he does. The song works exactly as advertised; the 11 p.m. version of Ulysses screams to be let loose. But the deal was struck by the calm, foresighted version, and that version had already taken the decision out of the weaker one’s hands.

Notice the structure, because it’s identical to every military standoff in this course: a party who currently has good judgment deliberately disables their own future ability to choose, so that a predictable future weakness can’t hijack the outcome.

The everyday menagerie of these devices is enormous:

Every one of these has the same shape as burning the ships. You spend something real — flexibility, money, face — to make a future retreat impossible or expensive. The only twist is who the “rival” is. In war it’s an enemy general; here it’s you, later, weaker. Same weapon, aimed at the mirror.

Bluffs that get called

Now the honest reckoning begins, and it starts with the most humiliating failure mode: the commitment you announce but can’t back up.

An empty threat isn’t neutral. It’s worse than saying nothing. Think back to Lesson 3’s credibility test: a threat only shapes behaviour if the other side believes you’ll actually carry it out. When you draw a line, dare someone to cross it, and then do nothing when they do, you haven’t merely failed — you’ve run an expensive public experiment that proves two things at once. You’ve revealed that you wanted to stop them (so now they know your preferences), and you’ve revealed that you won’t (so now they know your ceiling). You’ve handed a poker player both your desperation and your fold.

Worked example. A manager tells a persistently late employee: “If you’re late one more time, you’re fired.” The employee is late. The manager, short-staffed and conflict-averse, does nothing. What has the manager purchased? Not compliance — the opposite. Every future threat from that manager is now discounted to zero, and not just by this employee. The whole team watched the bluff get called. The next real ultimatum, the one the manager means, arrives pre-deflated. A boundary announced and not enforced is a training session in how to ignore you.

The commitment trap

The second failure is the mirror image of the first. Here the problem isn’t that you can’t follow through — it’s that you’ve locked yourself into following through on something that has become a disaster, and the lock has no key.

Cast your mind back to Lesson 2’s slider, where you discovered that commitment is a trap when the threat can’t hurt the rival enough. You handcuff yourself to an action, the action fails to move the other party, and now you’re stuck executing a plan that only damages you. Burning the ships is glorious when it forces the enemy to negotiate; it’s a death sentence when the enemy simply shrugs and waits for you to starve on a beach with no boats.

This is where commitment and one of the most notorious mental models in decision-making dangerously rhyme: sunk-cost escalation. The sunk-cost fallacy is throwing good resources after bad because of what you’ve already spent — money, time, lives, or pride that is gone no matter what you do next. “We can’t stop now, we’ve come too far” is its anthem. And commitment devices are practically engineered to trigger it, because their whole job is to make quitting feel impossible.

Worked example. A company is two years and a fortune into a doomed software project. Every rational sign says kill it. But the executives have staked their reputations on it publicly (a commitment device!), the budget is a monument to their judgment, and abandoning it means admitting the whole thing was a mistake. So they double down — another year, another fortune — not because the project will now succeed, but because stopping means booking the loss. The commitment that was meant to guarantee follow-through has metastasised into a machine for guaranteeing further loss.

The tell is the phrase itself. When “we’ve invested too much to quit” is the only argument left for continuing, you’re not committed — you’re trapped. A smart commitment binds you to a plan that’s still good. The trap binds you to a plan whose only remaining virtue is that you promised to finish it.

Doomsday devices that fire by accident

In Lesson 3 you learned that automaticity — removing your own discretion — is a way to buy credibility. A doomsday machine that retaliates automatically is terrifyingly believable precisely because no human hand can stay it. But that same removed discretion is the failure. The off-switch you tore out to convince your enemy is the off-switch you now don’t have when you need it.

Stanley Kubrick’s Dr. Strangelove built an entire film on this joke. The Soviets construct a doomsday machine rigged to end all life on Earth automatically if the USSR is attacked — the ultimate automatic commitment, immune to any leader’s second thoughts. It has two fatal flaws. First, they kept it secret, so it deters no one (a commitment nobody knows about influences nobody — Lesson 3’s visibility rule, violated fatally). Second, and worse, when a rogue American officer launches an attack by accident, the machine cannot be called off. There is no human left in the loop to say “wait, this was a mistake.” The credibility and the catastrophe are the same feature viewed from two angles.

The general lesson: every off-switch you remove to gain credibility is an off-switch you can’t reach in an emergency. Hair-trigger nuclear postures, auto-retaliation, “if I don’t check in by Friday, the lawyer sends the letter” — all of them buy believability by removing your ability to stop a mistake. Real-world false alarms in the Cold War (radar glitches reading a flock of geese or a rising moon as an incoming strike) are precisely the scenario where you’d desperately want the discretion you deliberately threw away. Removing your own brakes to prove you’ll crash into the wall works right up until the road unexpectedly curves.

The difficulty of un-committing when the world changes

The last failure is quieter and, for that reason, easier to walk into. A commitment that was genuinely brilliant on the day you made it can turn into a cage when the world moves and your binding doesn’t.

Credibility is bought with rigidity. That is the whole trade. You convince others (or your future self) that you cannot change course — and then the situation changes and you find that you truly cannot change course, exactly as advertised, exactly when you’d give anything to. The feature became the bug the moment the ground shifted.

Worked example. A country publicly pegs its currency to another at a fixed rate and stakes enormous credibility on defending it — a textbook commitment device that reassures investors and tames inflation. It works beautifully. Then a shock hits: the economy needs a cheaper currency to recover, but the peg that gave all that lovely credibility now forbids the very move that would save it. Defending the promise means grinding the economy into the dirt; breaking it means torching the credibility the peg was built to create. History is littered with pegs that were smart the day they were set and ruinous the day the world stopped cooperating. The same story recurs with long fixed contracts, rigid alliances, and constitutions written for a country that no longer exists.

This one is subtle because there’s no obvious mistake to point at. You didn’t bluff, you didn’t chase sunk costs, you didn’t remove a safety switch. You simply traded flexibility for credibility — a fair trade at the time — and then discovered that flexibility was the thing you needed most, and you’d already sold it. Every commitment is a bet that the future won’t require the option you’re destroying. Sometimes you lose that bet.

When to commit, and when to stay flexible

Step back and the entire course collapses into a single trade. A commitment device spends option value — the flexibility to react to whatever the future brings — to buy strategic influence over another party, or over your own future self. Options are valuable. Never give them up for free. So the question is always: is the influence worth more than the flexibility I’m burning?

That trade is worth making only when all three of these hold at once:

1. The influence gained exceeds the flexibility lost. The behaviour you change in the other party (or in future-you) must be worth more than every option you’re destroying — including options you can’t foresee. When in doubt, options are underpriced; the future is more surprising than you think.
2. The threat or promise genuinely bites. Commitment can’t manufacture leverage from nothing. If following through doesn’t actually hurt the other party enough to move them (Lesson 2’s trap) or reward them enough to move them (Lesson 4’s carrots), binding yourself to it just handcuffs you to a pointless act.
3. The commitment is visible and credible. A binding nobody can see influences nobody (the Strangelove secret), and a binding nobody believes influences nobody (the called bluff). If you can’t make it both known and believed, don’t make it at all.

Genius or arson? Sort the commitments

Each of the following is somebody tying their own hands. Some are textbook smart commitments — visible, credible, the threat bites, and the influence gained is worth the flexibility spent. Others are arson — bluffs that will get called, threats that don’t bite, sunk-cost escalation, rigid promises overtaken by events, or secret doomsday switches. Sort them.

The whole course in questions

Six lessons, one paradox, and a lot of controlled recklessness. Here’s the whole arc as a single recap.

The whole course in one picture