Annie Duke on quitting, sunk costs, and decision-making

Executive overview

We worship perseverance and treat quitting as weakness — but this miscalibration costs us time, money, and opportunity. Annie Duke argues that knowing when to walk away is a skill, and that most people quit too late rather than too soon.

The core problem is psychological: once we start something, we open a mental account and hate closing it in the red. This keeps us in failing jobs, bad investments, and dead-end projects long past the point of rational return.

Quitting is often the faster path to your goal — not a deviation from it.

Why we can't quit: the mental accounting trap

• Mental accounting: starting something opens a ledger entry; quitting converts an unrealized loss into a realized one.
• Being "in the losses" is a cognitive state, not an actual one — running 16 miles of a marathon means gaining 16 miles, not failing to reach 26.2.
• We quit too fast when ahead (locking in gains) and hold on too long when behind (hoping to recover).
• NYC cab driver study: drivers quit early on busy days and stayed late on slow ones — the opposite of rational. A fixed 6-hour rule would have earned 8% more; optimal switching would have earned 15% more.
• Retail investors cancel stop-loss orders when losing (to keep gambling) and take-profit orders early (to lock in gains) — same bias in both directions.

Sunk cost, ownership, and identity

• Sunk cost: fear of wasting past effort distorts forward-looking decisions. The right question is always "would I start this today?"
• Endowment effect: we value things we own more than equivalent things we don't — makes it harder to discard our own work.
• Identity: the hardest thing to quit is who you are. Sears sold off Allstate, Discover Card, Dean Witter, and Coldwell Banker — billions in market cap — to save a failing retail business because "Sears is a retailer."
• When a belief is outside the mainstream, disconfirming evidence makes people double down rather than update — stock analysts with contrarian forecasts cling harder when proven wrong.
• Cognitive dissonance resolves by rationalizing away facts, not by changing beliefs, because the cost of admission is too high.

When doubling down looks like courage

• Marathon runners finish on broken legs; we call them heroes. The rational move — stopping at mile three with sharp pain — would be mocked.
• We wait until quitting is no longer a choice (bone protruding, company out of cash, oxygen gone) because then no one questions it — including ourselves.
• Perseverance is courageous when the goal is worth the cost. Shackleton retreating to save his crew is different from continuing to vindicate a bad idea.
• Washington won the Revolution by being a strategic quitter: fought roughly 12 battles, won about half, and was expert at retreating to preserve his limited army.
• Grant understood that with superior resources, every engagement — win or lose — moved him closer to victory. Lee kept winning battles while losing the war.

How to quit faster and better

• Goals are motivating but inflexible — they become pass/fail finish lines that make people keep going past the rational exit point.
• Always add "unless" to any goal: Churchill's actual quote includes an exception for matters of honor and good sense.
• Monkeys and pedestals (Astro Teller, X): identify the hardest unsolved problem in any project and tackle it first. Build the pedestal only after you know the monkey can juggle.
  • Low-hanging fruit = pedestals. You already know you can do them. They produce the illusion of progress.
  • X evaluated a Hyperloop in 15 minutes: to test safe acceleration, you'd have to build the whole system first — not an X project.
  • Hyperloop developers spent years and millions before hitting the exact problems X spotted immediately.
• The goal isn't to avoid spending money on failures — it's to find out faster. Spending $2M to learn something is a $7M saving if the alternative is $9M.
• Ron Conway's rule: life's too short to keep resources in something unworthy of them.

Quitting as a creative practice

• Great writing requires constant quitting: cutting chapters, dumping fully written narratives, killing darlings.
• "Kill your darlings" is a sunk-cost problem plus an endowment problem plus an identity problem — the creator is attached on all three levels.
• Writing a proposal before the book is training the monkey first: it surfaces whether there is a book before you're four chapters into one that doesn't work.
• Stuart Butterfield shut down Glitch (a critically loved game) before anyone else could see it was failing — because employees working mainly for equity deserved to know their equity was worthless.