The Bias Checklist

An airline pilot with fifteen thousand hours in the cockpit does not take off from memory. Before every single flight — the routine hop she has flown a thousand times, the one where nothing could possibly be wrong — she reads a printed checklist out loud, item by item, and a second pilot confirms each one. Flaps set. Trim set. Fuel quantity checked. Not because she has forgotten how to fly. Precisely because she hasn’t — because expertise is exactly the thing that makes you skip the obvious step, and the obvious skipped step is what kills people. The checklist exists to defeat the pilot’s own confidence.

Surgeon and writer Atul Gawande built a whole book, The Checklist Manifesto, out of this single unglamorous observation: in fields where the stakes are high and the pressure is crushing, a stupid little written list of things to verify outperforms trained human judgement — not sometimes, but reliably. A nineteen-item surgical checklist, tested across eight hospitals, cut deaths by nearly half. Nineteen items. Things every surgeon already knew. The list didn’t add knowledge; it added the one thing pressure strips away, which is the discipline to actually check.

This lesson is the defensive core of the whole course, and it borrows Gawande’s insight wholesale. You have spent four lessons learning that Lollapalooza stacks are invisible from the inside and multiplicatively overpowering once they tip. That diagnosis has exactly one honest prescription, and it is not “try harder to notice.” It is: build a checklist, and run it on purpose, out loud, before you decide. Everything below is how.

The defence has to be deliberate — here’s the killer fact

Go back to the single most important sentence from the introduction and lesson 3, because the entire design of the checklist follows from it: no single bias in a stack announces itself. Each one, from the inside, feels like a perfectly sensible reaction to the situation in front of you.

Sit with how total that is. When social proof is operating, you don’t experience “I am being swayed by the crowd” — you experience “well, clearly a lot of smart people see something here.” When commitment bias is operating, you don’t feel “I’m trapped by my earlier choice” — you feel “I’m not the kind of person who quits halfway.” When loss aversion is operating, you don’t think “I’m overweighting the downside” — you think “walking away would be a genuine, painful loss.” Every strand of the stack disguises itself as the situation, never as your mind misfiring. And a stack is several of these at once, each reinforcing the others’ disguise.

That has a brutal consequence for defence. You cannot catch a Lollapalooza effect by instinct in the moment, because instinct is precisely the faculty the stack has switched off. Waiting to “feel” the pile-up is like waiting to feel your own blind spot — the whole problem is that it produces no sensation. By the time the pull is strong enough to notice as a pull, it has already tipped you; you’re now rationalising, not deciding.

So the defence must have three properties, and each one directly answers a way instinct fails. It must be deliberate (you run it on purpose, because it will never trigger itself). It must be in advance — written when you’re cool, because you can’t author good questions while a stack is cooking you. And it must be out loud or on paper — externalised, because a check that stays inside your head gets quietly overruled by the same biases you’re checking for. Silent self-assessment is theatre. A written list you physically work through is not.

Building the checklist: a roster of the usual suspects

Munger’s own tool is a list of roughly twenty-five “standard causes of human misjudgement” — the tendencies he insists you should run against any important decision, the way a doctor runs through a differential diagnosis. You do not need all twenty-five memorised; you need a compressed, scannable roster you’ll actually use. Here is a trimmed version — the tendencies that show up most often in real Lollapalooza stacks — each paired with the question to ask yourself when you run the list.

Two things about the table matter as much as its contents. First, notice how many of the questions are embarrassing to answer honestly — “am I only doing this because I already put money in?” is a question your commitment bias desperately wants you to skip. That’s the tell that it belongs on the list. Second, notice that each question is phrased to be answerable with evidence, not with a feeling. You’re not asking “do I feel biased?” (you never will). You’re asking “is there a favour I’m repaying, a crowd I’m following, a sunk cost I’m honouring?” — facts you can check.

How to run it: slow down, count, invert, remove yourself

A list is inert. The defence is the procedure you run it inside. Five moves, in order.

1. Slow down — buy time on purpose. Every Lollapalooza stack has an accelerator, usually scarcity (“the price ends tonight”). The single most powerful counter-move is to manufacture a delay: sleep on it, take twenty-four hours, say “I’ll get back to you.” If a decision genuinely cannot survive one night of thought, that fragility is itself a warning sign, not a reason to hurry. Time is the solvent that dissolves stacks, because most of the biases in them run on urgency.

2. Make the stack explicit — say each bias out loud. Walk the table above and name, for this specific decision, which tendencies are actually operating. Not “am I biased” but “social proof: yes, everyone in my feed is buying it. Commitment: yes, I already put down a deposit. Scarcity: yes, there’s a countdown timer.” Speaking them turns invisible strands into a visible bundle — which is the one thing the stack’s disguise cannot survive.

3. COUNT how many point the same way. This is the heart of it, and it ties straight back to lesson 2’s critical mass. One bias operating is normal life; you can lean against a single strand. But when you tally the checklist and find two, three, or more tendencies all shoving toward the same yes, you are in the danger zone — the regime where the pile-up stops adding and starts multiplying. The count is your threshold gauge. A high count doesn’t mean the decision is wrong; it means your judgement is compromised and cannot be trusted raw. Treat three-plus aligned tendencies as an automatic “stop and get outside help” trigger.

4. Invert. Munger’s favourite move, borrowed from the mathematician Jacobi: invert, always invert. Instead of asking “why should I say yes?”, ask the mirror question — “what would make me walk away from this?” List the conditions under which this is clearly a bad idea, then check honestly whether any of them are already true. Inversion is powerful here precisely because confirmation bias has your forward-facing search rigged to find yeses; flipping the question routes around it.

5. Physically remove yourself from the pressure. Biases run on the environment — the auctioneer’s chant, the salesperson at your elbow, the countdown timer, the room full of people bidding. So change the environment. Leave the auction room. Close the browser tab. Get a disinterested second opinion from someone with no stake in the outcome — the single best de-biasing tool there is, because their instinct hasn’t been switched off by your stack. A friend who wasn’t in the room can see the whole bundle you can’t.

The latticework, as a daily habit

Here is where the whole site turns back on itself. The thesis of a latticework of mental models is that you understand the world better by combining models than by using any one alone. The Lollapalooza effect is the dark mirror of that thesis: it’s what happens when the biases combine. And the checklist is the thesis pointed back at the problem — you defend against combined biases by combining models.

That’s not a slogan; it’s the literal mechanism. Each row of the checklist is a model — social proof, incentives, loss aversion, confirmation bias — and running the list is running your latticework across a single decision on purpose. A stack overwhelms you because it hits several of your models at once and you were only watching one. The fix is symmetric: watch all of them at once, deliberately, using the written list to hold in view what your attention can’t. The person who has internalised twenty models and runs them as a checklist is doing, defensively, exactly what Munger does offensively — thinking in combinations instead of one model at a time.

This is why the course sits where it does in the ladder. You needed the confirmation-bias, incentives, and loss-aversion courses in hand not as trivia but as checklist rows you already understand. The Lollapalooza checklist doesn’t work for someone who’s never met these biases — the questions would be hollow. It works because you can read “incentive-caused bias: how is my adviser paid?” and immediately know what you’re looking for. The daily habit is simply: before any decision that matters, run the models you already own as a list, and count the ones pointing the same way.

A worked example: the “limited-time” upgrade

Abstract advice slides off. Let’s run the actual procedure on one concrete decision, ticking each row.

The situation. You’re on a call with a slick software vendor. Halfway through, the rep says the “founder’s pricing” — 40% off — ends at midnight tonight, and only for people on this call. Three other companies you respect are named as customers. Your own boss forwarded you the vendor’s blog last week with “we should look at this.” You feel a rising urge to just sign now and stop overthinking it. Good — that urge is the cue to run the checklist. On paper. Out loud.

Now count. Seven of eight tendencies are pointing at “sign tonight.” That is not a close call — it’s a textbook Lollapalooza stack, four biases past the critical-mass line, and the count alone tells me my judgement on this decision is currently untrustworthy. The volume of the pressure is evidence of the stack, not evidence of the deal.

Invert. “What would make me walk away?” — If the product doesn’t actually solve my problem; if a real reference call turns up unhappy customers; if a genuine competitor is cheaper without the theatrics; if the “deadline” is fake. Checking the last one costs nothing: I say, “I need a week — will the price hold?” A real business deal survives that question. A stack-driven one collapses, because the whole point of the deadline was to stop me running this exact check.

Remove myself. I end the call without signing. I sleep on it. Tomorrow I ask a colleague with no stake — “does this actually solve a problem we have?” — and I do a reference call with one of those three respected customers myself, rather than taking the rep’s word. The environment that generated the pressure is gone, and with it most of the stack.

The decision. Not “no” — not yet. The checklist doesn’t tell me the software is bad; it tells me I was about to buy it for seven reasons that have nothing to do with whether it’s good. So I move the decision to a calm room and a sober track-one analysis. If the product genuinely fits and the price holds (it almost always does), I buy it next week, having decided instead of having been stampeded. If the deadline was real theatre and the price truly vanishes, I’ve lost 40% off a thing I hadn’t established I needed — a bargain of a lesson.

Pitfalls: a checklist you don’t run is theatre

Two failure modes, and both are ways the checklist quietly stops working while you feel protected.

The unrun checklist. A list you own but don’t actually work through — out loud, on paper, item by item — protects you exactly as much as a fire extinguisher still in its box protects you from a fire. The whole power of Gawande’s surgical checklist was that teams read it aloud and confirmed each item; a checklist framed on the wall and skipped saves no one. Under the pressure of a real stack, the temptation to think “I basically know what it’d say” and skip it is overwhelming — and that temptation is itself the stack talking. If you only ever run the list on decisions where you’re already calm, you’ve built a tool that’s absent exactly when you need it.

Checklist over-confidence. The subtler trap: “I have a checklist” becomes its own bias — a new coat of armour that makes you feel immune and therefore drop your guard. Recall from the introduction that the people who fall hardest for stacks are often the confident ones sure their judgement can’t be swamped. “I’m too systematic to be fooled” is that same over-confidence wearing a lab coat. The checklist is a tool, not a talisman; it works only while you keep treating your own judgement as fallible enough to need it. The day you feel bulletproof is the day you’ve re-armed the very bias the whole discipline exists to disarm.

When to use it

Run the full checklist whenever a decision is important, hard to reverse, and made under any kind of pressure or urgency — the exact conditions a Lollapalooza stack needs. High stakes plus a ticking clock plus a room full of people all pointing one way is the four-alarm case; run every row. For small, reversible, low-pressure choices you don’t need the whole ritual — but keep the two cheapest moves always loaded, because they cost nothing and catch most stacks: count how many forces point the same way, and if it’s more than two, manufacture a delay and get one disinterested opinion. Those two reflexes alone — the count and the pause — will save you from most of the pile-ups you’ll ever meet.

And keep the humility that makes the tool work: you run the checklist because you can be fooled, not because you’ve stopped being able to. Which raises the honest next question — where does this model itself mislead you? The final teaching lesson, “Where the Model Lies to You,” turns the checklist’s own skepticism back on the Lollapalooza idea: the hindsight trap of stamping “Lollapalooza!” on every disaster after the fact without naming the specific tendencies, and the real cases where biases cancel instead of compounding. A model you can criticise is a model you actually own.