Your decision toolkit
Turn eleven models into three habits: sort decisions by whether you can undo them, run a premortem before you commit, and keep a three-minute decision journal. Then apply four models to one real case.
HOW MUCH CARE
One-way doors and two-way doors
Before you spend effort on a decision, ask whether you can walk back through the door. Two-way: a new supplier, a software subscription, a month in a different neighbourhood, a haircut. Get it wrong and you undo it. One-way: a two-year lease, quitting a job in a small industry, a large payment you can't get back, telling someone exactly what you think in front of others. Jeff Bezos described these two kinds in a letter to Amazon's shareholders, and his point was that people treat the reversible ones far too seriously and the irreversible ones far too lightly.
Most of what slows people down is deliberating over choices they could reverse in a week. Make those fast. Save the slow, careful process for the doors that open one way.
Check yourself
Which door is each of these?
- Trying a different route to work for a week
- Signing a two-year lease
- Testing a product on twenty customers before printing a thousand
- Telling a colleague what you think of their work, in front of the team
- Switching to a different supplier for one order
- Spending most of your savings on equipment for a new business
Show the answer
Two-way (reversible): Trying a different route to work for a week, Testing a product on twenty customers before printing a thousand, Switching to a different supplier for one order
One-way (hard to undo): Signing a two-year lease, Telling a colleague what you think of their work, in front of the team, Spending most of your savings on equipment for a new business
A premortem in ten minutes
- Fix a date and state the failure as done
'It's a year from today. The shop has closed.' Say it as a fact, not a possibility. That one move is what makes the reasons arrive.
- Write the story in the past tense for five minutes
'We ran out of cash in month seven, because winter was dead and the rent kept coming.' Past tense, specific, no hedging words.
- Pick the two or three likeliest causes
Not the most dramatic ones, the likeliest. This is where base rates from lesson 4 earn their keep.
- Turn each cause into something you do now
A smaller first order. Three months of rent kept aside. A written cancellation clause. If a likely cause has no cheap prevention, that is very much worth knowing before you start.
- Do it with other people when you can
Stating the failure as a fact gives the quietest person in the room permission to say the thing everyone had been thinking.
Check yourself
Amir's team is about to start a six-month project. Which of these is an actual premortem?
- Writing, as though it is already next autumn: 'we shipped three months late. Here is how it happened.'
- Asking each person how confident they feel, out of ten
- Listing everything that could possibly go wrong, including the unlikely disasters
- Going through what went wrong on the previous project
Show the answer
Writing, as though it is already next autumn: 'we shipped three months late. Here is how it happened.'
Yes. The past tense is the whole trick. 'It failed, here's how' produces specific causes; 'what might go wrong' produces a shrug and a short silence.
LEARNING FROM YOURSELF
The decision journal
You can't learn from your decisions if you can't remember what you actually thought. Memory rewrites itself: after a good outcome you remember being sure, after a bad one you remember having doubts. A decision journal is the fix and it costs three minutes. Before you know how it turns out, write what you're deciding, what you expect, why, and what would show you were wrong.
Six months later you read it back. Sometimes you were right for a reason that turned out to be false, which is the most useful thing you can ever learn about your own thinking.
| The five lines | An example entry |
|---|---|
| The decision | Take the job at the smaller company |
| What I expect | More responsibility, similar pay, a calmer year |
| Why | Two people who left said the team is steady |
| What would show I was wrong | Still doing my old kind of work in month four |
| Review on | 1 March |
Check yourself
Nima's journal is only worth keeping if he writes each entry before he knows how things turned out. Filled in afterwards from memory, it records a tidier story than what he actually thought.
Show the answer
True
True. Once you know the outcome, your memory of your reasons quietly rearranges itself to match it. Written afterwards, the journal files a neat story and teaches him nothing. Written before, it is the only honest record of what he believed at the time, which is the whole point.
The case: Arman has a steady job and sells furniture he makes at weekends.
He wants to quit and do it full time.
Map or territory? (lesson 1)
His spreadsheet says the income works out.
The ground: 14 orders in a year, nearly all from friends of friends.
-> the spreadsheet quietly assumes strangers will buy. Nobody has tested that.
Base rate (lesson 4)
He finds three people in his trade who went full time.
Two are still going after two years; one went back after eight months.
-> not hopeless, not a sure thing. Start near "about even", then adjust.
Expected value and ruin (lesson 5)
A good year beats his salary. A bad year is about a third of it.
Savings: four months of costs.
-> the average looks fine; the bad case runs him out of money in month five.
Premortem (this lesson)
"It's 18 months later and I'm back in an office. What happened?"
1. Orders dried up once everyone I know had bought one
2. I spent the good months making, and never learned to sell
3. A large order was cancelled after I'd paid for the woodWhat he does instead of quitting:
- asks for three days a week for six months: a two-way door, not a one-way one
- spends the other two days selling to strangers only, the untested assumption
- one rule: no order started without a deposit
- a journal entry to review on 1 September:
"I expect 8 orders from people I've never met by then.
Fewer than 4 means I was wrong about demand, and I stay employed."No model gave him the answer. Together they turned 'should I quit?' into a smaller, reversible test of the one thing he didn't know.
Check yourself
Which part of Arman's plan does the most work?
- The journal entry, because writing things down helps him remember
- The base rate, since two of the three people he found are still going
- The deposit rule, because it stops him paying for materials on an order that gets cancelled
- Asking for three days a week: it turns a one-way door into a two-way one
Show the answer
Asking for three days a week: it turns a one-way door into a two-way one
Right. All four help, but the part-time arrangement changes what kind of decision this is. He finds out the same things at a fraction of the risk, and he can still walk back through the door.
Which model, when
- Deciding from a description (a listing, a CV, a plan): map and territory, then find the cheapest look at the ground.
- The usual answer is out of reach or keeps failing: first principles. Out of your depth: circle of competence, and borrow one.
- Judging a plan's chances: base rates first, then expected value, then check that you survive the bad case.
- Anything involving other people or rules: second-order thinking, and ask what the easiest way to win under the rule is.
- Before committing: inversion and a premortem. Choosing between options: instead of what? Already spent: it gets no vote.
Check yourself
Match each situation to the model that helps first
Show the answer
- A supplier's brochure promises everything → The map is not the territory
- You feel sure, but you've never done this before → Circle of competence
- A bonus scheme people could game → Second-order thinking
- Two years in and it still isn't working → Sunk cost: decide from here
Course recap
- Sort by the door: reversible decisions deserve speed, irreversible ones deserve a slow read and a premortem.
- A premortem states the failure as a fact and asks how it happened, which is what makes real causes appear.
- A three-minute decision journal, written before you know the outcome, is the only honest way to learn from your own judgement.
- No model decides for you. Together they turn a big yes-or-no into a smaller, testable, reversible step.
- Eleven models and three habits. Using two of them every week beats owning all of them neatly.