I’m writing Agency Is All You Need in public — every chapter ships as an essay, gets argued with by readers, then revised into the final book (Q1 2027).
Sunday, a few minutes past ten. Abdellatif — who came to Germany from Morocco, trained as an electrician, works in Munich logistics while the state slowly examines whether his credential counts, and wants to marry his girlfriend Lea — sits at his kitchen table, his phone propped against the fruit bowl, his brother’s face pixelating whenever the connection to Izmir dips. Between him and the phone lies a sheet of paper, folded once to make two columns. Over the left column he has written geht zurück — can be undone. Over the right: geht nicht zurück.
“You’ve been carrying this question for a year,” his brother says. “What’s different tonight?”
“Tonight I stopped trying to answer it in one piece.”
He turns the paper so the camera can see. In the right column, the short list: the marriage itself. Children, if they come. In the left column, everything else — where they live, his job, Lea’s working hours, the size of the wedding, even the flat.
“Mama is imagining three hundred people in Nador. That’s not in your left column?”
“The three hundred people are the left column. The wedding is not the marriage.” He pulls a second sheet closer, covered in numbers. “Eleven thousand four hundred saved. A celebration the size Mama imagines — most of it, gone. And you know what happens after a wedding like that. The calls change. Now you have made it. The two hundred a month becomes an expectation of four.”
“So you’re calculating whether you can afford Lea.”
“No. I’m calculating which of my fears are load-bearing. Whether I want to marry her — I don’t need a paper for that. I’ve known it longer than I’ve known the word Anerkennung. What I couldn’t see was everything stacked on top of it: the wedding, the money, her family, my family, whose country, whose rules. It arrived as one block, and the block was too heavy to lift. So I’m taking it apart.”
“And the right column doesn’t frighten you?”
“The right column is the only part I’m sure of. It’s the left column I’ve been afraid of. For a year I was afraid of things that can be undone.”
You only get one run
Most of what this series has offered so far — reading signals, building financial runway, mapping your institutional room — improves your information. This essay is about the moment information runs out, which it always does, exactly where the big decisions live. Marry or don’t. Leave or stay. Specialize or keep broad.
The instinct, especially in analytically trained people, is to answer uncertainty with more analysis: another spreadsheet, another six months of watching. For life-scale decisions, that hits a wall no diligence moves. Kahneman and Tversky spent careers documenting how badly human judgment performs under uncertainty, and beneath their catalogue of biases lies a structural point: where you have no distribution of past outcomes to learn from, prediction has nothing to stand on. A claims department can price a hundred thousand policies because it sees a hundred thousand outcomes. You get one marriage to this person, one exit from this employer, one life running forward at one speed. There is no base rate for your decision, and no second run to compare against.
The consequence is not despair but a change of target: if you cannot make the forecast better, make the decision structure better — build decisions that survive many futures instead of betting everything on one predicted future. That structure has five moves.
Separate what can be undone from what cannot
The first and most powerful move is the one Abdellatif made with a folded sheet of paper: sort by reversibility. The sorting question: if this goes wrong, what would it cost to get back to roughly where I started?
Reversible decisions can be undone at tolerable cost — money, time, some embarrassment: renting before buying, testing a business on weekends. Irreversible decisions close doors: children, spent savings, a burned relationship, certain silences that harden into permanence.
The rule: treat them with entirely different care. Irreversible decisions deserve slowness — deliberate delay, outside views, the full weight of your attention. Reversible decisions deserve speed, because the cost of a wrong reversible decision is small while the cost of deciding slowly is constant and compounding. Most people invert this. They agonize for months over reversible moves while sliding into irreversible ones by accumulation — the years of drift that quietly close doors no single day chose to close.
The deeper insight is that big life decisions are almost never single decisions. They are bundles, and bundles can be taken apart. “Should we get married?” arrives feeling monolithic and irreversible. Unbundled, it contains one genuinely irreversible core and a thick periphery of reversible attachments — location, work arrangements, ceremony, timing. The same holds for “should I quit?” and “should we move?”. Almost always, the irreversible core is smaller than the dread. The dread belongs to the bundle; the commitment belongs to the core. Seeing the difference is half the decision.
Keep next moves alive — at a price you have checked
The second move comes from Nassim Taleb’s work on decision-making under uncertainty: optionality. An option is a position with a small, bounded downside and a large or open upside — the right, but not the obligation, to make a future move. Since you cannot predict which future arrives, keep multiple viable next moves open, and pay only what those openings are honestly worth.
Sonja — this series’ other companion: mid-thirties, ten years of doing everything right at a large insurer whose restructuring memos have been deleting the ladder above her — holds a textbook option. Her four-day work arrangement costs a known 20 percent of salary; it buys a real weekly block in which a second income can be tested, with the corporate income still flowing. Abdellatif’s credential recognition is the same shape: bounded fees and evening paperwork, a large upside — a regulated profession, better pay, mobility — and no obligation to act on any particular date.
Two honest caveats, because optionality gets romanticized. Options are not free: every open door costs money, energy, attention — an option whose carrying cost you haven’t priced is not a strategy; it’s clutter. And options expire: recognition rules change, markets fill with competitors, people wait — Lea has been waiting. Keeping everything open forever and thereby choosing nothing is not maximized optionality. It is a sophisticated costume for drift.
Small real tests beat big analyses
Third move, and the cheapest: when a question can be answered by a small contact with reality, don’t answer it with a large projection instead.
Sonja could have spent a year modeling whether regional brokers would pay for AI-workflow help. Her first small paid project answered in three weeks what the model could never answer at all, because the model would have been made of her assumptions and the project was made of a stranger’s money. A serious conversation about money and children — the kind couples defer — outweighs years of private inference about what the other person probably wants.
A good test has real stakes (someone pays, something is actually revealed), small size (failure is information, not catastrophe), a defined question (“will a broker pay a day rate for this?” — not “is my business viable?”), and honest scoring — you decide before running it what result counts as bad news. That last property is the hardest, and it points straight to the fourth move.
Tripwires: decide now what will decide later
Drift is not the absence of decisions. It is decisions made by default, on the worst possible terms: unconsciously, at the pace of habit, reviewable never. The antidote is the tripwire: a decision point set in advance, converting vague intentions into if-then structure. If X happens, or by date Y, then I do Z.
Why decide in advance? Because at the moment of truth you will be a motivated reasoner. When the buffer dips, the mind that must decide whether to cut spending is the same mind that wants the nice apartment. A tripwire is a message from your calm self to your motivated self, sent while the calm one is still in charge — the oldest trick in the literature of self-command, the mast Ulysses had himself bound to, updated for careers and bank accounts.
Three design rules. Write tripwires down; unwritten ones renegotiate themselves. Attach dates, not just conditions — conditions without dates are hopes. And pre-commit the review, not necessarily the drastic action: “if fewer than two clients by March 1, recalibrate the offer with my crew” will actually fire; “…I give up” will be quietly repealed the night before.
Commitment is not the opposite of agency
The fifth move corrects a misreading the other four invite. If reversibility is good and options are good, is commitment — marriage, a move, a specialization, a child — therefore a loss of agency?
No. Get this wrong and the whole framework curdles into a philosophy of perpetual hedging, which is just fear with a methodology. Choosing to bind yourself is not the surrender of agency; it is its fullest exercise. The point of keeping doors open is to walk through one.
And commitment doesn’t merely spend agency — it creates room to act that no uncommitted position can reach. Two people who have bound their futures together can pool runway, take turns carrying risk, and plan on a horizon no individual hedger has. Albert Hirschman observed that loyalty is what makes voice worthwhile — it’s the commitment to stay that gives you standing and reason to shape a thing rather than merely shop among alternatives. The uncommitted can only ever exit. The committed can build.
The condition — and it is the whole condition — is open eyes. Legitimate commitment means you have looked at the irreversible core directly and accepted it: the person, the vocation, the place. Not the wedding, the title, the view. The failure pattern is committing to a periphery while hoping about the core — marrying the celebration, taking the job for the prestige — because the periphery is exactly the part that will change. Which is precisely what Abdellatif’s two columns were for: certainty about the core, flexibility about everything wrapped around it.
Three failure patterns stalk these moves — readers of my earlier book The Tautai Principle will recognize them as the traps of organizational strategy, translated to a life. Waiting for certainty: uncertainty at this scale is not a fog that lifts but a condition of the terrain; “when it’s clear” means never. Copying other people’s scripts: the cousin’s wedding as your template, the influencer’s exit as your plan — but scripts encode their author’s constraints, not yours. Projecting the trend straight: real change arrives in jumps and reversals; plans that require a trend to stay linear are predictions wearing a plan’s clothing.
Two decisions, one structure
Now watch both personas run the toolkit — on decisions that could not look more different and are structurally twins.
Abdellatif unbundled the marriage question and did what the analysis pointed to. Irreversible core: the commitment to Lea; children, if they come; and — semi-irreversible, worth naming honestly — the joining of two family systems, because you can end a marriage but in-laws never fully unweave. Reversible periphery: city, employer, working hours, the flat, and above all the wedding itself. From there, a staged plan. Civil marriage in Munich within the year — small, cheap, the core made real. The celebration in Nador the following summer, its scale a negotiation rather than a silent debt: a fixed budget, roughly a third of his savings, with a floor under the rest — six thousand untouched, the runway the new household starts with. The remittance question he decided before the wedding could decide it for him: a fixed monthly amount he and Lea would set together, communicated as a commitment rather than left as an open valve — because an unbounded expectation is a tripwire someone else installs in your life. Then he decided to propose. Not because the risks had disappeared — his status, the recognition timeline, the cross-pressure of two families remained exactly as real. Because he finally knew which risks he was accepting, and had chosen them.
Sonja ran the same structure toward the opposite move: not binding herself, but exposing herself. Her exit is designed to be reversible at every stage short of the last. The four-day arrangement — won after two meetings and one predictable speech about commitment — can be walked back. The side business runs as a sequence of tests, each with a defined question: will a second broker pay the day rate; will the first one renew. And the tripwires went on an index card, dated, taped inside a kitchen cupboard where visitors wouldn’t see them and she would. If the buffer falls below twelve months: pause discretionary spending on the business; review. If three paying clients and two months of pipeline: give notice. The card’s real function is that “Should I quit?” — the huge, unanswerable, midnight question — no longer exists in her life. It has been decomposed into small dated questions, each of which reality itself will answer.
His decision binds; hers exposes. But the structures mirror: both broke one paralyzing question into an irreversible core held with care and a reversible periphery run as experiments, and both wrote down what would trigger the next move rather than trusting their future motivated selves. Commitment and exposure are the two honest ways forward under uncertainty. The dishonest third way is drift — and drift, remember, is also a decision: the decision to let defaults, other people’s scripts, and the passage of time make your irreversible choices for you.
The left column
Two weeks after the night with the two columns, Abdellatif calls Izmir again. This call is short.
“April,” he says. “After her father’s birthday. There’s a place up the Isar she likes, where the river splits.”
His brother laughs, warm and disbelieving at once. “You made a project plan for a proposal. Staged rollout. My brother the engineer.”
“Other way around.” Abdellatif looks at the sheet, still on the table, softened from handling. “For a year I had a project I was hiding inside. Analysis, scenarios, one more thing to check. The plan isn’t the caution, Idris. The plan is what it took to stop being cautious about the wrong thing.”
A pause on the line, the particular pause of an older brother deciding not to make a joke.
“And the three hundred guests?”
“Are a negotiation. Next summer. Mama will get a number from us, and it will be a real number, and she will be angry for a month and then she will start cooking.” He smiles. “The left column, brother. It all goes in the left column.”
After they hang up he sits for a moment in the kitchen’s quiet. Down the hall, Lea’s light is still on. Nothing is certain — not the recognition, not the timeline. He has simply, finally, sorted what he is sure of from what he isn’t, and found that what he is sure of is enough to move.
He folds the paper along its worn crease and puts it in the drawer, where — though he doesn’t know it yet — Lea will find it on a Thursday in March.
Sonja and Abdellatif are constructed figures — personas built to carry real mechanisms, not case studies of real persons. The book’s method note explains how they’re used.