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).
The offer came on a Tuesday, in the glass corner office on the fourteenth floor, cathedral spires visible past the construction cranes. Sonja’s mentor slid a single page across the table: a role profile. Transformation Office. Reporting line: Board Member Operations. Grade increase effective Q3.
“You’d be on the inside of it,” he said. “Whatever happens to claims, to underwriting support — you’d be one of the people deciding how it happens, not one of the people it happens to.”
Sonja is in her mid-thirties, studied insurance economics, and has spent over a decade doing everything right at a large Cologne insurer whose restructuring memos have been quietly deleting the career ladder above her. By every metric she had used for twelve years, the page was a promotion. It was also a role whose entire purpose was to manage the shrinking of the departments she had been groomed to lead.
“I’ll think about it,” she said. “Thank you. Really.”
On the tram home she did what she had trained herself to do this past year: check the numbers instead of her feelings. Fourteen months of runway — the time her savings buy at current spending. One paying side client, a broker in Leverkusen who had paid her 1,800 euros to review his claims workflow and was already asking about a second project. A folder on her laptop named, without irony, Exit — options: three scenarios, a pricing sheet, forty regional brokerages, a draft email to her manager about reducing to four days.
The folder’s last-modified date was five weeks old.
That evening, role profile on one side of the kitchen table and runway spreadsheet on the other, she caught herself feeling something she had to sit with before she could name it. Relief. The offer meant she didn’t have to open the folder — meant years more in which every question about her future had an institutional answer. And the relief was the most alarming data point of her year, because nothing was forcing her to stay. Skills rebuilt, buffer built, first client tested: every dimension she knew how to measure said you can move.
And she was, she realized, about to say yes anyway.
The dimension you can’t see on a spreadsheet
The book this essay comes from works with a three-part model of agency — the capacity to direct your own course. Capability is what you can do: learning, judgment, working with intelligent machines. Room to act is the objective space you have: money, rights, networks. Sonja has done the work on both. Her skills are current, her runway is real, her options are genuine.
And she is not moving.
This is the third dimension, the one neither a curriculum nor a bank statement can supply: will — the disposition to act. Ownership of your course, tolerance for uncertainty, the readiness to trade the illusion of security for real options. The model’s interaction rule says agency is limited by its weakest dimension: capability without room is frustration; room without will is drift. Sonja is drifting in the most comfortable way imaginable — inside an institution that keeps handing her reasons to postpone the decision she has already, quietly, made.
It is tempting to call this a failure of nerve. It is not: what is happening to Sonja is lawful, following well-documented features of human judgment and the way institutions confer identity. What follows is why capable people with real options stay in situations they can see are declining — and it ends not with courage in the cinematic sense, but with a change of frame.
Losses loom larger
In the 1970s, Daniel Kahneman and Amos Tversky showed that people do not weigh gains and losses symmetrically: losses loom larger — in their experiments, roughly twice as heavily as equivalent gains. And people evaluate outcomes against a reference point — usually the status quo, or an expectation carried for years — not against an absolute standard. Apply this to a life decision and the mechanics of Sonja’s paralysis become visible.
Leaving the insurer is one visible loss. The title disappears on a specific date. The badge stops working. Colleagues know; her parents know; her old high-potential cohort knows. The loss is singular, attributable, and public — everything the loss-averse mind is built to avoid.
Staying is many invisible losses. Her market value drifts a little further from her internal grade each year. Her skills specialize toward processes that exist only inside this one company. Her network narrows to people whose fates are correlated with hers. Her options expire quietly, one by one; nothing is ever lost on a particular Tuesday.
So her intuition’s arithmetic is exactly backwards. The slow bleed feels safer than one clean cut, because a bleed is never a headline. A career can lose more value in five years of comfortable drift than in one badly timed exit — but the drift has no moment of pain, the exit has one, and the mind prices the moment, not the total.
A second distortion sits on top. Sonja’s reference point is not her actual situation; it is the trajectory she was promised in 2019, when she was selected for the high-potential program. Measured against fast-track Sonja — the department head she would be by now, in a version of the industry that no longer exists — every realistic move reads as a loss. Four days a week? A loss. Independent consulting? A loss. She is grieving a comparison point, and the comparison point is fictional. The most useful act of will available here is brutally unglamorous: reset the reference point to reality. Not “what am I giving up relative to the career I was supposed to have,” but “what are my actual options, from where I actually stand, in the industry that actually exists.”
“I am someone who was chosen”
Loss aversion explains the flinch. It does not explain the shame. For that, look at what twelve years inside have done to her identity.
Sonja’s sense of who she is has been fused with a conferred role. She was selected — assessed, ranked, chosen, mentored on the executive floor. “High potential” is not something you do; it is something an institution declares you to be. That is precisely its danger. Status conferred by a system can only be maintained inside that system: leave, and you don’t take the status with you — you leave the audience that produces it. This is why stepping sideways feels not like a strategic move but like a demotion of the self — before the cohort, the parents, the mentor who invested in her, and above all the internal narrator that has been saying I am someone who was chosen for over a decade.
Notice what the fear actually is. Not poverty — her runway rules that out. Not incapacity — she has a paying client. It is fear of being seen differently: recalibrating, off the track, ordinary. Status loss is processed in the same register as material loss, and loss aversion applies with full force.
Albert Bandura’s work on self-efficacy adds the final gear. The belief that you can execute what a situation demands is largely domain-specific and built primarily from mastery experiences — evidence of having done the thing. Sonja has twelve years of that evidence inside a corporate structure and one data point outside it. Her mind reads “little evidence” as “little ability” — a compelling error with a stern implication: the confidence she is waiting for cannot arrive before the actions that would generate it. In new territory, evidence follows action; it never precedes it. Waiting to feel ready is waiting for a delivery from a company that only ships after you’ve moved.
The cocoon
We can now say precisely what the cocoon is: the corporate or institutional shell that feels secure while quietly narrowing your options. Three features make it effective.
The door is open. A cocoon is not a prison. Nobody is keeping Sonja; she could resign tomorrow. The open door is what makes the arrangement so stable — it converts staying into an apparently free choice, renewed daily, so there is never a moment of coercion to push against. Prisons produce escape attempts. Cocoons produce gratitude.
Comfort compounds while options decay. Every year inside, the internal life gets marginally easier — seniority, routines, people who know your name — while the external position gets marginally weaker. The curves diverge slowly enough that no single year forces the comparison. The exit price rises at the exact pace at which the exit stops feeling necessary.
It retains through generosity, not confinement. The Transformation Office offer is the cocoon operating at peak performance. At precisely the moment a valuable person accumulates the means to leave, the institution produces a reason to stay — a title, a project, a signal of continued chosenness. The mentor likely means it kindly; institutions don’t need to intend the effect to produce it.
None of this makes staying wrong. Cocoons are genuinely comfortable, and for some people, at some stages, the trade is worth it — if made with open eyes. The failure mode is not staying. It is staying by default while believing you are still deciding.
The adventure stance
So what is the honest alternative? Not the movie version — not the desk quit, the leap, the burned boat. That is will without capability, which the interaction rule identifies as recklessness, and it fails at roughly the rate you’d expect.
The alternative is a stance: treating your working life as an expedition rather than an escalator.
The escalator — the linear life script, the assumption of steady, predictable ascent — carried an entire logic of evaluation. Progress meant altitude. Standing still meant failing. Stepping off meant falling. For two or three generations in wealthy countries this described many careers well, which is why the script sits so deep. But when the escalator itself is being dismantled — the first essay in this series laid out why — the script becomes actively misleading. It keeps scoring you by altitude on a machine that no longer goes anywhere, and it registers every lateral move as a fall.
An expedition runs on a different logic. The route is discovered en route. Planning still matters enormously — but as preparation, not prediction: provisions (runway), navigation ability (sensing, learning), checkpoints at which the route is reconsidered. Changing direction when the terrain demands it is not failure; it is the core skill. The relevant questions are not “how high am I?” and “am I ahead of my cohort?” but “am I oriented?”, “am I provisioned?”, “is my next move sound, given what I now know?”
Read Sonja through each frame and the same facts invert. Escalator: she is thirty-four, stalled, about to slide backward, and the Transformation Office is a rescue. Expedition: she is thirty-four, provisioned for fourteen months, newly re-skilled, carrying one verified discovery about outside demand — and the Transformation Office is a comfortable camp on a melting glacier, occupied mainly because leaving camp is cold.
Two clarifications, so the stance isn’t mistaken for its self-help imitation. It is more demanding than the escalator, not less: the provisioning, sensing, and course-setting the escalator did for you are now yours. And it promises nothing. Expeditions fail. It guarantees only that you are the one navigating — the difference between having a course and having a conveyance.
Will is not free
Everything above carries a risk: it can sound as if will were a decision one simply makes — as if the reader stuck in place should just want it harder. That would be false, and the falseness matters.
Will is a disposition, but it runs on capacity, and capacity is a finite, unevenly distributed resource. Distress and anxiety consume precisely the bandwidth self-direction demands — one of the mechanisms of the agency divide I traced in Education is Broken: two people with identical options and skills, one with the internal resources to convert them, one without. Exhaustion does the same on a shorter cycle; a person working rotating shifts is not “less motivated” than a rested strategist — they are spending their executive capacity on survival. Caregiving may be the most invisible drain of all: someone managing a parent’s decline alongside a job has often exhausted the day’s supply of planning and persisting before a minute of it can be spent on their own course.
Deci and Ryan’s research on self-determination points the same way: autonomous motivation is not summoned by exhortation; it grows where the needs for competence, autonomy, and relatedness are met — and several of those preconditions are external. So when will is depleted, the moralizing question — why don’t you want it enough? — is not just unkind. It is analytically wrong: a category error that mistakes a resource problem for a virtue problem.
What does the framework say to someone in that position? Shrink the frame: bandwidth that cannot carry a five-year question can often carry a thirty-minute one, and short-range acts compound. Treat restoring capacity as the first act of agency, not a detour from it. Borrow structure: standing routines, pre-made decisions, people who hold the course when you can’t — not weakness, engineering. And do not read depletion as verdict: the exhausted mind reliably reports that its state is permanent and that its state is you, and both reports are false.
The adventure stance, in other words, is not a demand that everyone summon boldness on schedule. It is a direction to face while moving at whatever speed the terrain of your actual life allows. Some seasons, that speed is barely above zero. The stance survives that. The escalator script does not — it has no category for a slow season except falling behind.
On Thursday, Sonja asked her mentor for ten minutes and told him no.
He took it better than she expected and worse than she’d hoped. “I want to make sure you understand what you’re declining,” he said, and she said that she did, and it was true — she could have written his side of the conversation herself. The track. The visibility. The once-in-a-restructuring chance. All of it accurate, all of it denominated in a currency that only spent inside the building.
At home that evening she did something she hadn’t planned. From the shoebox on top of the wardrobe she took the letter — printed, actual paper, as it had arrived in 2019. Dear Ms. — we are pleased to inform you that you have been selected for the Group High Potential Program. She had kept it the way you keep a diploma. She read it at the kitchen table and waited for the old warmth, and instead noticed, for the first time, that the letter wasn’t about her at all. It described a program, a process, a set of stages. Her name appeared twice. It was a boarding pass, and she had spent five years treating it as a portrait.
She didn’t throw it away. She put it back in the box — the right register for a document from an earlier expedition, when the maps were different.
Then she opened the laptop, opened the folder named Exit — options, and changed its name. Not to anything dramatic; she’d read enough of her own industry’s marketing to be allergic to drama. She named it Route, updated the draft email about four days, read it through twice, and did not send it — that was next week’s move, and there was a right way to make it. But she marked it, deliberately, the way you mark a checkpoint: this is where the route turns.
The question she’d been living inside for a year — what if I fall behind? — had quietly stopped compiling. It belonged to the escalator, and she wasn’t standing on the escalator anymore. The question that replaced it was smaller and had the great advantage of being answerable: what do I need to find out next?
Sonja is a constructed figure — a persona built to carry real mechanisms, not a case study of a real person. The book’s method note explains how the personas are used.