Education and career

Credentials are access tokens

A credential carries two kinds of value — the real capability it may represent, and the signal it sends to someone deciding whether to look at you at all. Entry tends to run on credential plus referral, in that order of screening and the reverse order of effectiveness: the credential gets you past the filter, the referral gets you the job. Some crowded roads stay correct anyway, because their value never depended on scarcity — a licence for a regulated profession still opens the door however many people hold one. But where value did depend on scarcity, inflation is real: the degree that once distinguished now merely qualifies.

The unwritten local rules

• The org chart describes reporting, not influence. The real map is who can get a decision reversed — usually learnable by watching whose objections cause meetings to be rescheduled.
• A performance review is a budget document written in the language of feedback. The distribution of ratings is often fixed before anyone is assessed, which does not make the feedback false so much as non-diagnostic about you in particular.
• Surprise is the thing managers are actually punished for. Most of what is called “managing up” is simply never being the reason your manager was surprised in front of their manager.
• “We’re like a family here” predicts worse severance, not better. Families do not have notice periods.

Two more are worth stating because they are the ones nobody says out loud. Almost every workplace has one person the chart undersells — not the most senior, but the one whose absence stops several things at once, usually because they hold context nobody wrote down. Finding out who that is, in your first month, is worth more than reading the handbook. And a complaint travels upward as information about the complainer at least as much as about the complaint, which is unjust and is also simply how the channel works: the person receiving it learns one new thing about the subject and one new thing about you. That is not an argument for silence. It is an argument for knowing the price before you pay it, and for choosing which complaints are worth it.

And one switch worth marking, because it happens without an announcement: your employer is a place you work — colleagues, a craft, somewhere to be — right up to the moment it becomes a counterparty with its own incentives, which is usually a severance conversation, a grievance, or a contract. Both descriptions are true; only one of them is true at a time, and reading the second situation with the first one’s assumptions is expensive. The same switch, and how to notice it, is on the health guide, where the clearest case lives.

Standing does not leave the building

The most valuable thing most workers own — credibility, the belief that your estimates are true — is entirely non-portable. It resets to zero at the boundary of the organisation. Knowing that the reset is structural, and not a sudden onset of incompetence, is what stops a capable person from concluding they have gone bad in the first six months of a new job. It is also why credibility is worth building before visibility: credibility buys attention reliably, and attention does not buy credibility at all, so spending the second before you have the first just makes it more expensive.

A related trap is legibility. An organisation can see shipped work with your name on it and decisions attributed to you; it cannot see the escalation you defused, the outage that did not happen, or the colleague who did not quit because you talked to them on a Thursday. That work is real, load-bearing, invisible at review, and unevenly distributed — which turns an accounting problem into an equity one. Exit is asymmetric for the same structural reason: the organisation can end the relationship in an afternoon, and you generally cannot, because your side needs a next position that takes months. Slack is what flattens that asymmetry.

The reason the reset is structural becomes obvious once you stop thinking of a reputation as something you have. It is not a quantity attached to you; it is a set of copies held in other people’s heads, each written at whatever moment they last paid attention, and updated only when that person meets new evidence or hears from somebody who has. You have no write access to any of them. Four things follow immediately. Every copy is stale, which is why somebody who has genuinely changed keeps meeting the old version of themselves in other people’s expectations — the change has to be re-transmitted, not merely made. The copies do not travel. Cross an organisation or a border and there are no copies there at all — not bad ones, none — so every claim now needs evidence attached. Damage propagates faster than repair, because one vivid negative event updates many copies at once and positive evidence updates them one at a time. And most entries are written by third parties rather than from direct evidence, which means the people who talk about you shape more of it than the people who know you.

This is also what separates an audience from a network. An audience is a large number of thin, stale copies held by people with no reason to act; a network is a small number of current ones held by people who might. They are different objects, they are acquired differently, and treating the first as though it were the second is the commonest disappointment in a career built online. The model describes how impressions travel — nothing more. It says nothing about whether any copy is accurate, and plenty of them are not.

There is a matching effect on the social side, and it is the one people take personally: a capable adult moves to a new organisation and becomes, for a few months, socially clumsy — reading the room wrong with complete confidence, because the pattern library that was trained somewhere else keeps firing. Why that happens, and what retrains it.

The meta, and why it degrades

the prevailing strategy degrades by being followed: a popular strategy divides its returns among everyone running it. The crowding arrives in a recognisable order — the strategy gets a name, the name gets courses sold about it, and the courses get sold by people whose success came from selling courses; by then the returns have moved. What this does not license is reflexive contrarianism, which is just the second-most-crowded strategy wearing a leather jacket. The point is not to flee crowds but to price them. The two reliable exits from meta-chasing are strategies whose returns are internal — loving the work itself — and genuine comparative advantage: the intersection of what you specifically are, which no course can teach and no crowd can arbitrage away.

Three shapes of a learning curve

“Steep learning curve” usually just means hard. The shape matters far more than the steepness, because it sets what the first few weeks are supposed to feel like. Front-loaded: fast early gains, then a long slow refinement — cooking, driving, conversational language, most practical skills. A modest investment gets you most of the useful part, and for a skill that serves something else that is genuinely enough. Threshold: nothing visible happens for a long time while foundations go in, and then it accelerates — reading music, an unfamiliar writing system, a whole new field. Step-function: flat stretches broken by sudden jumps, with the flat parts doing the structural work that makes the next jump possible.

The common error is quitting during a threshold curve’s slow start, having read the absence of progress as evidence the thing does not respond to effort. It is the same error in a step-function’s plateau. Both are worth knowing before you start rather than after you stop. And skills compound: one that connects to what you already know is worth more than its face value, so what to learn next is better answered by “what does this connect to” than by “what is most useful on its own”. Note which part of this is reversible and which is not — stopping is entirely reversible, and the months are not, which is the real argument for finding out the shape early.

The school’s rules that do not generalise

School teaches a set of rules by structure rather than by instruction, and they are local. That effort is assessed: it is there, and almost nowhere afterwards, where outcomes and visible artefacts are assessed and much of the effort is invisible. That work arrives from an authority, with a specified standard and a deadline: adult work usually has to be defined before it can be done, which is why people who were excellent at school so often find the transition harder than people who were not. That the criteria are published and honoured: rarely true elsewhere, where written criteria describe a decision rather than make it. That asking for help is a concession: structurally reinforced by individual assessment, and one of the more expensive habits to unlearn. And that your cohort is your comparison set: a fixed group of the same age, ranked continuously, which is not how any later setting works at all.

None of that was a lie. It was accurate for one place, and the transition out is where most people first discover that the rules were local rather than general — which is the cheapest available fix for “I was excellent there and I am struggling here”.

Changing direction is not starting over

A career change is not a return to level one; your base stats come with you. What transfers almost fully are the operating-system capabilities — how you learn, how you handle stress, how you communicate, how you think about a problem — which is why an experienced person changing fields is dramatically faster than a true beginner. What does not transfer cleanly is domain-specific knowledge, institutional relationships, credentials tied to the old field, and — most painfully — the status and recognition you had earned. So the real questions are which of your skills transfer, and what the change costs, denominated in time, income during the gap, the energy of learning while carrying an identity change, and social capital that has to be rebuilt.

Not every dissatisfaction is a signal to change: boredom can be a plateau, exhaustion can be broken energy management, restlessness can be a local-maximum problem where the way up first requires a step down. But sometimes the signal is real — a values mismatch that will not resolve, a ceiling that is structural rather than personal. The most reliable tell that a decision is being made by the past rather than the future is sunk cost. The question underneath all of it is whether the cost of staying now exceeds the cost of changing, and that calculation is personal.

Try this as a decision — the campaign runs twelve years of these tradeoffs under a budget that does not stretch.

One home per idea. Every mechanism on this site is explained in exactly one place, and everything else links to it. If you find an explanation here that exists nowhere else on the site, that is a bug — tell us on the corrections page.