Poor leadership and personal folly begin with, naivety, a lack of epistemic humility, an overestimation of one’s own abilities and contributions, as well as no feedback loops. This essay is about poor leadership, and systems by which people in positions of authority lose the ability to perceive the work they oversee, and the apparatus that protects them from finding out.
Assembler or Specialist
There is a real distinction between assembly and specialist work. Some labour is interchangeable - executing a documented process, operating known machinery. Some labour is not - designing the process, building something that did not exist yet, anticipating failure modes that have not happened.
Software has both. Gluing commonly-used APIs together to ship a feature is closer to assembly. Among other things, designing the system, choosing the abstractions, deciding what not to build, holding years of context about why the codebase looks the way it does - that is specialist work. It’s where the skill is significantly less interchangeable, and it compounds with experience.
The danger is when leadership cannot tell the difference. A leader who does not understand the work treats the people doing it as assemblers - fungible, replaceable by AI, doable by anyone with time and privilege. They are projecting their own ability to imagine doing it onto the people actually doing it.
This is what makes the AI-replacement narrative dangerous, and it has nothing to do with whether AI is actually good enough. AI can plausibly replace assembly. It cannot replace architecture, judgment, taste, or the compounded context of someone who has shipped real systems for 5+ years. But the replacement decision is not made by AI. It is made by the person in charge, who often cannot tell which kind of work they are looking at. The risk is not that the tool is good enough - the risk is that your boss thinks it is.
When the person above you believes you stick graphite in wood - that the work is trivial, that they could do it themselves if they had the time, that your experience is a kind of luck rather than a kind of accumulation - your job is in danger regardless of whether the work has actually become trivial.
The Apparatus That Removes Consensus
Every quarter, leadership faces the same question, dressed up as a technical one: which line items currently paid to humans could be automated? The pressure hits at every level - founders deciding which roles to renew, executives choosing what to outsource, middle managers deciding whose work the new tool absorbs, team leaders pushed to deliver more with less. Technological acceleration has made the question permanent. The pressure comes from stakeholders - who, in many cases, are the people answering the question (leadership positions often come with equity) - and it converts into an urgency that does not distinguish between the decision being good and the decision being made.
The standard defence is that those making the call carry the risk. In a stable environment, that holds. In one defined by constant re-calibration - genuine disruption, misinformation, memetic narratives about disruption, leaders slipping into media or AI-induced psychosis - the risk does not sit where the reward does. It sits with the people below, whose livelihoods depend on whether the person above them is reading the situation correctly or just reading it confidently.
A contract I was once proposed contained a clause stating that my performance would be evaluated against a list of reasonable metrics and “prevailing economic circumstances.” The metrics were specific but he economic clause was not. What the clause established was a unilateral option for the employer: if conditions soured, my evaluation could absorb the blame for events I had no hand in causing, and the employer could define those conditions after the fact as a result of my performance. What’s disrespectful is the implication that whatever skill I brought could be discounted by a factor the employer would define in their own interest, with no symmetric clause running the other way. The risk-reward asymmetry was written into the contract before I started.
Mass layoffs are framed as a difficult executive skill: reading the spreadsheet, identifying redundancy, communicating with affected employees, holding morale together among the survivors. This is the LinkedIn version. The version that actually runs is different.
People under quarterly pressure surround themselves with others who will not contradict them. The in-group’s function is to convert decisions into consensus before the decision has to defend itself. By the time a layoff plan reaches execution, the question of whether it was the right call has already been answered - not because anyone interrogated it, but because no one in the room had the standing or the incentive to.
Human Resources, historically, among other things did two things: hiring and payroll. Payroll has been absorbed by software. Hiring has been outsourced to recruiters and platforms. What HR does now, in most companies, is mediate conversations and internal conflicts, exercise authoritarian ‘diplomacy’, manage non-combative psyops, and send emails authored and authorised by others. It is a procedural shell - a function that lends the appearance of process to decisions made elsewhere, by people whose names will not appear on the email.
This is what doing layoffs as a ‘skill’ actually amounts to. The judgment was outsourced to an in-group that doesn’t disagree, executed through an HR function that doesn’t author, and absorbed by employees who signed contracts written to absorb it. The skill is designing a unilateral system of minimal friction, so a decision can float through without deeper consideration, and the thing being praised as competence is the systematic removal of every person and process that would have made the decision harder.
What the apparatus produces, where no one stops it, is a decision the people in charge could not defend if forced to. Companies fail and layoffs happen - that is not the issue. The issue is the framing the apparatus enables: that the cut is costless, the work was fungible, the original hire was a luxury rather than a necessity. The same person who hired a specialist because they could not do the work themselves declares - when the budget tightens - that the specialist is now replaceable. The same approvals that signed off on the hire sign off on the layoff without confronting the contradiction. The skill has not become less rare. The market has not produced a tool that erases years of accumulated context overnight. What has changed is the willingness to admit the work mattered.
If the work was fungible, the original hire would not have been necessary. The hire was an admission of skill the leadership didn’t have.
What Walks Out
Builders who get laid off do not stop building. They take their context with them - the compounded judgment about what to build and what to abandon, why the codebase looks the way it does, which failure modes are theoretical and which have actually happened at 3 AM. None of that is on the spreadsheet, and much of it doesn’t transfer in a handover document. It walks out the door and accumulates somewhere else.
Where it accumulates is mostly invisible, some join competitors, some go to startups, a few start their own - most disappear into other jobs, taking the institutional knowledge with them.
That export is the actual cost of the layoff. It does not appear on the quarterly report. The next time the organisation needs the knowledge - because the system has degraded, because a decision needs the judgment of someone who remembers why the previous one failed - it will be paid for again, at market rate, often through consultants or rebuilds that cost more than the salary deemed too expensive. Or at the cost of losing customers. The labour was the knowledge. So when the labour goes, the knowledge goes with it.
Side: This is what’s called context debt - AKA rationale debt or architectural amnesia. It’s much like a Chesterton’s Fence scenario. The distinction worth holding: technical debt lives in the artefact (product) and system that produces it, context debt lives in the absence of something in the artefact or system that produced it. Technical debt taxes every change continuously and can be measured (cycle time, defect rate, etc). Context debt is invisible until someone tries to make a decision that requires knowing why.
The Standing Defence
When pushback arrives - from a former employee, a critic outside the room, a colleague who refused to go along, a stakeholder asking hard questions - the response is rarely engagement. The response is reframing. The critic does not understand. They are not in the arena - the criticism is dismissed not on its merits but on the standing of the person making it.
This works in the short term. It also forecloses the feedback that would have improved the work. A leader who treats every critic as someone who doesn’t get it eventually stops getting any signal at all - they are surrounded by people who agree, by metrics they themselves selected, and by an internal narrative that hardens with every quarter the results hold. The harder things get, the more ‘visionary’ and ‘genius’ they must be - because the alternative explanation is that they were always running on grit and luck and getting both confused with insight.
Ask such a leader what evidence would change their mind, and there is rarely an answer. The category does not exist. Every contradiction can be reframed as misunderstanding, every loss as bad timing, every defection as disloyalty. Confidence and correctness become interchangeable in their own minds - not because the work has earned it, but because nothing has been allowed to count against them. This is all a result of miscalibration and drift.
What This Costs
Another contract I was proposed contained a clause stating the company would own all software I produced - inside or outside of work hours. The wage covered the role. The IP claim covered everything I might write with the skills the role required, regardless of where, when, or for whom. What I built on a Saturday for a personal project would belong to the company by default. There was no mechanism for valuing what was taken, and no symmetric clause running the other way. The credit transfer was being extended past the labour relation itself - into anything I could build, with the very skills the company had hired but not taught me.
In the case of poor leadership, credit for work done is taken - and the contracts are designed to facilitate the taking. Every hire is a wage-credit transfer. The byline on the work, the institutional memory of who solved what, the place in the story when the win gets retold - these were allocated by the same hand that signed the cheque. The wage is the price of the credit transfer, and the transfer is one-way because the contract was structured to make it one-way. The leader who later describes themselves as the builder is not lying about what the contract says; they are reciting a contract they helped shape. This is what makes the misattribution structural rather than incidental. The hubris is not in noticing the credit had to land somewhere. It is in pretending the landing site was an accident.
Good leadership writes contracts differently. It distributes credit downstream - bylines, equity, attribution, visibility - toward the people whose labour produced the win. It treats the credit as a settlement to be negotiated, not a verdict to be imposed. Poor leadership writes contracts that concentrate the credit upward, then defends the concentration as the natural shape of things. Having taken the assigned credit, the leader loses the ability to notice the assignment was a design choice. They cannot tell which of their decisions were skill and which were merely exercised positions of authority - because the position was structured to make the two indistinguishable. Each successful round of layoffs facilitate by an apparatus the removes consensus is taken as evidence of good judgment rather than evidence of structural asymmetry. The same apparatus that took the credit of people’s contribution upward to pure company IP no longer attributed to the original contributor is the apparatus that prevents the leader from evaluating whether they earned it.
The asymmetry pays. Where it runs, leaders and the company absorb credit for work done, severance settles contracts, the apparatus removes consensus from the cut, and critics get dismissed for not being in the arena. Each step is locally rational.
What it produces, where no one stops it, is a class of leaders who do not know what they actually built. They cannot tell which of their decisions were skill and which were luck. They cannot tell which of their hires were necessities and which were luxuries, because the apparatus has trained them not to look.
This is structural hubris. Not the personality of any individual in charge, but the emergent and predictable epistemic arrogance of people in a position of authority that has been engineered to remove every form of friction that would correct them. It is what happens when the seat itself is built to absorb credit, defer accountability, and dismiss the feedback that would force a reckoning. No wonder so many of the world’s leaders are narcissistic, coercive, and engage in misdirection - the seat selects for what survives in it.
The hire was an admission, the layoff retracts it for a formula, the credit stays misattributed and absorbed as company IP that came from nowhere. That is the steady state for any leadership that runs on such an apparatus.