Who is most exposed to the harm

The same three conditions apply to everyone inside an HDM organisation: a foreclosed response, a threat that cannot be anticipated, and an exit that costs more every year. What differs is how much of each a person gets. What they perceive sets how often the foreclosure reaches them. What their position is answerable for sets how much of the threat lands on them. What they have accumulated sets how tightly the exit holds them. All three rise with years in one place. They converge in the decade when the load outside work is heaviest.

The traits that became liabilities

All the traits and behaviours that had served Robert well became liabilities. He kept getting into trouble. So he changed how he went about things: to do the things he wanted to do, he learned to operate under the radar and avoid scrutiny. The approach worked. By his own reckoning it was very successful. It also made him smaller, and he did not get the credit he deserved for what he achieved.

His perceptiveness should have been a strength. It became a liability. Where the reasoning behind a binding choice has to be given and questioned, a person who perceives that an approach will not work and says so is the most useful thing in the room. That is what the questioning is for. Under HDM the person expressing it becomes the problem. The organisation’s rhetoric celebrates competence; its architecture punishes those who cannot pretend they do not see what is coming.

He had the playground first. The early years gave him occasions on which asking changed the work, which is what the detection of control is built from. He then entered a structure where it did not apply, and kept asking anyway. Until he no longer did. Not out of cowardice, and not out of disengagement. He stopped because holding it back was the only way left to do the work. In that single adaptation the organisation lost the perceptiveness it needed, and he lost the standing that would have come from contributing it. Performance cost and human cost are not two consequences of HDM here. They are one act, seen from two sides.

It is undeniably true, he says, that he had the problems he did because of who he is. It is not true that who he is was the cause of those problems. Characteristics determine how a structure affects a person. They do not make the person the cause.

Two timescales

Perceptiveness is the state of perceiving conditions and rules in a particular situation. Perceptive capacity is the stable underlying capability that determines how quickly perceptiveness builds from exposure to events (Verweij, 2026a; Verweij, 2026e).

Perceptiveness for this outcome produces the acute episode: this project, this failure, this attribution.

Perceptive capacity produces the chronic exposure. Someone with high capacity builds perceptiveness faster from the same events. So the episode recurs, and recurs sooner for them than for a colleague with identical tenure in the same role. Two people, the same years, different cumulative exposure — because one was seeing it coming from year three and the other from year eleven.

This yields a prediction that can be tested. Cumulative exposure to the mechanism should track perceptive capacity rather than years served, and perceptive capacity is assessable through systematic peer observation (Verweij, 2026e).

How exposed the position is

Carrying responsibility for outcomes that cannot be fully controlled is what having a job means. It is close to universal, and on its own it explains nothing. What varies is what the failure costs. What it costs is standing: a position among people who have known one another for years, in a place where the expected direction of travel with experience is upward, and where the job itself sits at the far end. People track their position in a group closely and involuntarily. A threat to it is not a small threat.

The experimental work is precise about which configuration produces a physiological response. Dickerson and Kemeny (2004) reviewed 208 laboratory studies of acute stressors and found that performance tasks raised cortisol when they were uncontrollable, or when performance could be negatively judged by others. Tasks containing both elements produced the largest cortisol and adrenocorticotropin changes and the longest times to recovery, at an effect size of 0.92 against 0.08, not significantly different from zero, where neither element was present. Their conclusion runs against the assumption that cortisol answers to stressors in general.

HDM supplies both elements, and supplies them as a standing arrangement rather than as an episode. The outcome is uncontrollable because nothing routes what a person perceives to the choice that binds them. The judgement is social because, with no reasoning on record, the failure lands on whoever stood nearest to it, in view of the people whose regard is what standing consists of. What that combination produces emotionally is shame, not anxiety. Shame is what a threat to standing produces in someone with no way to contest the judgement.

These are laboratory studies of acute exposure, and the claim in this paper is about years. They are used for what they establish: which configuration activates the response and which does not. What repeated activation does across a working life is a separate question.

Prins, Bates, Keyes, and Muntaner (2015) analysed data from 21,859 full-time workers in the United States. Around 19% of supervisors and 16% of managers reported lifetime depression, against 12% of workers and 11% of owners. Reading the pattern through Erik Olin Wright’s concept of contradictory class locations — positions combining authority over others with subordination to those above — they argued that structural position rather than demographic identity drives it. That is the middle of the chain, measured: the position with the most standing exposed and the least say in what the standing is spent on.

Robert describes losing it. In frequent reorganisations his group was an afterthought, never consulted, and once or twice left off the new organisation charts altogether. He went from being one of the go-to people in his field to being redundant and isolated. Formally he was not made redundant; he spent a year on the bench while he found himself another role.

The far end of the threat behaves the same way. Burgard, Brand, and House (2009) followed two nationally representative United States samples over periods running from three years to almost a decade, and found that persistent perceived job insecurity predicted poorer self-rated health in both, and depressive symptoms in one. Actual job losses and unemployment episodes were associated with the insecurity but did not account for its effect on health. The anticipation is the exposure rather than the event, which is why the searchlight does its work in the months when nothing is happening.

Seniority does not protect against any of this. Saud and Rice (2024), using Australian Public Service census data, found role seniority associated with higher burnout likelihood rather than lower. Structurally that is what should be expected: a senior position has more standing to lose, is visible to more people when it loses it, and is answerable for more that it does not control.

Standing is built from the same material as investment. The years that accumulate what leaving would cost are the years that accumulate what the threat can take. The second and third conditions do not vary independently.

What tenure does to the mechanism

The years did not all feel like this. The early ones felt like progress — advancing, learning, building something. The mechanism tightens later, as the investment that accumulated during the good years becomes the thing that makes leaving impossible. By the time the full weight of it is on him, he has already built his life around staying.

Ford, Matthews, Wooldridge, Mishra, Kakar, and Strahan (2014) established meta-analytically that occupational stressor-strain effects increase over time. Igic, Keller, Elfering, Tschan, Kalin, and Semmer (2017) tracked 483 employees across five waves over ten years and found that unfavourable constellations of work stressors and resources predicted poorer health and wellbeing even after controlling for current conditions. The effect is chronic, and it does not fully reverse.

The Whitehall II analysis (Bosma, Marmot, Hemingway et al., 1997) found that job strain assessed at two time points had a cumulative effect on coronary heart disease risk that single-time-point measurement missed entirely. Godin, Kittel, Coppieters, and Siegrist (2005) found that workers exposed to high strain at both measurement occasions showed a clear dose-response gradient across depression, anxiety, somatisation, and chronic fatigue.

At the biological level, Cuitun Coronado, Chandola, and Steptoe (2018) found that employees reporting effort-reward imbalance on two or more occasions had measurably higher allostatic load scores than those who never reported imbalance. Repeated exposure shows in the body. Years of HDM exposure do not add up to a proportional multiple of the harm of one year. The harm compounds. The source does not change, so individual-level interventions such as awareness, resilience training or mindfulness programmes address the response without removing the cause.

Exposure also does not stop at the exit. Robert eventually acquired a manager whose conduct was abusive by any measure, which added a second source of harm on top of the one already running. He left, and then left two more employers, and by the end he was so punch-drunk that he no longer knew who he was or what he stood for. He describes making poor choices in that period because he had lost his sense of himself and was deluded about the resources he actually had available. The damage travels.

What leaving takes

What leaving costs is one question. Whether the person has what leaving takes is another, and it is different in kind.

The mechanism described in this paper tightens with tenure, which places its heaviest weight on people in their forties and fifties. That is also when demands outside work tend to peak: children still at home, parents beginning to need care, and the losses that start arriving in that decade. In the Netherlands, one in five people in their fifties is in the sandwich generation: parents aged eighty or over and children still living at home (CBS, 2024). Separately, 2.7 million people combine paid work with informal care (De Boer, Zegers, and Broese van Groenou, 2026). Lei, Leggett, and Maust (2023) found that a quarter of Americans caring for an ageing parent were also caring for a child under eighteen. They spent about as many hours on the elder care as everyone else, with the child care sitting on top of it rather than in place of it. They were more likely to be in paid work (69.4% against 53.9%). And they reported substantial financial difficulty at roughly twice the rate (23.5% against 12.2%), scoring higher on caregiver role overload.

Hobfoll’s framework supplies the connection. Resources depleted in one domain reduce the capacity to absorb loss in another. Changing employer is itself a resource-intensive act: it costs energy, attention, risk tolerance, and the willingness to be new at something. A person carrying a heavy load elsewhere may face a bearable cost of leaving and still not have what executing the change requires.

So the two curves align. The point at which exposure has accumulated most is the point at which the person has least capacity to act on it. This is a resource statement, not a statement about weakness, and it is testable: exposure should be hardest to escape precisely where concurrent load is highest.

Robert names a version of this that does not come from outside at all. The low confidence and low energy the situation itself produced were what he had to conduct a job search on, at the point when searching had become necessary. The condition supplies the reason to leave and takes away part of what leaving runs on.

The career literature documents what follows. Verbruggen and De Vos (2020) theorise career inaction, the failure to realise a desired career change. They identify job embeddedness (Mitchell, Holtom, Lee, Sablynski, and Erez, 2001) as an inertia-enhancing condition: the pay-offs for staying raise anxiety about loss, and the social norms around a stable career make departure costlier still. They also identify what comes afterwards: people who did not act tend to blame themselves for not having acted while they still could. That is a second self-attribution laid over the first. Not only did I fail at the work; I failed to leave.

Why nobody above sees it

If the harm is this predictable, why does nobody above notice?

The answer is structural, and it is not indifference.

Seeing what state someone is in is an act of perception. It draws on the same capability as seeing that an approach will not work: perceptiveness of human and organisational conditions, built up from exposure to events. What HDM rewards on the way up does not build it. It rewards issuing choices confidently without having to account for them, and it turns what the people closest to the work perceive into advice, which can be set aside without anyone having to say why. Someone who advanced through that is not being unkind when they do not see what Robert sees. Nothing on the way up required them to build that perceptiveness, and nothing in the position requires them to bring it to bear.

Robert gives two reasons of his own. The first: there was nothing on the surface to notice. He largely defined his own job, was a subject expert, and knew how to work the system, so the work kept coming out at an acceptable level while what it was costing him stayed out of sight. The second is that people often are not paying attention, and would rather not address what they notice. His summary: he kept hoping someone would notice, and no one ever did.

The first reason closes a loop. The architecture forced the adaptation: asking had stopped working, so he went under the radar and worked the system. That adaptation is what held the output up. The organisation had already lost his perceptiveness; what it had not lost was his throughput, and throughput is what it was watching.

This runs straight back into forced internal attribution. The absence of any response from above is not read as a perceptual limit in someone else. It is read as confirmation. If nobody above is troubled by this, perhaps there is nothing to be troubled by, and the problem is me.