Experienced-worker injuries
A Socratic walk-through of experienced-worker injuries — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why do the most experienced workers get hurt doing things a beginner would never risk?
Every trade has the story. The man who had run that machine for twenty years reached in without isolating it. The electrician who could have recited the lock-out procedure in his sleep did not use it. Everyone says the same sentence afterwards: he of all people knew better. And he did. So what is it that competence changes, if it plainly is not knowledge?
Reasoning it through
REASONING #Begin with what a novice actually does. He does not know which parts of the task are dangerous, so he treats all of them as though they might be — slowly, by the book, with someone watching. That caution is not wisdom. It is the expensive default of somebody who cannot yet tell signal from noise, and it happens to cover the real hazard because it covers everything.
Now let competence arrive. Two things change, and they change at different speeds. Perceived risk drops fast, because every uneventful repetition is evidence — a thousand safe reaches into that machine is a genuinely strong sample. Actual risk drops too, but less, because a good part of the hazard lives in the machine and not in the man: the guard that fails, the offcut that binds, the colleague who restarts the line. Skill buys down the portion of risk that skill governs, and leaves the rest where it was.
What happens to the gap between those two curves? Ask what a person does with a margin they believe they have. Usually they spend it — on speed, on convenience, on doing the job without stopping to fetch the isolation key. That is the risk-compensation argument: safety improvements are partly consumed as performance rather than banked as safety. It shows up wherever the operator can feel the improvement, and the size of the effect is argued about, but the direction of it is not controversial.
There is a second mechanism running alongside, and it is not about beliefs at all. Practice converts a deliberate sequence into an automatic one — that is what practice is for, and it is why the expert is faster and more accurate. But the deliberate sequence contained a check step, and automation does not preserve steps by their importance. It preserves them by their contribution to the output, and a safety check contributes nothing to the output on any given day. So the check quietly falls out of the routine, not by decision, and nothing signals that it has gone.
Now hold the story to the data, because this is where the folk version overreaches. In most industries, injury rates are highest in the first months on a job. Short tenure is one of the strongest predictors of being hurt that we have. So the claim cannot be that experience is dangerous. Something narrower is going on: among experienced workers, the injuries that do occur tend to follow a different path — not ignorance of the hazard, but a knowing deviation from a procedure that has been skipped safely many times before. And there are settings where the veteran rate genuinely does rise, typically ones where the routine is long-established and the hazard is intermittent.
One more caution, because it is the sort of thing that quietly inverts a finding. Workers who remain in a hazardous trade after twenty years are the ones who were not injured out of it. Comparing tenure groups compares survivors with entrants, and that alone would make experience look protective even if it were not.
The analogy
THE ANALOGY #Think of proofreading a page you wrote yourself. The first pass is slow and letter by letter. By the fifth you are no longer reading words at all — you are recognising the sentences you meant to write, and the typo sits in plain sight, unread. Nothing about your ability has declined. Checking has been replaced by recognition, which is faster and almost always right, and which fails precisely on the case that does not match what you expect.
A missed typo costs nothing and can be caught later, whereas the industrial version of that substitution has an irreversible tail. And no one hands a proofreader harder pages because she is good at it — whereas the experienced worker is routinely given the awkward job, the night call-out, the machine nobody else will touch, so part of what looks like complacency is simply worse exposure.
Clarifying the model
THE MODEL #The mechanism is a gap between two things that both improve, at different rates — perceived risk and actual risk — plus the erosion of a check step that never justified itself on any single day. Neither part requires anybody to be careless, which is why "complacency" is a poor word for it if it is heard as a character judgement.
That framing also predicts where you should and should not see it. It should be strongest where the hazard is intermittent rather than continuous, where the shortcut has an immediate payoff in time or comfort, and where the safe procedure is separable from the productive task — an isolation step, a harness, a permit. It should be weak where the safety measure is inseparable from doing the job at all.
And the honest caveat is worth stating plainly rather than in a footnote. The literature does not support "experienced workers are more dangerous". It supports something more specific: that the kind of injury shifts with tenure, from not-knowing to knowingly-deviating, and that in particular settings long routine erodes deliberate caution. Confounds sit all over that finding — task assignment, hours of exposure, survivor effects, and different reporting habits between new and long-serving staff — and it is worth holding the conclusion at the strength the evidence actually gives it.
A picture of it
THE PICTURE #How to readStart at the black dot and follow the ordinary career: skill rises, and the deliberate check is replaced by recognition. The self-loop on Routinised is the engine — each uneventful repetition is real evidence that the shortcut is safe, so the margin narrows without anyone deciding to narrow it. Two arrows reach Injured, and their labels are the point: the novice is hurt by a hazard he has not learned, the veteran by one that did not look like the shape he had learned. The loop through Re-alerted is why a safety talk works for a while and then stops working.
What became clearer
WHAT CLEARED #Experience does not make people reckless; it makes caution unnecessary in almost every instance, and a habit that is unnecessary almost always is a habit that decays. Perceived risk falls faster than real risk, the surplus gets spent on speed and convenience, and the deliberate check drops out of a routine that no longer seems to need it. The blunt version of the claim is wrong — beginners are hurt more often — but the specific version survives: what changes with tenure is not how much people know, it is which injuries remain possible.
Where to go next
ONWARD #- Why near misses are such weak teachers, and what reporting systems do to make them count.
- How task assignment inside a crew reallocates hazard toward the most experienced worker, independently of anything they choose.
Key terms
TERMS #| Term | What it means |
|---|---|
| Risk compensation | the tendency to spend an improvement in safety on greater performance, so the realised risk falls by less than the improvement. |
| Habituation | the decline in response to a stimulus that has repeatedly proved harmless. |
| Automaticity | the conversion of a deliberate step-by-step sequence into a fast one requiring little attention, which is what practice produces. |
| Healthy worker survivor effect | the bias created when the people who remain in a hazardous job are those it has not yet harmed. |
Every term the collection defines is gathered in the glossary.