This is my own record. I have lived inside the pattern Care is describing, and the pattern has been the shape of my adult life.
What follows are seven moments from thirty-two years inside a wet network. They are not the worst moments, and they are not the only moments. They are the ones where the pattern becomes visible — where a finding is labelled in a way that determines what happens next, where a rule is applied without regard to what the rule was for, where the network cannot see what it is doing because the artefact it is looking at does not contain the thing that matters.
They are in order. The reader is not asked to judge the individual decisions. The reader is asked to notice the recurrence.
1994 — the finding that was not investigated
Before the endocarditis, before the aortic valve replacement, before any of it, a GP had documented an unusual fourth heart sound in my notes. The finding was recorded and not investigated. It was not passed on when I was referred for a routine procedure. It was not factored into the decision about whether to give antibiotics before that procedure.
The procedure was a vasectomy, performed without prophylactic antibiotics. At the time, antibiotic prophylaxis was recommended for certain patients considered at risk of infective endocarditis undergoing procedures likely to cause bacteraemia. A patient with an unusual cardiac finding and a previously undocumented bicuspid aortic valve was such a patient. No one connected the finding to the procedure, because the finding was in one part of the record and the procedure was in another.
Within weeks I was in hospital with acute infective endocarditis, a femoral embolus, and an aortic valve that would need to be replaced.
The finding was not wrong. The procedure was not wrong in isolation. The connection between them was the thing that no one made, because the network had no mechanism for making it.
2002 — the registrar who called it a misinterpretation
Eight years later, a follow-up echocardiogram showed a paravalvular leak around the mechanical aortic valve. The finding was recorded, and then dismissed by a cardiology registrar as a misinterpretation of the imaging.
The registrar was not lying and was not careless. He was applying a reasonable rule: the previous echo had been normal, the new echo did not match, therefore the new echo was probably wrong. It is the rule that protects against overreaction. It is also the rule that produces the freeze, in the form of a finding that is dismissed rather than investigated.
A second opinion was sought from a more senior cardiologist. He confirmed the leak and the associated haemolysis. The finding was real. The registrar's rule had been correct in general and wrong in this case, and the network had no way of knowing which until the second opinion arrived.
The leak, having been confirmed, was then managed conservatively for several years, because surgery was considered too risky and percutaneous closure was not yet available at my centre.
2009 — the brain that rendered the missing half
In September 2008, after the third percutaneous closure of the paravalvular leak, I lost the upper half of the visual field in my right eye. A retinal artery embolism, probably cardioembolic. I wrote about it in the notebook (Artificial Reality) much later, trying to make sense of what my brain did in those few seconds. But what stayed with me was not the loss. It was the fact that the world did not go black. The top half of the room remained intact. My brain, cut off from the data, continued to render what ought to be there. It filled the gap with expectation. It kept the lights on in a room that had already gone dark.
The network does exactly the same thing. When a patient does not fit the protocol, the network continues to render the patient it expects to see. The scan says mild, because the system expects mild. The blood pressure is noted on the left arm because the right arm does not fit the algorithm. The network fills the gap with the artefact, and the artefact is coherent, and the patient is not.
This is not a failure of individual perception. It is the structural condition of a system that prefers a coherent model to a fragmented reality. The doctor in 2011 saw a mild leak because the system rendered a mild leak. The consultant in 2013 saw conservative management because the system rendered a stable trajectory. The system was doing what a brain does. It was keeping the lights on.
2011 — the leak described as mild
An echocardiogram in October 2011 described the paravalvular leak as mild. The word went into the record. Two weeks later I was in severe heart failure, unable to lie flat, and by December a cardiac catheterisation showed severe paravalvular regurgitation, raised left-ventricular end-diastolic pressure, and pulmonary hypertension.
The word mild was not a lie. It was a description of what the scan showed at the moment it was taken, by the criteria the scan was using. It was also the artefact the network acted on, and the network acted on it as if it were the state of the patient. The patient was deteriorating. The word said mild. The word was believed, because the word was in the record, and the record is what the network can see.
What followed was a cascade of investigations, admissions, and eventually a percutaneous closure that stabilised me. The cascade was necessary. It would also not have been necessary if the earlier finding had been acted on as a trajectory rather than as a snapshot. The network can act on a snapshot. It cannot act on a trajectory, because a trajectory is not an artefact.
2013 — conservative management because no target was available
By 2013 the paravalvular leak had recurred and was severe, and an attempt at further percutaneous closure was unsuccessful. The clinicians recorded that breathlessness was not the principal limitation, that the left-upper-chest pain was not considered typically ischaemic, and that conservative observation was more prudent than immediate high-risk surgery.
Each sentence was true. The breathlessness was not the principal limitation at that moment. The pain was not typical. The surgery was high-risk, and the risk was real. The network was doing what the network does: it was applying the rule that says do not act when the expected harm of acting exceeds the expected harm of waiting.
What the rule could not see was that the trajectory was not stable. The patient was getting worse. The rule is a snapshot rule. It protects against unnecessary intervention. It does not protect against the slow accumulation of damage that happens when intervention is deferred and the deferral is recorded as conservative management.
Eight months later, after a multidisciplinary meeting with my family present, the decision was made to proceed with a third sternotomy, a bioprosthetic aortic valve replacement, and coronary artery bypass grafts. The surgery was described as uneventful but extremely difficult. I recovered. The piece of me that recovered was not quite the same as the piece of me that had gone in.
I wrote about that period in my notebook (The Measure of a Thing) at the time, trying to understand the shift I was watching. I noted that the surgeon himself never changed. He was direct, focused, unhurried — someone who saw a person with a problem to solve. But the hospital around him had changed entirely. By 2014, it was a logistics operation, and I was the item being processed. Discharge was discussed before recovery. Movement mattered more than wellbeing. “Compliance eclipsed care,” I wrote at the time. I didn’t have the language of wet networks then, but I could see the artefact replacing the care.
2026 — the blood pressure that does not belong to the algorithm
This year, my GP and cardiologist documented a consistent and marked difference between my left-arm and right-arm blood pressure. The readings over several days showed the left arm consistently higher and the right arm consistently lower. The difference is large enough to suggest arterial narrowing on the right side.
I queried this. The response was that future measurements should be taken on the left arm only, because the left arm is the more reliable reading for medication decisions. The right-arm finding was noted and not investigated, on the basis that the CT head scan arranged by nephrology had already been done and would not explain the inter-arm difference.
This is the same pattern at a different scale. A finding that does not fit the protocol is noted, and then the protocol is adjusted so the finding does not need to be acted on. The adjustment is reasonable. The finding is real. And no one has yet suggested investigating the right-arm pressure, because the protocol the network runs on does not have a slot for it.
The inter-arm difference is a small thing compared with what came before. It is in the piece because it is now, and because the pattern is still operating. It is not a historical artefact of a less careful era. It is what the network does when the finding does not fit the shape of the thing the network can act on.
2026 — the wisdom that gets filtered out on the way to the top
The last moment is not an incident. It is a structure.
Throughout those thirty-two years, the people who have helped me most were not the people with the most authority. They were the people with the most experience of the specific problem in front of them — the clinician who had seen this pattern before, the nurse who knew what I looked like when I was deteriorating, the GP who noticed something the specialist had missed. Every one of them was doing work the network did not reward. Every one of them was applying a pattern-recognition that the network could not audit.
And every one of them was operating within a system that was increasingly shaped around the artefact rather than the pattern. The targets, the dashboards, the audits, the thresholds — they were all doing their job, and their job was to make the service legible. The service became legible. The patterns became less so, because patterns are not legible in the same way. A pattern is a shape. An artefact is a number. The network can rank numbers. It cannot rank shapes.
Which is why the elder wisdom matters, and why it is being filtered out. The people with the most slow depth are the ones whose judgement is least auditable. The network does not need to exclude them. It only needs to reward the auditable, and wait.
What the seven moments have in common
None of the seven moments is a scandal. No one in this piece acted in bad faith. Every decision the network made was defensible on the information available at the time it was made. The consultants were competent. The registrars were competent. The system was doing what the system does.
The pattern is not individual. The pattern is structural. It is what happens when a wet network runs on the artefacts it can see, and the artefact is never the thing that matters. A heart sound in one part of the record and a procedure in another. A leak labelled mild. A finding dismissed as a misinterpretation. A missing visual field rendered whole by a brain that could not accept the gap. A trajectory treated as a snapshot. A measurement excluded because the protocol has no slot for it. A pattern that cannot be audited, and so cannot be rewarded.
Every one of these is the same mechanism, wearing a different face. The network can only act on what it can see, and what it can see is never the oath that matters.
I am not writing this to complain. The network that failed me also saved me, repeatedly, and the people in it were doing their best. I am writing it because the pattern is not specific to my case, and because the network has no mechanism for learning from it. Thirty-two years of the same structural failure, in a system that cannot see the failure as structural, because the failure is not an artefact.
Which is where the next piece begins.
What comes next
The next piece is the anatomy of the mechanism underneath all seven moments. Not one target, not one rule, not one registrar. The way the network's own oaths collide, and the way the selection mechanisms push the wrong one to the top.
Outlying is the ethical centre of Care — three oaths, and the person the network fails to reconcile them for.