Modern epidemiology was born from an act of seeing. When John Snow traced the cholera deaths of 1854 to a single water pump on Broad Street, his instrument was not a microscope but a map — a way of looking that made visible a pattern others had walked past for weeks. The lesson endures, and it is more radical than it first appears: a health system does not respond to the world as it is. It responds to the world as it has arranged to perceive it.
We tend to speak of surveillance as though it were a neutral mirror, faithfully reflecting whatever moves through a population. It is nothing of the kind. Every surveillance system is a series of decisions made in advance — which conditions are reportable, how a case is defined, what counts as the denominator, which deaths are investigated and which are simply recorded. These choices are made quietly, often years before any outbreak, by people who could not have known what they were ruling in or out. By the time a crisis arrives, the system can only see what it was earlier built to notice.
The actionable and the invisible
This matters because what a system counts becomes, in time, what it is able to act upon. A condition with a case definition, a reporting line, and a column in the weekly return enters the field of management: it can be tracked, budgeted, argued over, reduced. A condition without those things does not merely go unmeasured. It drifts out of the institutional imagination altogether. It becomes, in a sense, weather — present, unremarked, endured rather than addressed.
A system acts only on what it agrees to count. Everything else becomes weather: present, unremarked, and survived rather than addressed.
The pandemic years made this visible at scale. Health systems that had never been data-poor discovered they were sight-poor. Dashboards multiplied; understanding did not keep pace. Numbers arrived in great volume and yet failed to answer the questions that decisions actually turned on — who was being missed, where transmission was concentrating, which signal deserved a response and which was noise. The instinct, almost everywhere, was to gather more. The harder discipline — asking what the data were structurally incapable of showing — was rarer.
The distance between the two curves is not error — it is design. It is the sum of what a system chose not to define, not to reach, or not to look for. A surveillance system can be entirely accurate about what it measures and still leave most of the burden in shadow.
From recording to intelligence
The distinction worth holding onto is between surveillance and intelligence. Surveillance records what happened. Intelligence is built backward from a decision someone has to make — and it is willing to look precisely where the institution would rather not. The two are easily confused, because they often use the same data. They are separated by a question: not what does our information show, but what are we, by the design of our systems, structurally unable to see — and what are we therefore unable to act upon?
This is not a technical refinement. It is closer to a moral posture. The conditions a register omits, the populations a sampling frame never reaches, the deaths counted as something more convenient than their cause — these omissions are rarely neutral. They tend to fall on the people already least visible to an institution. Over time, what begins as a gap in measurement hardens into a kind of structural silence, and a system can come to mistake the limits of its own vision for the limits of the problem.
Resilience, then, is partly a function of sight. A system designed to see clearly — including the parts of its own performance it would prefer not to examine — recovers from shock faster than one that is merely well-instrumented. The first question a serious institution should ask of its surveillance is not how much data it holds, but how honest that data is permitted to be.
For the ministries, universities, and investors we work with, the practical implication is steadying rather than alarming. Better intelligence rarely begins with a larger system. It begins with a more honest one: an account of what is currently invisible, a clear line from each measurement to a decision it is meant to inform, and the institutional courage to count the things that are inconvenient to know. Clear sight is not the whole of good judgment. But nothing good is decided without it.
An essay from Cordelia Advisory · Epidemic Intelligence & Surveillance