Water management inherits its professional confidence from engineering disciplines where that confidence is earned by the completeness of what is known. In hydrology, the completeness is an illusion. For the models that carry this illusion into practice, see No Fat Modelling and When the River Disagrees. For the professional culture that responds to uncertainty with more data rather than more humility, see The Hungry Model.

Hydrology is routinely practised as if it were a closed system of known components governed by exact laws — when in fact the components are partially known, the laws are mostly empirical, and uncertainty is not a problem to be solved but a condition to be worked with.

In electronics, the engineer works with a closed system of their own making. Every component was chosen, specified, and placed deliberately. The laws governing their interaction are derived from first principles and exact in their application. Uncertainty is in principle solvable by better instruments and tighter specifications. The engineer designed the system, knows the system, and can predict its behaviour with confidence.

This mental model — closed system, known components, exact laws, solvable uncertainty — is the mental model that water management has imported into hydrology. It is the wrong model for the wrong system.

The hydrologist is not working with a system they designed. The soil has a hydraulic conductivity — but it varies across the catchment, changes with moisture content, and cannot be measured everywhere. The rainfall has an intensity — but it varies across the storm and across the return period in ways described by probability distributions rather than specifications. The laws governing these interactions are not derived from first principles. They are empirical observations formalised into equations and applied beyond the conditions in which they were observed.

We have, to use an honest phrase, only the faintest idea about most of the components.

The electronics mental model responds to uncertainty with more: more sensors, more data, more model complexity, more computational resolution. But in hydrology, uncertainty is not primarily a measurement problem. It is a condition of the system. What gets lost in the relentless pursuit of more is the most important skill in hydrology: the ability to read a system whose complexity is irreducible but whose behaviour is deeply recognisable. Rivers flood in predictable seasons. Catchments generate runoff in patterns that persist across decades. The system cannot be fully known — but it can be read, by someone who has watched it carefully enough and long enough. This knowledge cannot be entered into a model. It arrives without peer review. It is listened to politely and set aside.

The result is a profession simultaneously overconfident in its models and underconfident in its judgment — one that mistakes the precision of its outputs for the accuracy of its understanding.

Acknowledge openly that the components are partially known, the laws are empirical, and the uncertainty is irreducible — and design with margins that reflect genuine ignorance rather than measurement tolerance. Protect the practice of reading the system alongside the practice of modelling it. Treat divergence between model output and observed behaviour as information about the limits of the model, not as error to be explained away. And before reaching for more data, ask whether the uncertainty is a measurement problem or a condition of the system itself.

For the model that mistakes precision for accuracy, see No Fat Modelling. For the professional culture that ranks model over reality, see When the River Disagrees. For the budget consumed by data collection before the questions have been asked, see The Hungry Model. For the single future the model produces from a system that contains many, see The Single Future Pitfall. For the experienced knowledge the electronics mental model cannot process, see On His Land and The Loop of Reinvented Knowledge.