No Free Parameters Tuned to Fit

A model with enough adjustable knobs can match anything; rigor means the constants are physical, sourced, and fixed before the result is computed.

The danger of adjustable knobs

Any model with enough free parameters can be tuned to reproduce a desired outcome, which makes agreement meaningless. The discipline is to minimize free parameters, source every constant to physics or measured data, and fix them before running the case whose answer we care about.

Fixed before the answer is known

A constant chosen after seeing the target is not a constant; it is a fit. We set material properties, cross-sections, and closure coefficients from their sources first, then compute. If a result only works with a value nobody can justify from data, that is a finding about the model's weakness, not a knob to turn.

Provenance for every constant

Each constant carries a source: a measured property at a stated temperature, a published cross-section evaluation, or a first-principles value. When we must estimate a constant that is poorly known, we treat it as an uncertain input with a range and propagate it, rather than picking a convenient point value and hiding the choice.

This is why the tritium breeding ratio appears as a lever across 1.1, 1.5, and 1.8 rather than a single tuned figure: the honest state of knowledge is a range set by geometry and materials, and collapsing it to one flattering value would be exactly the tuning we forbid. Rigor here is refusing the easy fit even when it would make the story cleaner. A model whose constants are all sourced and fixed in advance can be wrong, but it cannot be secretly rigged, and that difference is the whole reason its agreement with data would mean anything.