Error Bars, Not Point Claims

Kronos prefers a value with an honest error bar to a clean-looking single number, because the clean number implies a precision the evidence does not support.

The false comfort of a single number

A single number looks authoritative. That is exactly why it can mislead. Reporting '88.7 MW' without a range invites the reader to treat it as pinned down, when in fact it is a design point with assumptions attached. An error bar restores the honesty by showing how much the number could move.

When we quote a precise figure

Some frozen-canon values are quoted to several digits — Q_sci 3.424, neutron fraction 5.44%. Those digits describe the model output, not measurement precision. The precision is real in the sense that the calculation reproducibly returns that value; it is not a claim that a built machine will land on exactly that figure. We say which kind of precision we mean.

Reading our numbers correctly

• A frozen design value = the reproducible output of a specific model
• Its uncertainty = how far hardware could plausibly land from it
• A precise-looking figure is not a promise of precise hardware
• When measurements exist, the figure gains a genuine measurement error bar

The goal is calibration: a reader should come away neither over-trusting a tidy number nor dismissing a well-founded one. The error bar is how we communicate the difference.