Honest Extrapolation
Extrapolating beyond measured or validated conditions is sometimes necessary, but Kronos labels it as extrapolation rather than presenting it as established.
When you must go beyond the data
Design requires reaching beyond conditions anyone has measured — that is what makes it design. Extrapolation is legitimate. What is not legitimate is presenting an extrapolated result with the same confidence as a validated one. Kronos labels extrapolation as extrapolation and states how far past the validated range it reaches.
Extrapolations we flag
- Confinement scalings taken to the breeder's regime beyond existing devices
- A two-point-to-three-point extrapolation flagged explicitly in our burner analysis
- Fleet-level projections built from single-machine design points
- Long-term lifetime inferred from short-term irradiation models
The honest way to extrapolate
State the anchor points, the direction of the extrapolation, and the assumption that carries it. Show how the result moves if that assumption is wrong. Then let the reader weigh it. The failure mode is a smooth curve that hides where the data stopped and the assumption began — so we mark that boundary explicitly.
An extrapolation honestly labelled is a hypothesis worth testing. The same extrapolation presented as fact is a claim waiting to embarrass its author.