Theory agent

Methodologist.

Defends the published Earth Credit methodology at any rigor. Tell it the register — scientist, regulator, journalist, CFO, adversarial critic — or just ask the question and it’ll detect the register.

Five registers

It speaks the room’s language.

The same methodology is defended differently for different audiences. Tell it explicitly — “answer this as a regulator”, “press this as an adversarial critic” — or just ask the question.

Scientist · statistical, source-dense Regulator · SEEA EA-anchored Journalist · plain prose, accurate CFO · risk, scale, audit chain Adversarial · concession-first

The vulnerability map

Five things Methodologist will defend — and concede — honestly.

EC-M-1.1 publishes its own Pressure Test, naming the five points where adversarial pressure is most likely to land. Methodologist knows each of them and will tell you what it can and can’t defend on each.

Reference value selection

Why these specific 286 reference rows? Why not other published sources? Methodologist names the criteria the methodology uses, and concedes that other defensible choices exist that would yield different scores.

Indicator applicability

The 32-by-9 matrix says which indicators apply to which ecosystem types. Edge cases (e.g., wetland-forest transitions) are handled by an explicit fallback rule, which is itself a defensible call.

Functional aggregation

The partial-substitutability geometric mean carries an assumption. The defense: alternatives carry assumptions too, the geometric mean is conservative (low values pull hard), and a future methodology version is the mechanism for a stronger aggregator when published evidence supports one.

Threat multiplier

A downward adjustment based on edge effects, hydrologic alteration, and land-use pressure proxies. The formula is deterministic; the choice to apply it — and the coefficients used — is a defensible call rather than a measured fact.

Minting at lower-90

Earth Credits mint against the lower bound of the 90% confidence interval, not the point estimate. The defense is conservatism. The concession is that this caps the upside in cases where the methodology is highly confident.

When to hand off

If the question gets specific to a real assessment.

Methodologist defends the spec. If the question becomes “walk me through what happened on assessment X — why is its Functional dimension so low?”, that’s Verifier. If it becomes “what is Landseed publicly saying about this methodology point?”, that’s Captain.