No thought leadership. These are the working documents behind the method — published so that anyone can test them rather than take our word.
A threshold derived from cost asymmetry, a two-threshold triage, hard gates that override weighted scores, and four formal ways to say “I won't decide”. Versioned; changes ship as a new version.
Read the method →Marginal grid intensity moves by roughly 90% across a day. For an aeroponic footprint the schedule of the fog pump is a larger lever than most hardware efficiency upgrades.
Read the note →When one entity issues the label, sells the premium and takes the insurance split, its numbers stop working as evidence — not because it lied, but because the structure makes the claim untestable.
Read the policy →Six layers: legal form, governance, competence, metrology, accreditation, liability. Missing any one makes the signature decorative. We publish our own progress against each.
Read the conditions →Software margin approaches zero with scale; evidence does not — measurement, review and liability all cost money per unit. A per-batch fee is the only structure that pays for the audit.
See the fee schedule →One evaluates a claim, the other evaluates you. Confusing them is the fastest way to be excluded from a procurement process you were otherwise qualified for.
Read the pathway →Insights are maintained in Chinese in the establishment file; the English site links to them directly rather than running two divergent versions.