nx_commonsbench.nx

nx_commonsbench.nx -- THE RULER for neighbour-network standing: does a reciprocity rule actually reward the person who shares and starve the person who skims? Method (per feedback-match-known-good-then-improve): we do NOT invent a ruler. We take the estate's EXISTING reciprocity rule -- the pairwise tolerance-ratio audit shipped in nx_mycorrhiza (nx_my_audit_kind) and nx_hypha (nx_hy_reciprocity_verdict) -- and measure it against a labelled population of economic archetypes with declared ground truth. The bench is CALIBRATED TO FAIL THE INCUMBENT: a ruler that the current rule passes would measure nothing. THE STRUCTURAL CLAIM UNDER TEST: AN EXTRACTOR IS BALANCED WITH EVERY COUNTERPARTY AND NET-NEGATIVE ON THE NETWORK. A PAIRWISE AUDIT IS THEREFORE STRUCTURALLY BLIND TO THE CLASS. The skimmer always gives each counterparty *something* fee-justifying, so every pairwise audit returns BALANCED. The extraction is only visible in the network aggregate. Symmetrically, the most generous member (a nurse: gives 7x what she takes) is the one a ratio rule flags as anomalous. If that is true, the incumbent rule does not merely miss the predator -- it INVERTS the ordering. Ground truth is declared per archetype and is the thing later rules are scored against. Integer only, no floats, no content read. license_tier: ORIGINAL expect_exit: 0