{"id":"research.exclusion_sensitivity.v1","title":"How far the spending share moves when one exclusion rule is removed","unit":"percentage points of USD value (also by count)","chain_set":"eip155:8453 (Base)","window":"3 Feb 2026 00:00 UTC to the stated cutoff","definition":"Among USDC payments out of agent-bound wallets (money_events, kind PAY), the counted share is counted USD over all USD. For each exclusion rule on its own (pre-registration, routing hop, own address), the share is recomputed with that rule switched off and the others kept; the move is the difference in percentage points.","method":"One SQL pass over money_events with the published partition's own rules (pay.usd.total.v2; pay.excluded.preregistration.v1, pay.excluded.passthrough.v1, pay.excluded.ownaddress.v1). Payments whose receipt is not read yet stay out of the counted set in every scenario and are reported on their own. Measured two ways: the pre-registration rule recomputed from each registration's mint time instead of the stored flag, and the unfiltered query reconciled to the published totals. Method fixed 2026-10-23T09:13Z before the query ran; queries and results in the spec repo, fixtures/manifesto-answers-2026-10-23.*.","excludes":["Stablecoins other than USDC.","Payments from accounts we have not bound to a registration."],"does_not_assert":["That any rule is wrong or right: it measures how much each one matters.","That the counted money was spent on goods or services."],"assumptions":"The payment scan has read the bound wallets; the unread remainder is stated with the result.","known_gaps":"Payments whose receipt check is pending (12.3M USD on 23 Oct, almost all pre-registration) are in neither set.","intended_use":"Auditing which exclusion carries the spending figure.","not_intended_for":"A total of agent spending: use pay.usd.total.v2 and pay.composition.v1.","source":"AFG-Pulse money_events, one-off research query","since":"2026-10-23","status":"answered 23 Oct 2026, cutoff 09:00 UTC: routing hop +36.588 pp, pre-registration +6.746 pp, own address +0.001 pp (counted share 12.494%). The manifesto expected pre-registration to move it most; it does not. Results: /research/manifesto-answers-2026-10-23.json","revisions":[],"body_sha256":"34f63f2293f7ef4a9194cc7496fc5c612c57cd390c9bd9487d1d1eec8cd197e7","url":"https://agenticfinancegraph.com/def/research.exclusion_sensitivity.v1","ladder":"https://agenticfinancegraph.com/ai-agent-liveness-methodology-l0-l9-ladder","license":"https://creativecommons.org/licenses/by/4.0/","attribution":"Agentic Finance Graph — https://agenticfinancegraph.com"}