x402 Treasury Experiments

Supplementary dataset for a pre-registered measurement of the cost and friction of converting accumulated L2 USDC micropayment revenue into spendable fiat on Base mainnet. The sweep experiment (usdc-fiat-sweep-api) was committed to a public repository before any conversion: protocol, operator runbook, tranche ladder, and figure recipes sealed in git; one execution pass; results committed regardless of outcome. Eight pre-registered tranches (1, 1, 5, 5, 20, 20, 100, and the full remaining balance) totaling USD 157.25908 were converted to USD via the Coinbase Advanced Trade convert API, swept from the production receiver wallet (9 run records: 8 executed, 1 pre-registered not_executed). Endpoints per tranche: realized fee, quoted fee, spread against the USD mid at execution time, and per-hop latency (send, Coinbase credit, fiat spendable). Headline results: convert fee quoted and charged USD 0.00, realized fee 0.00%, and spread 0% on every tranche; each tranche converted 1:1 to fiat; the only realized cost was Base gas, 0.00000654 ETH spent from a 0.00014 ETH funding float (on-chain transfer values verified against Base mainnet at wei precision). Send-to-spendable latency for the three tranches with uncontaminated credit timestamps: 236 s, 263 s, and 5 s; the five earlier tranches' credit timestamps are recovery-time artifacts of a disclosed mid-run remediation and are excluded from latency findings; every execution deviation is recorded verbatim in the repository METHODOLOGY. Includes the sealed protocol and operator runbook, the run records with verbatim raw API artifacts and transaction receipts, the frozen figure recipes, and the figures. The operator maintains the production site whose revenue is measured; findings are about this measurement, not any product.