Answers are grounded in this paper and cite their sources.
Customs administrations are asked to trust declarations they cannot verify. Blockchain has been offered as a remedy for a decade, yet a 2022 survey of 110 administrations found only two full deployments. We argue that one contributing cause is a mismatch between the problem and the proposed remedy. A ledger makes a record tamper-evident without establishing that the record was true when it was written. We present CANDID, a Customs Attestation Network for Declaration Integrity and Disclosure. It commits salted field-group hashes rather than payloads and records which off-chain source attested to which declared field. Its reference implementation sustained 3500 transactions per second with durable storage on two cores. This rate, measured with real cryptography within a one-second latency objective, is well above the estimated peak demand of a very large administration. Selectivity was evaluated on 54,000 synthetic Korean import declarations generated from real data, with twenty draws of a synthetic attestation layer. Under the baseline generator, the attestation record raised the area under the curve from 0.649 to 0.909, and a sensitivity analysis showed this gain shrinking as sources became less informative. We then tested whether weighting provenance by source credibility helps. Seven policies were implemented, including Beta reputation, truth discovery, and the likelihood-ratio rule that is optimal under conditional independence. No deployable policy improved on passing the raw per-source indicators to the classifier in any setting examined. In the evaluated setting, the attestation record itself carries the value, and deployable weighting adds nothing that we could measure.
No takes yet. Share an insight, caveat, or question.
Nedal Ababneh (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: