notes: events/talks/kericonf26-d2-11-hardman-general-evidence-problems

This data as json

path title type status classification words body
events/talks/kericonf26-d2-11-hardman-general-evidence-problems Daniel Hardman: Applying KERI to General Evidence Problems (KERIConf 2026) summary v0.1 — 2026-10-15: drafted by Claude from the curated transcript; unread by Mark; marked partner on Mark's word 2026-10-21, paraphrase only, no quotation partner 769 # Daniel Hardman: Applying KERI to General Evidence Problems (KERIConf 2026) **Abstract.** Daniel Hardman presented the dossier, a form of ACDC (Authentic Chained Data Container) for assembling rich, durable collections of evidence, drawing on his work in telecommunications but arguing the idea applies far more widely. He covered why simple credential presentations fall short, how a dossier is structured and referenced, how it differs from a verifiable presentation, and features in the draft dossier specification such as joint issuance. The talk is for specification writers and implementers interested in ACDC beyond single-credential use. ## Summary Hardman opened by saying that dossiers are a topic on which the KERI (Key Event Receipt Infrastructure) community could lead. He recalled Samuel Smith describing ACDCs around 2019 as a verifiable data graph, a phrase he has come to see as profound. He described the problem as he met it in telecommunications. Verifiable credentials assume a holder in direct, two-way interaction with a verifier. His use cases needed one-way verification by several parties along a path, with no return channel, and verification out of band later by an auditor or regulator. He also needed a richer evidence model. Most credential demonstrations involve one credential, yet real processes such as applications, research, investigations and documentation call for large collections of evidence that accrete over time. He stressed two overlooked points. Evidence has lifespans, and a collection is only as durable as its shortest-lived piece. Pieces of evidence also serve different purposes under different rules. A mortgage file, his example, combines consecutive recent pay stubs, recent proof of employment and a credit report, each with its own issuer and governance, all evaluated at funding and kept afterward so one can later ask whether the borrower qualified at the time. He called such a curated collection a dossier. Because it is a verifiable data graph, it has a SAID (self-addressing identifier), so a small reference, optionally as an OOBI (Out-Of-Band Introduction), can travel where large data cannot. In voice calls, the SIP invite carries a signed token in an identity header, and that token can point to the evidence, so every call carries a new signature over the same stable evidence. Structurally, a dossier is an ACDC whose attributes carry curation metadata (who assembled it, when, under what governance, for what purpose) while every piece of evidence sits in edges pointing to other ACDCs. Non-credential data such as PDFs, audio, images or hashes of large files can be embedded, and foreign credentials such as X.509 certificates, W3C verifiable credentials or mobile driver's licenses can be wrapped: verified in their own stack and attested in a new ACDC that states its stance on revocation. He contrasted dossiers with verifiable presentations. A presentation is ephemeral and proves entitlement; a dossier is kept, and signing it asserts collection and purpose rather than the truth of every item. In questions he said edge operators come from ACDC itself, that a dossier can point to another dossier, and welcomed an audience suggestion to put OOBIs in edges. Samuel Smith explained that verifiable data graphs come from automated reasoning, where aggregation operators combine inputs to make decisions. An audience member linked dossiers to purpose limitation under fair information practices. He introduced joint issuance, in which independent parties anchor the same dossier asynchronously until a threshold is met, with an optional finalizer, and showed how one dossier can be disclosed sparsely to one audience and fully to another under the same root hash. The specification is at version 0.6 and he invited contributions. ## Key terms dossier, verifiable data graph, evidence lifespan, curation, edges, embedding, wrapping, joint issuance, finalizer, graduated disclosure ## Claims and their evidence - A dossier is a flavor of ACDC whose attributes carry curation metadata and whose edges carry the evidence. [informative] - Because a dossier has a SAID, a compact reference such as an OOBI can travel in small headers like a SIP identity token while the evidence stays stable. [informative] - The lifespan of a body of evidence is bounded by the shortest lifespan among its parts. [presenter-reported] - The draft dossier specification introduces joint issuance with asynchronous signers, a threshold and an optional finalizer, and new edge operators. [presenter-reported] - The dossier specification is at version 0.6 and open for contributions. [presenter-reported] ## Related The dossier specification (version 0.6), the ACDC specification, verifiable credentials and verifiable presentations (W3C), X.509, ISO mobile driver's license, SIP and its identity header, Trust over IP working groups. ## Provenance Drafted by Claude on 2026-10-15 from the curated transcript (d2-11-daniel-hardman-applying-keri-to-general-evidence-problems.md), all minutes read. No quotations; paraphrase only. Not yet read by Mark.

Links from other tables