notes: research/papers/papers-acdc-ipex-negotiation

This data as json

path title type status classification words body
research/papers/papers-acdc-ipex-negotiation ACDC IPEX Negotiation summary v0.1 — 2026-10-15: drafted by Claude from the PDF in the Papers repository; unread by Mark; marked partner on Mark's word 2026-10-21, paraphrase only partner 600 # ACDC IPEX Negotiation **Abstract.** Working design notes on how parties negotiate graduated disclosure of Authentic Chained Data Containers (ACDCs) under the Issuance and Presentation Exchange (IPEX) protocol, using a composed schema identifier as the credential type and lists of attribute paths to request disclosure. It also clarifies where the issuee identifier sits in aggregate (selectively disclosable) ACDCs and raises open questions on presentation registries in the context of SEDI (State Endorsed Digital Identity). For implementers of ACDC and IPEX. ## Summary The notes compare mechanisms for negotiating graduated disclosure: attribute path lists, aggregate block label lists, metadata ACDCs, schemas and schema decompositions. Negotiating with schema decompositions is judged complicated because attributes appear disjoint in a schema, while in a metadata ACDC attribute paths are not disjoint and can be named with a path string. The proposal is that negotiation use the composed schema as the type: only the schema's Self-Addressing Identifier (SAID) is given, the full schema is assumed cached or attached, attribute paths request the level of partial disclosure, and metadata ACDCs negotiate terms before disclosure. Requesting a simple field discloses all simple fields in its block, with nested blocks disclosed only as their SAIDs; requesting a block discloses the whole subtree. A worked example on a student record shows paths such as `/a/grades/gpa`. The IPEX flow is described as apply, offer (which may imply the same path list and carries a metadata ACDC), agree, and grant, which delivers the uncompacted ACDC at the agreed disclosure level. For chains of ACDCs, a dictionary keyed by schema SAID maps each to its path list; paths may also reach edge and rule sections. Path delimiters could follow CESR (Composable Event Streaming Representation) pathing. A clarifications section states that the issuee field `i` is optional, that an ACDC is either attributive (`a` section) or aggregative (`A` section), that in the aggregative case the issuee is the `i` field of the second block, and that the I2I edge operator fails when no issuee is present. It notes the author's suggestion that the core SEDI identity credential use the aggregate section, while recording that the state of Utah had not decided. A section on presentation registries asks whether they should be optional per citizen and suggests looking for a high-stakes use case, such as a drug prescription. The PDF ends with diagram titles whose images did not come through in the text. ## Key terms graduated disclosure, IPEX, composed schema, attribute path list, metadata ACDC, aggregate section, issuee AID, I2I operator, presentation registry, untargeted ACDC ## Claims and their evidence - Negotiating disclosure with schema decompositions makes attribute pathing complex, while metadata ACDCs allow a single path string at any nesting level. [informative] - An ACDC either has a cryptographically verifiable binding to its issuee through an `i` field or has none; the binding is boolean. [informative] - The keripy class SerderACDC exposes the issuee through its `.iseaid` and `.issuee` properties, as shown in a named test. [informative] - The state of Utah had not decided between attributive and aggregative sections for the core SEDI credential at the time of writing. [presenter-reported] ## Related ACDC specification (Trust over IP), IPEX (section of the ACDC specification), KERI Architectures for Group Issuance (KAGI), Revocation, New Default logic for Unary Edge Operators. ## Provenance Drafted by Claude on 2026-10-15 from the PDF `whitepapers/ACDC_IPEX_Negotiation.pdf`, text extracted in full (about 1,450 words; diagrams not readable as text). The PDF names no author; it is attributed to Samuel M. Smith by the repository. A longer markdown version in the same repository was not summarized. Not yet read by Mark.

Links from other tables