]> ProSapien LLC

sam@prosapien.com

TODO TODO Working Group Internet-Draft An authentic chained data container (ACDC) is an IETF internet draft focused specification being incubated at the ToIP (Trust over IP) foundation . An ACDC is a variant of the W3C Verifiable Credential (VC) specification . The W3C VC specification depends on the W3C DID (Decentralized IDentifier) specification . A major use case for the ACDC specification is to provide GLEIF vLEIs (verifiable Legal Entity Identifiers) . GLEIF is the Global Legal Entity Identifier Foundation . ACDCs are dependent on a suite of related IETF focused standards associated with the KERI (Key Event Receipt Infrastructure) specification. These include CESR , SAID , PTEL , CESR-Proof , IPEX , did:keri , and OOBI . Some of the major distinguishing features of ACDCs include normative support for chaining, use of composable JSON Schema , multiple serialization formats, namely, JSON , CBOR , MGPK , and CESR , support for Ricardian contracts , support for chain-link confidentiality , a well defined security model derived from KERI , compact formats for resource constrained applications, simple partial disclosure mechanisms and simple selective disclosure mechanisms.

Introduction The primary purpose of the ACDC protocol is to provide granular provenanced proof-of-authorship (authenticity) of their contained data via a tree or chain of linked ACDCs (technically a directed acyclic graph or DAG). Similar to the concept of a chain-of-custody, ACDCs provide a verifiable chain of proof-of-authorship of the contained data. With a little additional syntactic sugar, this primary facility of chained (treed) proof-of-authorship (authenticity) is extensible to a chained (treed) verifiable authentic proof-of-authority (proof-of-authorship-of-authority). A proof-of-authority may be used to provide verifiable authorizations or permissions or rights or credentials. A chained (treed) proof-of-authority enables delegation of authority and delegated authorizations. The dictionary definition of credential is evidence of authority, status, rights, entitlement to privileges, or the like. Appropriately structured ACDCs may be used as credentials when their semantics provide verifiable evidence of authority. Chained ACDCs may provide delegated credentials. Chains of ACDCs that merely provide proof-of-authorship (authenticity) of data may be appended to chains of ACDCs that provide proof-of-authority (delegation) to enable verifiable delegated authorized authorship of data. This is a vital facility for authentic data supply chains. Furthermore, any physical supply chain may be measured, monitored, regulated, audited, and/or archived by a data supply chain acting as a digital twin . Therefore ACDCs provide the critical enabling facility for an authentic data economy and by association an authentic real (twinned) economy. ACDCs act as securely attributed (authentic) fragments of a distributed property graph (PG) . Thus they may be used to construct knowledge graphs expressed as property graphs . ACDCs enable securely-attributed and privacy-protecting knowledge graphs. The ACDC specification (including its partial and selective disclosure mechanisms) leverages two primary cryptographic operations namely digests and digital signatures . These operations when used in an ACDC MUST have a security level, cryptographic strength, or entropy of approximately 128 bits . (See the appendix for a discussion of cryptographic strength and security) An important property of high-strength cryptographic digests is that a verifiable cryptographic commitment (such as a digital signature) to the digest of some data is equivalent to a commitment to the data itself. ACDCs leverage this property to enable compact chains of ACDCs that anchor data via digests. The data contained in an ACDC may therefore be merely its equivalent anchoring digest. The anchored data is thereby equivalently authenticated or authorized by the chain of ACDCs.

ACDC Fields An ACDC may be abstractly modeled as a nested key: value mapping. To avoid confusion with the cryptographic use of the term key we instead use the term field to refer to a mapping pair and the terms field label and field value for each member of a pair. These pairs can be represented by two tuples e.g (label, value). We qualify this terminology when necessary by using the term field map to reference such a mapping. Field maps may be nested where a given field value is itself a reference to another field map. We call this nested set of fields a nested field map or simply a nested map for short. A field may be represented by a framing code or block delimited serialization. In a block delimited serialization, such as JSON, each field map is represented by an object block with block delimiters such as {} . Given this equivalence, we may also use the term block or nested block as synonymous with field map or nested field map. In many programming languages, a field map is implemented as a dictionary or hash table in order to enable performant asynchronous lookup of a field value from its field label. Reproducible serialization of field maps requires a canonical ordering of those fields. One such canonical ordering is called insertion or field creation order. A list of (field, value) pairs provides an ordered representation of any field map. Most programming languages now support ordered dictionaries or hash tables that provide reproducible iteration over a list of ordered field (label, value) pairs where the ordering is the insertion or field creation order. This enables reproducible round trip serialization/deserialization of field maps. ACDCs depend on insertion ordered field maps for canonical serialization/deserialization. ACDCs support multiple serialization types, namely JSON, CBOR, MGPK, and CESR but for the sake of simplicity, we will only use JSON herein for examples . The basic set of normative field labels in ACDC field maps is defined in the following table.

Field Label Table

Label	Title	Description
v	Version String	Regexable format: ACDCvvSSSShhhhhh_ that provides protocol type, version, serialization type, size, and terminator.
d	Digest (SAID)	Self-referential fully qualified cryptographic digest of enclosing map.
i	Identifier (AID)	Semantics are determined by the context of its enclosing map.
u	UUID	Random Universally Unique IDentifier as fully qualified high entropy pseudo-random string, a salted nonce.
ri	Registry Identifier (AID)	Issuance and/or revocation, transfer, or retraction registry for ACDC.
s	Schema	Either the SAID of a JSON Schema block or the block itself.
a	Attribute	Either the SAID of a block of attributes or the block itself.
A	Attribute Aggregate	Either the Aggregate of a selectively disclosable block of attributes or the block itself.
e	Edge	Either the SAID of a block of edges or the block itself.
r	Rule	Either the SAID a block of rules or the block itself.
n	Node	SAID of another ACDC as the terminating point of a directed edge that connects the encapsulating ACDC node to the specified ACDC node as a fragment of a distributed property graph (PG).
l	Legal Language	Text of Ricardian contract clause.

Compact Labels The primary field labels are compact in that they use only one or two characters. ACDCs are meant to support resource-constrained applications such as supply chain or IoT (Internet of Things) applications. Compact labels better support resource-constrained applications in general. With compact labels, the over-the-wire verifiable signed serialization consumes a minimum amount of bandwidth. Nevertheless, without loss of generality, a one-to-one normative semantic overlay using more verbose expressive field labels may be applied to the normative compact labels after verification of the over-the-wire serialization. This approach better supports bandwidth and storage constraints on transmission while not precluding any later semantic post-processing. This is a well-known design pattern for resource-constrained applications.

Version String Field The version string, v, field MUST be the first field in any top-level ACDC field map. It provides a regular expression target for determining the serialization format and size (character count) of a serialized ACDC. A stream-parser may use the version string to extract and deserialize (deterministically) any serialized ACDC in a stream of serialized ACDCs. Each ACDC in a stream may use a different serialization type. The format of the version string is ACDCvvSSSShhhhhh_. The first four characters ACDC indicate the enclosing field map serialization. The next two characters, vv provide the lowercase hexadecimal notation for the major and minor version numbers of the version of the ACDC specification used for the serialization. The first v provides the major version number and the second v provides the minor version number. For example, 01 indicates major version 0 and minor version 1 or in dotted-decimal notation 0.1. Likewise 1c indicates major version 1 and minor version decimal 12 or in dotted-decimal notation 1.12. The next four characters SSSS indicate the serialization type in uppercase. The four supported serialization types are JSON, CBOR, MGPK, and CESR for the JSON, CBOR, MessagePack, and CESR serialization standards respectively . The next six characters provide in lowercase hexadecimal notation the total number of characters in the serialization of the ACDC. The maximum length of a given ACDC is thereby constrained to be 2²⁴ = 16,777,216 characters in length. The final character - is the version string terminator. This enables later versions of ACDC to change the total version string size and thereby enable versioned changes to the composition of the fields in the version string while preserving deterministic regular expression extractability of the version string. Although a given ACDC serialization type may have a field map delimiter or framing code characters that appear before (i.e. prefix) the version string field in a serialization, the set of possible prefixes is sufficiently constrained by the allowed serialization protocols to guarantee that a regular expression can determine unambiguously the start of any ordered field map serialization that includes the version string as the first field value. Given the version string, a parser may then determine the end of the serialization so that it can extract the full ACDC from the stream without first deserializing it. This enables performant stream parsing and off-loading of ACDC streams that include any or all of the supported serialization types.

AID (Autonomic IDentifier) Fields Some fields, such as the i and ri fields, MUST each have an AID (Autonomic IDentifier) as its value. An AID is a fully qualified Self-Certifying IDentifier (SCID) that follows the KERI protocol . A SCID is derived from one or more (public, private) key pairs using asymmetric or public-key cryptography to create verifiable digital signatures . Each AID has a set of one or more controllers who each control a private key. By virtue of their private key(s), the set of controllers may make statements on behalf of the associated AID that is backed by uniquely verifiable commitments via digital signatures on those statements. Any entity may then verify those signatures using the associated set of public keys. No shared or trusted relationship between the controllers and verifiers is required. The verifiable key state for AIDs is established with the KERI protocol . The use of AIDS enables ACDCs to be used in a portable but securely attributable, fully decentralized manner in an ecosystem that spans trust domains.

Namespaced AIDs Because KERI is agnostic about the namespace for any particular AID, different namespace standards may be used to express KERI AIDs within AID fields in an ACDC. The examples below use the W3C DID namespace specification with the did:keri method . But the examples would have the same validity from a KERI perspective if some other supported namespace was used or no namespace was used at all. The latter case consists of a bare KERI AID (identifier prefix).

SAID (Self-Addressing IDentifier) Fields Some fields in ACDCs may have for their value either a field map or a SAID. A SAID follows the SAID protocol . Essentially a SAID is a Self-Addressing IDentifier (self-referential content addressable). A SAID is a special type of cryptographic digest of its encapsulating field map (block). The encapsulating block of a SAID is called a SAD (Self-Addressed Data). Using a SAID as a field value enables a more compact but secure representation of the associated block (SAD) from which the SAID is derived. Any nested field map that includes a SAID field (i.e. is, therefore, a SAD) may be compacted into its SAID. The uncompacted blocks for each associated SAID may be attached or cached to optimize bandwidth and availability without decreasing security. Several top-level ACDC fields may have for their value either a serialized field map or the SAID of that field map. Each SAID provides a stable universal cryptographically verifiable and agile reference to its encapsulating block (serialized field map). Specifically, the value of top-level s, a, e, and r fields may be replaced by the SAID of their associated field map. When replaced by their SAID, these top-level sections are in compact form. Recall that a cryptographic commitment (such as a digital signature or cryptographic digest) on a given digest with sufficient cryptographic strength including collision resistance is equivalent to a commitment to the block from which the given digest was derived. Specifically, a digital signature on a SAID makes a verifiable cryptographic non-repudiable commitment that is equivalent to a commitment on the full serialization of the associated block from which the SAID was derived. This enables reasoning about ACDCs in whole or in part via their SAIDS in a fully interoperable, verifiable, compact, and secure manner. This also supports the well-known bow-tie model of Ricardian Contracts . This includes reasoning about the whole ACDC given by its top-level SAID, d, field as well as reasoning about any nested sections using their SAIDS.

Selectively Disclosable Attribute Aggregate Field The top-level selectively-disclosable attribute aggregate section, A, field value is an aggregate of cryptographic commitments used to make a commitment to a set (bundle) of selectively-disclosable attributes. The value of the attribute aggregate, A, field depends on the type of selective disclosure mechanism employed. For example, the aggregate value could be the cryptographic digest of the concatenation of an ordered set of cryptographic digests, a Merkle tree root digest of an ordered set of cryptographic digests, or a cryptographic accumulator.

UUID (Universally Unique IDentifier) Fields The purpose of the UUID, u, field in any block is to provide sufficient entropy to the SAID, d, field of the associated block to make computationally infeasible any brute force attacks on that block that attempt to discover the block contents from the schema and the SAID. The UUID, u, field may be considered a salty nonce . Without the entropy provided the UUID, u, field, an adversary may be able to reconstruct the block contents merely from the SAID of the block and the schema of the block using a rainbow or dictionary attack on the set of field values allowed by the schema . The effective security level, entropy, or cryptographic strength of the schema-compliant field values may be much less than the cryptographic strength of the SAID digest. Another way of saying this is that the cardinality of the power set of all combinations of allowed field values may be much less than the cryptographic strength of the SAID. Thus an adversary could successfully discover via brute force the exact block by creating digests of all the elements of the power set which may be small enough to be computationally feasible instead of inverting the SAID itself. Sufficient entropy in the u field ensures that the cardinality of the power set allowed by the schema is at least as great as the entropy of the SAID digest algorithm itself. A UUID, u field may optionally appear in any block (field map) at any level of an ACDC. Whenever a block in an ACDC includes a UUID, u, field then it's associated SAID, d, field makes a blinded commitment to the contents of that block. The UUID, u, field is the blinding factor. This makes that block securely partially-disclosable or even selectively-disclosable notwithstanding disclosure of the associated schema of the block. The block contents can only be discovered given disclosure of the included UUID field. Likewise when a UUID, u, field appears at the top level of an ACDC then that top-level SAID, d, field makes a blinded commitment to the contents of the whole ACDC itself. Thus the whole ACDC, not merely some block within the ACDC, may be disclosed in a privacy-preserving (correlation minimizing) manner.

Full, Partial, and Selective Disclosure The difference between partial disclosure and selective disclosure of a given field map is determined by the correlatability of the disclosed field(s) after full disclosure of the detailed field value with respect to its enclosing block (map or array of fields). A partially disclosable field becomes correlatable after full disclosure. Whereas a selectively disclosable field may be excluded from the full disclosure of any other selectively disclosable fields in the selectively disclosable block (array). After such selective disclosure, the selectively disclosed fields are not correlatable to the so-far undisclosed but selectively disclosable fields in that block. When used in the context of selective disclosure, full disclosure means detailed disclosure of the selectively disclosed attributes not detailed disclosure of all selectively disclosable attributes. Whereas when used in the context of partial disclosure, full disclosure means detailed disclosure of the field map that was so far only partially disclosed. Partial disclosure is an essential mechanism needed to support both performant exchange of information and chain-link confidentiality on exchanged information . The exchange of only the SAID of a given field map is a type of partial disclosure. Another type of partial disclosure is the disclosure of validatable metadata about a detailed field map e.g. the schema of a field map. The SAID of a field map provides a compact cryptographically equivalent commitment to the yet to be undisclosed field map details. A later exchange of the uncompacted field map detail provides full disclosure. Any later full disclosure is verifiable to an earlier partial disclosure. Partial disclosure via compact SAIDs enables the scalable repeated verifiable exchange of SAID references to cached full disclosures. Multiple SAID references to cached fully disclosed field maps may be transmitted compactly without redundant retransmission of the full details each time a new reference is transmitted. Likewise, partial disclosure via SAIDs also supports the bow-tie model of Ricardian contracts . Similarly, the schema of a field map is metadata about the structure of the field map this is validatable given the full disclosure of the field map. The details ofcompact and/or confidential exchange mechanisms that leverage partial disclosure are explained later. Selective disclosure, on the other hand, is an essential mechanism needed to unbundle in a correlation minimizing way a single commitment by an Issuer to a bundle of fields (i.e. a nested array or list or tuple of fields) as a whole. This allows separating a "stew" (bundle) of "ingredients" (attributes) into its constituent "ingredients" (attributes) without correlating the constituents via the Issuer's commitment to the "stew" (bundle) as a whole.

Schema Section

Schema is Type Notable is the fact that there are no top-level type fields in an ACDC. This is because the schema, s, field itself is the type field. ACDCs follow the design principle of separation of concerns between a data container's actual payload information and the type information of that container's payload. In this sense, type information is metadata, not data. The schema dialect is JSON Schema 2020-12 . JSON Schema support for composable schema (sub-schema), conditional schema (sub-schema), and regular expressions in schema enable a validator to ask and answer complex questions about the type of even optional payload elements while maintaining isolation between payload information and type (structure) information about the payload . ACDC's use of JSON Schema MUST be in accordance with the ACDC defined profile as defined herein. The exceptions are defined below.

Schema ID Field Label The usual field label for SAID fields in ACDCs is d. In the case of the schema section, however, the field label for the SAID of the schema section is $id. This repurposes the schema id field label, $id as defined by JSON Schema . The top-level id, $id, field value in a JSON Schema provides a unique identifier of the schema instance. In a usual (non-ACDC) schema the value of the id, $id, field is expressed as a URI. This is called the Base URI of the schema. In an ACDC schema, however, the top-level id, $id, field value is repurposed. Its value MUST include the SAID of the schema. This ensures that the ACDC schema is static and verifiable to their SAIDS. A verifiably static schema satisfies one of the essential security properties of ACDCs as discussed below. There are several ACDC supported formats for the value of the top-level id, $id, field but all of the formats MUST include the SAID of the schema (see below). Correspondingly, the value of the top-level schema, s, field MUST be the SAID included in the schema's top-level $id field. The detailed schema is either attached or cached and maybe discovered via its SAIDified, id, $id, field value. When an id, '$id', field appears in a sub-schema it indicates a bundled sub-schema called a schema resource . The value of the id, '$id', field in any ACDC bundled sub-schema resource MUST include the SAID of that sub-schema using one of the formats described below. The sub-schema so bundled MUST be verifiable against its referenced and embedded SAID value. This ensures secure bundling.

Static Schema For security reasons, the full schema of an ACDC must be completely self-contained and statically fixed (immutable) for that ACDC. By this, we mean that no dynamic schema references or dynamic schema generation mechanisms are allowed. Should an adversary successfully attack the source that provides the dynamic schema resource and change the result provided by that reference, then the schema validation on any ACDC that uses that dynamic schema reference may fail. Such an attack effectively revokes all the ACDCs that use that dynamic schema reference. We call this a schema revocation attack. More insidiously, an attacker could shift the semantics of the dynamic schema in such a way that although the ACDC still passes its schema validation, the behavior of the downstream processing of that ACDC is changed by the semantic shift. This we call a semantic malleability attack. It may be considered a new type of transaction malleability attack . To prevent both forms of attack, all schema must be static, i.e. schema MUST be SADs and therefore verifiable against their SAIDs. To elaborate, the serialization of a static schema may be self-contained. A compact commitment to the detailed static schema may be provided by its SAID. In other words, the SAID of a static schema is a verifiable cryptographic identifier for its SAD. Therefore all ACDC compliant schema must be SADs. In other words, they MUST therefore be SAIDified. The associated detailed static schema (uncompacted SAD) is cryptographically bound and verifiable to its SAID. The JSON Schema specification allows complex schema references that may include non-local URI references . These references may use the $id or $ref keywords. A relative URI reference provided by a $ref keyword is resolved against the Base URI provided by the top-level $id field. When this top-level Base URI is non-local then all relative $ref references are therefore also non-local. A non-local URI reference provided by a $ref keyword may be resolved without reference to the Base URI. In general, schema indicated by non-local URI references ($id or $ref) MUST NOT be used because they are not cryptographically end-verifiable. The value of the underlying schema resource so referenced may change (mutate). To restate, a non-local URI schema resource is not end-verifiable to its URI reference because there is no cryptographic binding between URI and resource. This does not preclude the use of remotely cached SAIDified schema resources because those resources are end-verifiable to their embedded SAID references. Said another way, a SAIDified schema resource is itself a SAD (Self-Address Data) referenced by its SAID. A URI that includes a SAID may be used to securely reference a remote or distributed SAIDified schema resource because that resource is fixed (immutable, nonmalleable) and verifiable to both the SAID in the reference and the embedded SAID in the resource so referenced. To elaborate, a non-local URI reference that includes an embedded cryptographic commitment such as a SAID is verifiable to the underlying resource when that resource is a SAD. This applies to JSON Schema as a whole as well as bundled sub-schema resources. There ACDC supported formats for the value of the top-level id, $id, field are as follows:

• Bare SAIDs may be used to refer to a SAIDified schema as long as the JSON schema validator supports bare SAID references. By default, many if not all JSON schema validators support bare strings (non-URIs) for the Base URI provided by the top-level $id field value.
• The sad: URI scheme may be used to directly indicate a URI resource that safely returns a verifiable SAD. For example sad:SAID where SAID is replaced with the actual SAID of a SAD that provides a verifiable non-local reference to JSON Schema as indicated by the mime-type of schema+json.
• The IETF KERI OOBI internet draft specification provides a URL syntax that references a SAD resource by its SAID at the service endpoint indicated by that URL . Such remote OOBI URLs are also safe because the provided SAD resource is verifiable against the SAID in the OOBI URL. Therefore OOBI URLs are also acceptable non-local URI references for JSON Schema.
• The did: URI scheme may be used safely to prefix non-local URI references that act to namespace SAIDs expressed as DID URIs or DID URLs. DID resolvers resolve DID URLs for a given DID method such as did:keri and may return DID docs or DID doc metadata with SAIDified schema or service endpoints that return SAIDified schema. A verifiable non-local reference in the form of DID URL that includes the schema SAID is resolved safely when it dereferences to the SAD of that SAID. For example, the resolution result returns an ACDC JSON Schema whose id, $id, field includes the SAID and returns a resource with JSON Schema mime-type of schema+json.

To clarify, ACDCs MUST NOT use complex JSON Schema references which allow *dynamically generated *schema resources to be obtained from online JSON Schema Libraries . The latter approach may be difficult or impossible to secure because a cryptographic commitment to the base schema that includes complex schema (non-relative URI-based) references only commits to the non-relative URI reference and not to the actual schema resource which may change (is dynamic, mutable, malleable). To restate, this approach is insecure because a cryptographic commitment to a complex (non-relative URI-based) reference is NOT equivalent to a commitment to the detailed associated schema resource so referenced if it may change. ACDCs MUST use static JSON Schema (i.e. SAIDifiable schema). These may include internal relative references to other parts of a fully self-contained static (SAIDified) schema or references to static (SAIDified) external schema parts. As indicated above, these references may be bare SAIDs, DID URIs or URLs (did: scheme), SAD URIs (sad: scheme), or OOBI URLs. Recall that a commitment to a SAID with sufficient collision resistance makes an equivalent secure commitment to its encapsulating block SAD. Thus static schema may be either fully self-contained or distributed in parts but the value of any reference to a part must be verifiably static (immutable, nonmalleable) by virtue of either being relative to the self-contained whole or being referenced by its SAID. The static schema in whole or in parts may be attached to the ACDC itself or provided via a highly available cache or data store. To restate, this approach is securely end-verifiable (zero-trust) because a cryptographic commitment to the SAID of a SAIDified schema is equivalent to a commitment to the detailed associated schema itself (SAD).

Schema Dialect The schema dialect for ACDC 1.0 is JSON Schema 2020-12 and is indicated by the identifier "https://json-schema.org/draft/2020-12/schema" . This is indicated in a JSON Schema via the value of the top-level $schema field. Although the value of $schema is expressed as a URI, de-referencing does not provide dynamically downloadable schema dialect validation code. This would be an attack vector. The validator MUST control the tooling code dialect used for schema validation and hence the tooling dialect version actually used. A mismatch between the supported tooling code dialect version and the $schema string value should cause the validation to fail. The string is simply an identifier that communicates the intended dialect to be processed by the schema validation tool. When provided, the top-level $schema field value for ACDC version 1.0 must be "https://json-schema.org/draft/2020-12/schema".

Schema Availablity The composed detailed (uncompacted) (bundled) static schema for an ACDC may be cached or attached. But cached, and/or attached static schema is not to be confused with dynamic schema. Nonetheless, while securely verifiable, a remotely cached, SAIDified, schema resource may be unavailable. Availability is a separate concern. Unavailable does not mean insecure or unverifiable. ACDCs MUST be verifiable when available. Availability is typically solvable through redundancy. Although a given ACDC application domain or eco-system governance framework may impose schema availability constraints, the ACDC specification itself does not impose any specific availability requirements on Issuers other than schema caches SHOULD be sufficiently available for the intended application of their associated ACDCs. It's up to the Issuer of an ACDC to satisfy any availability constraints on its schema that may be imposed by the application domain or eco-system.

Composable JSON Schema A composable JSON Schema enables the use of any combination of compacted/uncompacted attribute, edge, and rule sections in a provided ACDC. When compact, any one of these sections may be represented merely by its SAID . When used for the top-level attribute, a, edge, e, or rule, r, section field values, the oneOf sub-schema composition operator provides both compact and uncompacted variants. The provided ACDC MUST validate against an allowed combination of the composed variants, either the compact SAID of a block or the full detailed (uncompacted) block for each section. The validator determines what decomposed variants the provided ACDC MUST also validate against. Decomposed variants may be dependent on the type of disclosure, partial, full, or selective. Unlike the other compactifiable sections, it is impossible to define recursively the exact detailed schema as a variant of a oneOf composition operator contained in itself. Nonetheless, the provided schema, whether self-contained, attached, or cached MUST validate as a SAD against its provided SAID. It MUST also validate against one of its specified oneOf variants. The compliance of the provided non-schema attribute, a, edge, e, and rule, r, sections MUST be enforced by validating against the composed schema. In contrast, the compliance of the provided composed schema for an expected ACDC type MUST be enforced by the validator. This is because it is not possible to enforce strict compliance of the schema by validating it against itself. ACDC specific schema compliance requirements are usually specified in the eco-system governance framework for a given ACDC type. Because the SAID of a schema is a unique content-addressable identifier of the schema itself, compliance can be enforced by comparison to the allowed schema SAID in a well-known publication or registry of ACDC types for a given ecosystem governance framework (EGF). The EGF may be solely specified by the Issuer for the ACDCs it generates or be specified by some mutually agreed upon eco-system governance mechanism. Typically the business logic for making a decision about a presentation of an ACDC starts by specifying the SAID of the composed schema for the ACDC type that the business logic is expecting from the presentation. The verified SAID of the actually presented schema is then compared against the expected SAID. If they match then the actually presented ACDC may be validated against any desired decomposition of the expected (composed) schema. To elaborate, a validator can confirm compliance of any non-schema section of the ACDC against its schema both before and after uncompacted disclosure of that section by using a composed base schema with oneOf pre-disclosure and a decomposed schema post-disclosure with the compact oneOf option removed. This capability provides a mechanism for secure schema validation of both compact and uncompacted variants that require the Issuer to only commit to the composed schema and not to all the different schema variants for each combination of a given compact/uncompacted section in an ACDC. One of the most important features of ACDCs is support for Chain-Link Confidentiality . This provides a powerful mechanism for protecting against un-permissioned exploitation of the data disclosed via an ACDC. Essentially an exchange of information compatible with chain-link confidentiality starts with an offer by the discloser to disclose confidential information to a potential disclosee. This offer includes sufficient metadata about the information to be disclosed such that the disclosee can agree to those terms. Specifically, the metadata includes both the schema of the information to be disclosed and the terms of use of that data once disclosed. Once the disclosee has accepted the terms then full disclosure is made. A full disclosure that happens after contractual acceptance of the terms of use we call permissioned disclosure. The pre-acceptance disclosure of metadata is a form of partial disclosure. As is the case for compact (uncompacted) ACDC disclosure, Composable JSON Schema, enables the use of the same base schema for both the validation of the partial disclosure of the offer metadata prior to contract acceptance and validation of full or detailed disclosure after contract acceptance . A cryptographic commitment to the base schema securely specifies the allowable semantics for both partial and full disclosure. Decomposition of the base schema enables a validator to impose more specific semantics at later stages of the exchange process. Specifically, the oneOf sub-schema composition operator validates against either the compact SAID of a block or the full block. Decomposing the schema to remove the optional compact variant enables a validator to ensure complaint full disclosure. To clarify, a validator can confirm schema compliance both before and after detailed disclosure by using a composed base schema pre-disclosure and a decomposed schema post-disclosure with the undisclosed options removed. These features provide a mechanism for secure schema-validated contractually-bound partial (and/or selective) disclosure of confidential data via ACDCs.

ACDC Variants There are several variants of ACDCs determined by the presence/absence of certain fields and/or the value of those fields. At the top level, the presence (absence), of the UUID, u, field produces two variants. These are private (public) respectively. In addition, a present but empty UUID, u, field produces a private metadata variant.

Public ACDC Given that there is no top-level UUID, u, field in an ACDC, then knowledge of both the schema of the ACDC and the top-level SAID, d, field may enable the discovery of the remaining contents of the ACDC via a rainbow table attack . Therefore, although the top-level, d, field is a cryptographic digest, it may not securely blind the contents of the ACDC when knowledge of the schema is available. The field values may be discoverable. Consequently, any cryptographic commitment to the top-level SAID, d, field may provide a fixed point of correlation potentially to the ACDC field values themselves in spite of non-disclosure of those field values. Thus an ACDC without a top-level UUID, u, field must be considered a public (non-confidential) ACDC.

Private ACDC Given a top-level UUID, u, field, whose value has sufficient cryptographic entropy, then the top-level SAID, d, field of an ACDC may provide a secure cryptographic digest that blinds the contents of the ACDC . An adversary when given both the schema of the ACDC and the top-level SAID, d, field, is not able to discover the remaining contents of the ACDC in a computationally feasible manner such as through a rainbow table attack . Therefore the top-level, UUID, u, field may be used to securely blind the contents of the ACDC notwithstanding knowledge of the schema and top-level, SAID, d, field. Moreover, a cryptographic commitment to that that top-level SAID, d, field does not provide a fixed point of correlation to the other ACDC field values themselves unless and until there has been a disclosure of those field values. Thus an ACDC with a sufficiently high entropy top-level UUID, u, field may be considered a private (confidential) ACDC. enables a verifiable commitment to the top-level SAID of a private ACDC to be made prior to the disclosure of the details of the ACDC itself without leaking those contents. This is called partial disclosure. Furthermore, the inclusion of a UUID, u, field in a block also enables selective disclosure mechanisms described later in the section on selective disclosure.

Metadata ACDC An empty, top-level UUID, u, field appearing in an ACDC indicates that the ACDC is a metadata ACDC. The purpose of a metadata ACDC is to provide a mechanism for a Discloser to make cryptographic commitments to the metadata of a yet to be disclosed private ACDC without providing any point of correlation to the actual top-level SAID, d, field of that yet to be disclosed ACDC. The top-level SAID, d, field, of the metadata ACDC, is cryptographically derived from an ACDC with an empty top-level UUID, u, field so its value will necessarily be different from that of an ACDC with a high entropy top-level UUID, u, field value. Nonetheless, the Discloser may make a non-repudiable cryptographic commitment to the metadata SAID in order to initiate a chain-link confidentiality exchange without leaking correlation to the actual ACDC to be disclosed . A Disclosee (verifier) may validate the other metadata information in the metadata ACDC before agreeing to any restrictions imposed by the future disclosure. The metadata includes the Issuer, the schema, the provenancing edges, and the rules (terms-of-use). The top-level attribute section, a, field value of a metadata ACDC may be empty so that its value is not correlatable across disclosures (presentations). Should the potential Disclosee refuse to agree to the rules then the Discloser has not leaked the SAID of the actual ACDC or the SAID of the attribute block that would have been disclosed. Given the metadata ACDC, the potential Disclosee is able to verify the Issuer, the schema, the provenanced edges, and rules prior to agreeing to the rules. Similarly, an Issuer may use a metadata ACDC to get agreement to a contractual waiver expressed in the rule section with a potential Issuee prior to issuance. Should the Issuee refuse to accept the terms of the waiver then the Issuer has not leaked the SAID of the actual ACDC that would have been issued nor the SAID of its attributes block nor the attribute values themselves. When a metadata ACDC is disclosed (presented) only the Discloser's signature(s) is attached not the Issuer's signature(s). This precludes the Issuer's signature(s) from being used as a point of correlation until after the Disclosee has agreed to the terms in the rule section. When chain-link confidentiality is used, the Issuer's signatures are not disclosed to the Disclosee until after the Disclosee has agreed to keep them confidential. The Disclosee is protected from forged Discloser because ultimately verification of the disclosed ACDC will fail if the Discloser does not eventually provide verifiable Issuer's signatures. Nonetheless, should the potential Disclosee not agree to the terms of the disclosure expressed in the rule section then the Issuer's signature(s) is not leaked.

Unpermissioned Exploitation of Data An important design goal of ACDCs is they support the sharing of provably authentic data while also protecting against the un-permissioned exploitation of that data. Often the term privacy protection is used to describe similar properties. But a narrow focus on "privacy protection" may lead to problematic design trade-offs. With ACDCs, the primary design goal is not data privacy protection per se but the more general goal of protection from the un-permissioned exploitation of data. In this light, a given privacy protection mechanism may be employed to help protect against unpermissioned exploitation of data but only when it serves that more general-purpose and not as an end in and of itself. There are three primary mechanisms ACDCs use to protect against unpermissioned exploitation of data. These are:

• Chain-link Confidentiality
• Partial Disclosure
• Selective Disclosure

Principle of Least Disclosure ACDCs are designed to satisfy the principle of least disclosure.

• The system should disclose only the minimum amount of information about a given party needed to facilitate a transaction and no more.

For example, the partial disclosure of portions of an ACDC to enable chain-link confidentiality of the subsequent full disclosure is an application of the principle of least disclosure. Likewise, unbundling only the necessary attributes from a bundled commitment using selective disclosure to enable a correlation minimizing disclosure from that bundle is an application of the principle of least disclosure.

Three Party Exploitation Model Unpermission exploitation is characterized using a three-party model. The three parties are as follows:

• First-Party = Discloser of data.
• Second-Party = Disclosee of data received from First Party (Discloser).
• Third-Party = Observer of data disclosed by First Party (Discloser) to Second Party (Disclosee).

Second-Party (Disclosee) Exploitation

• implicit permissioned correlation.
  • no contractual restrictions on the use of disclosed data.
• explicit permissioned correlation.
  • use as permitted by contract
• explicit unpermissioned correlation with other second parties or third parties.
  • malicious use in violation of contract

Third-Party (Observer) Exploitation

• implicit permissioned correlation.
  • no contractual restrictions on use of observed data.
• explicit unpermissioned correlation via collusion with second parties.
  • malicious use in violation of second party contract

Chain-link Confidentiality Exchange Chain-link confidentiality imposes contractual restrictions and liability on any Disclosee (Second-Party) . The exchange provides a fair contract consummation mechanism. The steps in a chain-link confidentiality exchange are as follows:

• Discloser provides a non-repudiable Offer with verifiable metadata (sufficient partial disclosure) which includes any terms or restrictions on use.
• Disclosee verifies Offer against composed schema and metadata adherence to desired data.
• Disclosee provides non-repudiable Accept of terms that are contingent on compliant disclosure.
• Discloser provides non-repudiable Disclosure with sufficient compliant detail.
• Disclosee verifies Disclosure using decomposed schema and adherence of disclosed data to Offer.

Disclosee may now engage in permissioned use and carries liability as a deterrent against unpermissioned use.

Compact ACDC The top-level section field values of a compact ACDC are the SAIDs of each uncompacted top-level section. The section field labels are s, a, e, and r.

Compact Public ACDC A fully compact public ACDC is shown below.

Compact Private ACDC A fully compact private ACDC is shown below.

Compact Private ACDC Schema The schema for the compact private ACDC example above is provided below.

Attribute Section The attribute section in the examples above has been compacted into its SAID. The schema of the compacted attribute section is as follows, Two variants of an ACDC, namely, namely, private (public) attribute are defined respectively by the presence (absence) of a UUID, u, field in the uncompacted attribute section block. Two other variants of an ACDC, namely, targeted (untargeted) are defined respectively by the presence (absence) of an issuee, i, field in the uncompacted attribute section block.

Public-Attribute ACDC Suppose that the un-compacted value of the attribute section as denoted by the attribute section, a, field is as follows, The SAID, d, field at the top level of the uncompacted attribute block is the same SAID used as the compacted value of the attribute section, a, field. Given the absence of a u field at the top level of the attributes block, then knowledge of both SAID, d, field at the top level of an attributes block and the schema of the attributes block may enable the discovery of the remaining contents of the attributes block via a rainbow table attack . Therefore the SAID, d, field of the attributes block, although, a cryptographic digest, does not securely blind the contents of the attributes block given knowledge of the schema. It only provides compactness, not privacy. Moreover, any cryptographic commitment to that SAID, d, field provides a fixed point of correlation potentially to the attribute block field values themselves in spite of non-disclosure of those field values via a compact ACDC. Thus an ACDC without a UUID, u, field in its attributes block must be considered a public-attribute ACDC even when expressed in compact form.

Public Uncompacted Attribute Section Schema The subschema for the public uncompacted attribute section is shown below,

Composed Schema for both Public Compact and Uncompacted Attribute Section Variants Through the use of the JSON Schema oneOf composition operator the following composed schema will validate against both the compact and un-compacted value of the attribute section field.

Private-Attribute ACDC Consider the following form of an uncompacted private-attribute block, Given the presence of a top-level UUID, u, field of the attribute block whose value has sufficient cryptographic entropy, then the top-level SAID, d, field of the attribute block provides a secure cryptographic digest of the contents of the attribute block . An adversary when given both the schema of the attribute block and its SAID, d, field, is not able to discover the remaining contents of the attribute block in a computationally feasible manner such as a rainbow table attack . Therefore the attribute block's UUID, u, field in a compact ACDC enables its attribute block's SAID, d, field to securely blind the contents of the attribute block notwithstanding knowledge of the attribute block's schema and SAID, d field. Moreover, a cryptographic commitment to that attribute block's, SAID, d, field does not provide a fixed point of correlation to the attribute field values themselves unless and until there has been a disclosure of those field values. To elaborate, when an ACDC includes a sufficiently high entropy UUID, u, field at the top level of its attributes block then the ACDC may be considered a private-attributes ACDC when expressed in compact form, that is, the attribute block is represented by its SAID, d, field and the value of its top-level attribute section, a, field is the value of the nested SAID, d, field from the uncompacted version of the attribute block. A verifiable commitment may be made to the compact form of the ACDC without leaking details of the attributes. Later disclosure of the uncompacted attribute block may be verified against its SAID, d, field that was provided in the compact form as the value of the top-level attribute section, a, field. Because the Issuee AID is nested in the attribute block as that block's top-level, issuee, i, field, a presentation exchange (disclosure) could be initiated on behalf of a different AID that has not yet been correlated to the Issuee AID and then only correlated to the Issuee AID after the Disclosee has agreed to the chain-link confidentiality provisions in the rules section of the private-attributes ACDC .

Composed Schema for Both Compact and Uncompacted Private-Attribute ACDC Through the use of the JSON Schema oneOf composition operator the following composed schema will validate against both the compact and un-compacted value of the private attribute section, a, field. As described above in the Schema section of this specification, the oneOf sub-schema composition operator validates against either the compact SAID of a block or the full block. A validator can use a composed schema that has been committed to by the Issuer to securely confirm schema compliance both before and after detailed disclosure by using the fully composed base schema pre-disclosure and a specific decomposed variant post-disclosure. Decomposing the schema to remove the optional compact variant (i.e. removing the oneOf compact option) enables a validator to ensure complaint full disclosure.

Untargeted ACDC Consider the case where the issuee, i, field is absent at the top level of the attribute block as shown below, This ACDC has an Issuer but no Issuee. Therefore, there is no provably controllable Target AID. This may be thought of as an undirected verifiable attestation or observation of the data in the attributes block by the Issuer. One could say that the attestation is addressed to "whom it may concern". It is therefore an untargeted ACDC, or equivalently an unissueed ACDC. An untargeted ACDC enables verifiable authorship by the Issuer of the data in the attributes block but there is no specified counter-party and no verifiable mechanism for delegation of authority. Consequently, the rule section may only provide contractual obligations of implied counter-parties. This form of an ACDC provides a container for authentic data only (not authentic data as authorization). But authentic data is still a very important use case. To clarify, an untargeted ACDC enables verifiable authorship of data. An observer such as a sensor that controls an AID may make verifiable non-repudiable measurements and publish them as ACDCs. These may be chained together to provide provenance for or a chain-of-custody of any data. These ACDCs could be used to provide a verifiable data supply chain for any compliance-regulated application. This provides a way to protect participants in a supply chain from imposters. Such data supply chains are also useful as a verifiable digital twin of a physical supply chain . A hybrid chain of one or more targeted ACDCs ending in a chain of one or more untargeted ACDCs enables delegated authorized attestations at the tail of that chain. This may be very useful in many regulated supply chain applications such as verifiable authorized authentic datasheets for a given pharmaceutical.

Targeted ACDC When present at the top level of the attribute section, the issuee, i, field value provides the AID of the Issuee of the ACDC. This Issuee AID is a provably controllable identifier that serves as the Target AID. This makes the ACDC a targeted ACDC or equivalently an issueed ACDC. Targeted ACDCs may be used for many different purposes such as an authorization or a delegation directed at the Issuee AID, i.e. the Target. In other words, a targeted ACDC provides a container for authentic data that may also be used as some form of authorization such as a credential that is verifiably bound to the Issuee as targeted by the Issuer. Furthermore, by virtue of the targeted Issuee's provable control over its AID, the targeted ACDC may be verifiably presented (disclosed) by the controller of the Issuee AID. For example, the definition of the term credential is evidence of authority, status, rights, entitlement to privileges, or the like. To elaborate, the presence of an attribute section top-level issuee, i, field enables the ACDC to be used as a verifiable credential given by the Issuer to the Issuee. One reason the issuee, i, field is nested into the attribute section, a, block is to enable the Issuee AID to be private or partially or selectively disclosable. The Issuee may also be called the Holder or Subject of the ACDC. But here we use the more semantically precise albeit less common terms of Issuer and Issuee. The ACDC is issued from or by an Issuer and is issued to or for an Issuee. This precise terminology does not bias or color the role (function) that an Issuee plays in the use of an ACDC. What the presence of Issuee AID does provide is a mechanism for control of the subsequent use of the ACDC once it has been issued. To elaborate, because the issuee, i, field value is an AID, by definition, there is a provable controller of that AID. Therefore that Issuee controller may make non-repudiable commitments via digital signatures on behalf of its AID. Therefore subsequent use of the ACDC by the Issuee may be securely attributed to the Issuee. Importantly the presence of an issuee, i, field enables the associated Issuee to make authoritative verifiable presentations or disclosures of the ACDC. A designated Issueealso better enables the initiation of presentation exchanges of the ACDC between that Issuee as Discloser and a Disclosee (verifier). In addition, because the Issuee is a specified counter-party the Issuer may engage in a contract with the Issuee that the Issuee agrees to by virtue of its non-repudiable signature on an offer of the ACDC prior to its issuance. This agreement may be a pre-condition to the issuance and thereby impose liability waivers or other terms of use on that Issuee. Likewise, the presence of an issuee, i, field, enables the Issuer to use the ACDC as a contractual vehicle for conveying an authorization to the Issuee. This enables verifiable delegation chains of authority because the Issuee in one ACDC may become the Issuer in some other ACDC. Thereby an Issuer may delegate authority to an Issuee who may then become a verifiably authorized Issuer that then delegates that authority (or an attenuation of that authority) to some other verifiably authorized Issuee and so forth.

Edge Section In the compact ACDC examples above, the edge section has been compacted into merely the SAID of that section. Suppose that the un-compacted value of the edge section denoted by the top-level edge, e, field is as follows, The edge section's top-level SAID, d, field is the SAID of the edge block and is the same SAID used as the compacted value of the ACDC's top-level edge, e, field. Each edge in the edge section gets its field with its own local label. In the example above, the edge label is "boss". Note that each edge does NOT include a type field. The type of each edge is provided by the schema vis-a-vis the label of that edge. This is in accordance with the design principle of ACDCs that may be succinctly expressed as "schema is type". This approach varies somewhat from many property graphs which often do not have a schema . Because ACDCs have a schema for other reasons, however, they leverage that schema to provide edge types with a cleaner separation of concerns. Each edge sub-block has one required node, n, field. The value of the node, n, field is the SAID of the ACDC to which the edge connects. A main distinguishing feature of a property graph (PG) is that both nodes but edges may have a set of properties . These might include modifiers that influence how the connected node is to be used such as a weight. Weighted directed edges represent degrees of confidence or likelihood. These types of PGs are commonly used for machine learning or reasoning under uncertainty. The following example adds a weight property to the edge sub-block as indicated by the weight, w, field.

Globally Distributed Secure Graph Fragments Abstractly, an ACDC with one or more edges may be a fragment of a distributed property graph. However, the local label does not enable the direct unique global resolution of a given edge including its properties other than a trivial edge with only one property, its node, n field. To enable an edge with additional properties to be globally uniquely resolvable, that edge's block may have a SAID, d, field. Because a SAID is a cryptographic digest it will universally and uniquely identify an edge with a given set of properties . This allows ACDCs to be used as secure fragments of a globally distributed property graph (PG). This enables a property graph to serve as a global knowledge graph in a secure manner that crosses trust domains . This is shown below.

Compact Edge Given that an individual edge's property block includes a SAID, d, field then a compact representation of the edge's property block is provided by replacing it with its SAID. This may be useful for complex edges with many properties. This is called a compact edge. This is shown as follows,

Private Edge Each edge's properties may be blinded by its SAID, d, field (i.e. be private) if its properties block includes a UUID, u field. As with UUID, u, fields used elsewhere in ACDC, if the UUID, u, field value has sufficient entropy then the values of the properties of its enclosing block are not discoverable in a computationally feasible manner merely given the schema for the edge block and its SAID, d field. This is called a private edge. When a private edge is provided in compact form then the edge detail is hidden and is partially disclosable. An uncompacted private edge is shown below. When an edge points to a private ACDC, a Discloser may choose to use a metadata version of that private ACDC when presenting the node, n, field of that edge prior to acceptance of the terms of disclosure. The Disclosee can verify the metadata of the private node without the Discloser exposing the actual node contents via the actual node SAID or other attributes. Private ACDCs (nodes) and private edges may be used in combination to prevent an un-permissioned correlation of the distributed property graph.

Simple Compact Edge When an edge sub-block has only one field that is its node, n, field then the edge block may use an alternate simplified compact form where the labeled edge field value is the value of its node, n, field. The schema for that particular edge label, in this case, "boss", will indicate that the edge value is a node SAID and not the edge sub-block SAID as would be the case for the normal compact form shown above. This alternate compact form is shown below.

Node Discovery In general, the discovery of the details of an ACDC referenced as a node, n field value, in an edge sub-block begins with the node SAID or the SAID of the associated edge sub-block. Because a SAID is a cryptographic digest with high collision resistance it provides a universally unique identifier to the referenced ACDC as a node. The Discovery of a service endpoint URL that provides database access to a copy of the ACDC may be bootstrapped via an OOBI (Out-Of-Band-Introduction) that links the service endpoint URL to the SAID of the ACDC . Alternatively, the Issuer may provide as an attachment at the time of issuance a copy of the referenced ACDC. In either case, after a successful exchange, the Issuee or recipient of any ACDC will have either a copy or a means of obtaining a copy of any referenced ACDCs as nodes in the edge sections of all ACDCs so chained. That Issuee or recipient will then have everything it needs to make a successful disclosure to some other Disclosee. This is the essence of percolated discovery.

Rule Section In the compact ACDC examples above, the rule section has been compacted into merely the SAID of that section. Suppose that the un-compacted value of the rule section denoted by the top-level rule, r, field is as follows, The purpose of the rule section is to provide a Ricardian Contract . The important features of a Ricardian contract are that it be both human and machine-readable and referenceable by a cryptographic digest. A JSON encoded document or block such as the rule section block is a practical example of both a human and machine-readable document. The rule section's top-level SAID, d, field provides the digest. This provision supports the bow-tie model of Ricardian Contracts . Ricardian legal contracts may be hierarchically structured into sections and subsections with named or numbered clauses in each section. The labels on the clauses may follow such a hierarchical structure using nested maps or blocks. These provisions enable the rule section to satisfy the features of a Ricardian contract. To elaborate, the rule section's top-level SAID, d, field is the SAID of that block and is the same SAID used as the compacted value of the rule section, r, field that appears at the top level of the ACDC. Each clause in the rule section gets its own field. Each clause also has its own local label. The legal, l, field in each block provides the associated legal language. Note there are no type fields in the rule section. The type of a contract and the type of each clause is provided by the schema vis-a-vis the label of that clause. This follows the ACDC design principle that may be succinctly expressed as "schema is type". Each rule section clause may also have its own clause SAID, d, field. Clause SAIDs enable reference to individual clauses, not merely the whole contract as given by the rule section's top-level SAID, d, field. An example rule section with clause SAIDs is provided below.

Compact Clauses The use of clause SAIDS enables a compact form of a set of clauses where each clause value is the SAID of the corresponding clause. For example,

Private Clause The disclosure of some clauses may be pre-conditioned on acceptance of chain-link confidentiality. In this case, some clauses may benefit from partial disclosure. Thus clauses may be blinded by their SAID, d, field when the clause block includes a sufficiently high entropy UUID, u, field. The use of a clause UUID enables the compact form of a clause to NOT be discoverable merely from the schema for the clause and its SAID via rainbow table attack . Therefore such a clause may be partially disclosable. These are called private clauses. A private clause example is shown below.

Simple Compact Clause An alternate simplified compact form uses the value of the legal, l, field as the value of the clause field label. The schema for a specific clause label will indicate that the field value, for a given clause label is the legal language itself and not the clause block's SAID, d, field as is the normal compact form shown above. This alternate simple compact form is shown below. In this form individual clauses are not compactifiable and are fully self-contained.

Clause Discovery In compact form, the discovery of either the rule section as a whole or a given clause begins with the provided SAID. Because the SAID, d, field of any block is a cryptographic digest with high collision resistance it provides a universally unique identifier to the referenced block details (whole rule section or individual clause). The discovery of a service endpoint URL that provides database access to a copy of the rule section or to any of its clauses may be bootstrapped via an OOBI (Out-Of-Band-Introduction) that links the service endpoint URL to the SAID of the respective block. Alternatively, the issuer may provide as an attachment at issuance a copy of the referenced contract associated with the whole rule section or any clause. In either case, after a successful issuance exchange, the Issuee or holder of any ACDC will have either a copy or a means of obtaining a copy of any referenced contracts in whole or in part of all ACDCs so issued. That Issuee or recipient will then have everything it needs to subsequently make a successful presentation or disclosure to a Disclosee. This is the essence of percolated discovery.

Informative Example of an ACDC

Public Compact Variant

Public Uncompacted Variant

Composed Schema that Supports both Public Compact and Uncompacted Variants

Selective Disclosure As explained previously, the primary difference between partial disclosure and selective disclosure is determined by the correlatability with respect to its encompassing block after full disclosure of the detailed field value. A partially disclosable field becomes correlatable to its encompassing block after its full disclosure. Whereas a selectively disclosable field may be excluded from the full disclosure of any other selectively disclosable fields in its encompassing block. After selective disclosure, the selectively disclosed fields are not correlatable to the so-far undisclosed but selectively disclosable fields in the same encompassing block. In this sense, full disclosure means detailed disclosure of the selectively disclosed attributes not detailed disclosure of all selectively disclosable attributes. Recall that partial disclosure is an essential mechanism needed to support chain-link confidentiality . The chain-link confidentiality exchange offer requires partial disclosure, and full disclosure only happens after acceptance of the offer. Selective disclosure, on the other hand, is an essential mechanism needed to unbundle in a correlation minimizing way a single commitment by an Issuer to a bundle of fields (i.e. a nested block or array of fields). This allows separating a "stew" of "ingredients" (attributes) into its constituent "ingredients" (attributes) without correlating the constituents via the stew. ACDCs, as a standard, benefit from a minimally sufficient approach to selective disclosure that is simple enough to be universally implementable and adoptable. This does not preclude support for other more sophisticated but optional approaches. But the minimally sufficient approach should be universal so that at least one selective disclosure mechanism be made available in all ACDC implementations. To clarify, not all instances of an ACDC must employ the minimal selective disclosure mechanisms as described herein but all ACDC implementations must support any instance of an ACDC that employs the minimal selective disclosure mechanisms as described above. The ACDC chaining mechanism reduces the need for selective disclosure in some applications. Many non-ACDC verifiable credentials provide bundled precisely because there is no other way to associate the attributes in the bundle. These bundled credentials could be refactored into a graph of ACDCs. Each of which is separately disclosable and verifiable thereby obviating the need for selective disclosure. Nonetheless, some applications require bundled attributes and therefore may benefit from the independent selective disclosure of bundled attributes. This is provided by selectively disclosable attribute ACDCs. The use of a revocation registry is an example of a type of bundling, not of attributes in a credential, but uses of a credential in different contexts. Unbundling the usage contexts may be beneficial. This is provided by bulk-issued ACDCs. In either case, the basic selective disclosure mechanism is comprised of a single aggregated blinded commitment to a list of blinded commitments to undisclosed values. Membership of any blinded commitment to a value in the list of aggregated blinded commitments may be proven without leaking (disclosing) the unblinded value belonging to any other blinded commitment in the list. This enables provable selective disclosure of the unblinded values. When a non-repudiable digital signature is created on the aggregated blinded commitment then any disclosure of a given value belonging to a given blinded commitment in the list is also non-repudiable. This approach does not require any more complex cryptography than digests and digital signatures. This satisfies the design ethos of minimally sufficient means. The primary drawback of this approach is verbosity. It trades ease and simplicity and adoptability of implementation for size. Its verbosity may be mitigated by replacing the list of blinded commitments with a Merkle tree of those commitments where the Merkle tree root becomes the aggregated blinded commitment. Given sufficient cryptographic entropy of the blinding factors, collision resistance of the digests, and unforgeability of the digital signatures, either inclusion proof format (list or Merkle tree digest) prevents a potential disclosee or adversary from discovering in a computationally feasible way the values of any undisclosed blinded value details from the combination of the schema of those value details and either the aggregated blinded commitment and/or the list of aggregated blinded commitments . A potential disclosee or adversary would also need both the blinding factor and the actual value details. Selective disclosure in combination with partial disclosure for chain-link confidentiality provides comprehensive correlation minimization because a discloser may use a non-disclosing metadata ACDC prior to acceptance by the disclosee of the terms of the chain-link confidentiality expressed in the rule section . Thus only malicious disclosees who violate chain-link confidentiality may correlate between independent disclosures of the value details of distinct members in the list of aggregated blinded commitments. Nonetheless, they are not able to discover any as of yet undisclosed (unblinded) value details.

Selectively Disclosable Attribute ACDC In a selectively disclosable attribute ACDC, the set of attributes is provided as an array of blinded blocks. Each attribute in the set has its own dedicated blinded block. Each block has its own SAID, d, field and UUID, u, field in addition to its attribute field or fields. When an attribute block has more than one attribute field then the set of fields in that block are not independently selectively disclosable but MUST be disclosed together as a set. Notable is that the field labels of the selectively disclosable attributes are also blinded because they only appear within the blinded block. This prevents un-permissioned correlation via contextualized variants of a field label that appear in a selectively disclosable block. For example, localized or internationalized variants where each variant's field label(s) each use a different language or some other context correlatable information in the field labels themselves. A selectively-disclosable attribute section appears at the top level using the field label A. This is distinct from the field label a for a non-selectively-disclosable attribute section. This makes clear (unambiguous) the semantics of the attribute section's associated schema. This also clearly reflects the fact that the value of a compact variant of selectively-disclosable attribute section is an "aggregate" not a SAID. As described previously, the top-level selectively-disclosable attribute aggregate section, A, field value is an aggregate of cryptographic commitments used to make a commitment to a set (bundle) of selectively-disclosable attributes. The derivation of its value depends on the type of selective disclosure mechanism employed. For example, the aggregate value could be the cryptographic digest of the concatenation of an ordered set of cryptographic digests, a Merkle tree root digest of an ordered set of cryptographic digests, or a cryptographic accumulator. The Issuer attribute block is absent from an uncompacted untargeted selectively disclosable ACDC as follows: The Issuer attribute block is present in an uncompacted untargeted selectively disclosable ACDC as follows:

Blinded Attribute Array Given that each attribute block's UUID, u, field has sufficient cryptographic entropy, then each attribute block's SAID, d, field provides a secure cryptographic digest of its contents that effectively blinds the attribute value from discovery given only its Schema and SAID. To clarify, the adversary despite being given both the schema of the attribute block and its SAID, d, field, is not able to discover the remaining contents of the attribute block in a computationally feasible manner such as a rainbow table attack . Therefore the UUID, u, field of each attribute block enables the associated SAID, d, field to securely blind the block's contents notwithstanding knowledge of the block's schema and that SAID, d, field. Moreover, a cryptographic commitment to that SAID, d, field does not provide a fixed point of correlation to the associated attribute (SAD) field values themselves unless and until there has been specific disclosure of those field values themselves. Given a total of N elements in the attributes array, let a_i represent the SAID, d, field of the attribute at zero-based index i. More precisely the set of attributes is expressed as the ordered set, {a_i for all i in {0, ..., N-1}}. The ordered set of a_i may be also expressed as a list, that is, [a₀, a₁, ...., a_N-1].

Composed Schema for Selectively Disclosable Attribute Section Because the selectively-disclosable attributes are provided by an array (list), the uncompacted variant in the schema uses an array of items and the anyOf composition operator to allow one or more of the items to be disclosed without requiring all to be disclosed. Thus both the oneOf and anyOf composition operators are used. The oneOf is used to provide compact partial disclosure of the aggregate, A, as the value of the top-level selectively-disclosable attribute section, A, field in its compact variant and the nested anyOf operator is used to enable selective disclosure in the uncompacted selectively-disclosable variant.

Inclusion Proof via Aggregated List Digest All the a_i in the list are aggregated into a single aggregate digest denoted A by computing the digest of their ordered concatenation. This is expressed as follows: A = H(C(a_i for all i in {0, ..., N-1})) where H is the digest (hash) operator and C is the concatentation operator. To be explicit, using the targeted example above, let a₀ denote the SAID of the Issuee attribute, a₁ denote the SAID of the score attribute, and a₂ denote the SAID of the name attribute then the aggregated digest A is computed as follows: A = H(C(a₀, a₁, a₂)). Equivalently using + as the infix concatenation operator, we have, A = H(a₀ + a₁ + a₂) Given sufficient collision resistance of the digest operator, the digest of an ordered concatenation is not subject to a birthday attack on its concatenated elements . In compact form, the value of the selectively-disclosable top-level attribute section, A, field is set to the aggregated value A. This aggregate A makes a blinded cryptographic commitment to the all the ordered elements in the list, [a₀, a₁, ...., a_N-1]. Moreover because each a_i element also makes a blinded commitment to its block's (SAD) attribute value(s), disclosure of any given a_i element does not expose or disclose any discoverable information detail about either its own or another block's attribute value(s). Therefore one may safely disclose the full list of a_i elements without exposing the blinded block attribute values. Proof of inclusion in the list consists of checking the list for a matching value. A computationally efficient way to do this is to create a hash table or B-tree of the list and then check for inclusion via lookup in the hash table or B-tree. To protect against later forgery given a later compromise of the signing keys of the Issuer, the issuer MUST anchor an issuance proof digest seal to the ACDC in its KEL. This seal binds the signing key state to the issuance. There are two cases. In the first case, an issuance/revocation registry is used. In the second case, an issuance/revocation registry is not used. When the ACDC is registered using an issuance/revocation TEL (Transaction Event Log) then the issuance proof seal digest is the SAID of the issuance (inception) event in the ACDC's TEL entry. The issuance event in the TEL includes the SAID of the ACDC. This binds the ACDC to the issuance proof seal in the Issuer's KEL through the TEL entry. When the ACDC is not registered using an issuance/revocation TEL then the issuance proof seal digest is the SAID of the ACDC itself. In either case, this issuance proof seal makes a verifiable binding between the issuance of the ACDC and the key state of the Issuer at the time of issuance. Because aggregated value A provided as the attribute section, A, field, value is bound to the SAID of the ACDC which is also bound to the key state via the issuance proof seal, the attribute details of each attribute block are also bound to the key state. The requirement of an anchored issuance proof seal means that the forger Must first successfully publish in the KEL of the issuer an inclusion proof digest seal bound to a forged ACDC. This makes any forgery attempt detectable. To elaborate, the only way to successfully publish such a seal is in a subsequent interaction event in a KEL that has not yet changed its key state via a rotation event. Whereas any KEL that has changed its key state via a rotation must be forked before the rotation. This makes the forgery attempt either both detectable and recoverable via rotation in any KEL that has not yet changed its key state or detectable as duplicity in any KEL that has changed its key state. In any event, the issuance proof seal ensures detectability of any later attempt at forgery using compromised keys. Given that aggregate value A appears as the compact value of the top-level attribute section, A, field, the selective disclosure of the attribute at index j may be proven to the disclosee with four items of information. These are:

• The actual detailed disclosed attribute block itself (at index j) with all its fields.
• The list of all attribute block digests, [a₀, a₁, ...., a_N-1] that includes a_j.
• The ACDC in compact form with selectively-disclosable attribute section, A, field value set to aggregate A.
• The signature(s), s, of the Issuee on the ACDC's top-level SAID, d, field.

The actual detailed disclosed attribute block is only disclosed after the disclosee has agreed to the terms of the rules section. Therefore, in the event the potential disclosee declines to accept the terms of disclosure, then a presentation of the compact version of the ACDC and/or the list of attribute digests, [a₀, a₁, ...., a_N-1]. does not provide any point of correlation to any of the attribute values themselves. The attributes of block j are hidden by a_j and the list of attribute digests [a₀, a₁, ...., a_N-1] is hidden by the aggregate A. The partial disclosure needed to enable chain-link confidentiality does not leak any of the selectively disclosable details. The disclosee may then verify the disclosure by: * computing a_j on the selectively disclosed attribute block details. * confirming that the computed a_j appears in the provided list [a₀, a₁, ...., a_N-1]. * computing A from the provided list [a₀, a₁, ...., a_N-1]. * confirming that the computed A matches the value, A, of the selectively-disclosable attribute section, A, field value in the provided ACDC. * computing the top-level SAID, d, field of the provided ACDC. * confirming the presence of the issuance seal digest in the Issuer's KEL * confirming that the issuance seal digest in the Issuer's KEL is bound to the ACDC top-level SAID, d, field either directly or indirectly through a TEL registry entry. * verifying the provided signature(s) of the Issuee on the provided top-level SAID, d field value. The last 3 steps that culminate with verifying the signature(s) require determining the key state of the Issuer at the time of issuance, this may require additional verification steps as per the KERI, PTEL, and CESR-Proof protocols. A private selectively disclosable ACDC provides significant correlation minimization because a presenter may use a metadata ACDC prior to acceptance by the disclosee of the terms of the chain-link confidentiality expressed in the rule section . Thus only malicious disclosees who violate chain-link confidentiality may correlate between presentations of a given private selectively disclosable ACDC. Nonetheless, they are not able to discover any undisclosed attributes.

Inclusion Proof via Merkle Tree Root Digest The inclusion proof via aggregated list may be somewhat verbose when there are a large number of attribute blocks in the selectively disclosable attribute section. A more efficient approach is to create a Merkle tree of the attribute block digests and let the aggregate, A, be the Merkle tree root digest . Specifically, set the value of the top-level selectively-disclosable attribute section, A, field to the aggregate, A whose value is the Merkle tree root digest . The Merkle tree needs to have appropriate second-pre-image attack protection of interior branch nodes . The discloser then only needs to provide a subset of digests from the Merkle tree to prove that a given digest, a_j contributed to the Merkle tree root digest, A. For ACDCs with a small number of attributes the added complexity of the Merkle tree approach may not be worth the savings in verbosity.

Hierarchical Derivation at Issuance of Selectively Disclosable Attribute ACDCs The amount of data transferred between the Issuer and Issuee (or recipient in the case of an untargeted ACDC) at issuance of a selectively disclosable attribute ACDC may be minimized by using a hierarchical deterministic derivation function to derive the value of the UUDI, u, fields from a shared secret salt . There are several ways that the Issuer may securely share that secret salt. Given that an Ed25519 key pair(s) controls each of the Issuer and Issuee AIDs, (or recipient AID in the case of an untargeted ACDC) a corresponding X15519 asymmetric encryption key pair(s) may be derived from each controlling Ed25519 key pair(s) . An X25519 public key may be derived from an Ed25519 public key. Likewise, an X25519 private key may be derived from an Ed25519 private key . In an interactive approach, the Issuer derives a public asymmetric X25519 encryption key from the Issuee's published Ed25519 public key and the Issuee derives a public asymmetric X25519 encryption key from the Issuer's published Ed25519 public key. The two then interact via a Diffie-Hellman (DH) key exchange to create a shared symmetric encryption key . The shared symmetric encryption key may be used to encrypt the secret salt or the shared symmetric encryption key itself may be used has high entropy cryptographic material from which the secret salt may be derived. In a non-interactive approach, the Issuer derives an X25519 asymmetric public encryption key from the Issuee's (recipient's) public Ed25519 public key. The Issuer then encrypts the secret salt with that public asymmetric encryption key and signs the encryption with the Issuer's private Ed25519 signing key. This is transmitted to the Issuee, who verifies the signature and decrypts the secret salt using the private X25519 decryption key derived from the Issuee's private Ed25519 key. This non-interactive approach is more scalable for AIDs that are controlled with a multi-sig group of signing keys. The Issuer can broadcast to all members of the Issuee's (or recipient's) multi-sig signing group individually asymmetrically encrypted and signed copies of the secret salt. In addition to the secret salt, the Issuer provides to the Issuee (recipient) a template of the ACDC but with empty UUID, u, and SAID, d, fields in each block with such fields. Each UUID, u, field value is then derived from the shared salt with a path prefix that indexes a specific block. Given the UUID, u, field value, the SAID, d, field value may then be derived. Likewise, both compact and uncompacted versions of the ACDC may then be generated. The derivation path for the top-level UUID, u, field (for private ACDCS), is the string "0" and derivation path the the the zeroth indexed attribute in the attributes array is the string "0/0". Likewise, the next attribute's derivation path is the string "0/1" and so forth. In addition to the shared salt and ACDC template, the Issuer also provides its signature(s) on its own generated compact version ACDC. The Issuer may also provide references to the anchoring issuance proof seals. Everything else an Issuee (recipient) needs to make a verifiable presentation/disclosure can be computed at the time of presentation/disclosure by the Issuee.

Bulk-Issued Private ACDCs The purpose of bulk issuance is to enable the Issuee to use unique ACDC more efficiently SAIDs to isolate and minimize correlation across different usage contexts of essentially the same ACDC while allowing public commitments to the ACDC SAIDs. A private ACDC may be issued in bulk as a set. In its basic form, the only difference between each ACDC is the top-level SAID, d, and UUID, u field values. To elaborate, bulk issuance enables the use of un-correlatable copies while minimizing the associated data transfer and storage requirements. Essentially each copy (member) of a bulk issued ACDC set shares a template that both the Issuer and Issuee use to generate a given ACDC in that set without requiring that the Issuer and Issuee exchange and store a unique copy of each member of the set independently. This minimizes the data transfer and storage requirements for both the Issuer and the Issuee. An ACDC provenance chain is connected via references to the SAIDs given by the top-level SAID, d, fields of the ACDCs in that chain. A given ACDC thereby makes commitments to other ACDCs. Expressed another way, an ACDC may be a node in a directed graph of ACDCs. Each directed edge in that graph emanating from one ACDC includes a reference to the SAID of some other connected ACDC. These edges provide points of correlation to an ACDC via their SAID reference. Private bulk issued ACDCs enable the Issuee to control better the correlatability of presentations using different presentation strategies. For example, the Issuee could use one copy of a bulk-issued private ACDC per presentation even to the same verifier. This strategy would consume the most copies. It is essentially a one-time-use ACDC strategy. Alternatively, the Issuee could use the same copy for all presentations to the same verifier and thereby only permit the verifier to correlate between presentations it received directly but not between other verifiers. This limits the consumption to one copy per verifier. In yet another alternative, the Issuee could use one copy for all presentations in a given context with a group of verifiers, thereby only permitting correlation among that group. In this context, we are talking about permissioned correlation. Any verifier that has received a complete presentation of a private ACDC has access to all the fields disclosed by the presentation but the terms of the chain-link confidentiality agreement may forbid sharing those field values outside a given context. Thus an Issuee may use a combination of bulk issued ACDCs with chain-link confidentiality to control permissioned correlation of the contents of an ACDC while allowing the SAID of the ACDC to be more public. The SAID of a private ACDC does not expose the ACDC contents to an un-permissioned third party. Unique SAIDs belonging to bulk issued ACDCs prevent third parties from making a provable correlation between ACDCs via their SAIDs in spite of those SAIDs being public. This does not stop malicious verifiers (as second parties) from colluding and correlating against the disclosed fields but it does limit provable correlation to the information disclosed to a given group of malicious colluding verifiers. To restate unique SAIDs per copy of a set of private bulk issued ACDC prevent un-permissioned third parties from making provable correlations in spite of those SAIDs being public unless they collude with malicious verifiers (second parties). In some applications, chain-link-confidentiality is insufficient to deter un-permissioned correlation. Some verifiers may be malicious with sufficient malicious incentives to overcome whatever counter incentives the terms of the contractual chain-link confidentiality may impose. In these cases, more aggressive technological anti-correlation mechanisms such as bulk issued ACDCs may be useful. To elaborate, in spite of the fact that chain-link confidentiality terms of use may forbid such malicious correlation, making such correlation more difficult technically may provide better protection than chain-link confidentiality alone [[41]]. It is important to note that any group of colluding malicious verifiers may always make a statistical correlation between presentations despite technical barriers to cryptographically provable correlation. In general, there is no cryptographic mechanism that precludes statistical correlation among a set of colluding verifiers because they may make cryptographically unverifiable or unprovable assertions about information presented to them that may be proven as likely true using merely statistical correlation techniques.

Basic Bulk Issuance The amount of data transferred between the Issuer and Issuee (or recipient of an untargeted ACDC) at issuance of a set of bulk issued ACDCs may be minimized by using a hierarchical deterministic derivation function to derive the value of the UUID, u, fields from a shared secret salt . As described above, there are several ways that the Issuer may securely share a secret salt. Given that the Issuer and Issuee (or recipient when untargeted) AIDs are each controlled by an Ed25519 key pair(s), a corresponding X15519 asymmetric encryption key pair(s) may be derived from the controlling Ed25519 key pair(s) . An X25519 public key may be derived from an Ed25519 public key. Likewise, an X25519 private key may be derived from an Ed25519 private key . In an interactive approach, the Issuer derives a public asymmetric X25519 encryption key from the Issuee's published Ed25519 public key and the Issuee derives a public asymmetric X25519 encryption key from the Issuer's published Ed25519 public key. The two then interact via a Diffie-Hellman (DH) key exchange to create a shared symmetric encryption key . The shared symmetric encryption key may be used to encrypt the secret salt or the shared symmetric encryption key itself may be used has high entropy cryptographic material from which the secret salt may be derived. In a non-interactive approach, the Issuer derives an X25519 asymmetric public encryption key from the Issuee's (or recipient's) public Ed25519 public key. The Issuer then encrypts the secret salt with that public asymmetric encryption key and signs the encryption with the Issuer's private Ed25519 signing key. This is transmitted to the Issuee, who verifies the signature and decrypts the secret salt using the private X25519 decryption key derived from the Issuee's private Ed25519 key. This non-interactive approach is more scalable for AIDs that are controlled with a multi-sig group of signing keys. The Issuer can broadcast to all members of the Issuee's (or recipient's) multi-sig signing group individually asymmetrically encrypted and signed copies of the secret salt. In addition to the secret salt, the Issuer also provides a template of the private ACDC but with empty UUID, u, and SAID, d, fields at the top-level of each nested block with such fields. Each UUID, u, field value is then derived from the shared salt with a deterministic path prefix that indexes both its membership in the bulk issued set and its location in the ACDC. Given the UUID, u, field value, the associated SAID, d, field value may then be derived. Likewise, both full and compact versions of the ACDC may then be generated. This generation is analogous to that described in the section for selective disclosure ACDCs but extended to a set of private ACDCs. The initial element in each deterministic derivation path is the string value of the bulk-issued member's copy index k, such as "0", "1", "2" etc. Specifically, if k denotes the index of an ordered set of bulk issued private ACDCs of size M, the derivation path starts with the string "k" where k is replaced with the decimal or hexadecimal textual representation of the numeric index k. Furthermore, a bulk-issued private ACDC with a private attribute section uses "k" to derive its top-level UUID and "k/0" to derive its attribute section UUID. This hierarchical path is extended to any nested private attribute blocks. This approach is further extended to enable bulk issued selective disclosure ACDCs by using a similar hierarchical derivation path for the UUID field value in each of the selectively disclosable blocks in the array of attributes. For example, the path "k/j" is used to generate the UUID of attribute index j at bulk-issued ACDC index k. In addition to the shared salt and ACDC template, the Issuer also provides a list of signatures of SAIDs, one for each SAID of each copy of the associated compact bulk-issued ACDC. The Issuee (or recipient) can generate on-demand each compact or uncompacted ACDC from the template, the salt, and its index k. The Issuee does not need to store a copy of each bulk issued ACDC, merely the template, the salt, and the list of signatures. The Issuer MUST also anchor in its KEL an issuance proof digest seal of the set of bulk issued ACDCs. The issuance proof digest seal makes a cryptographic commitment to the set of top-level SAIDS belonging to the bulk issued ACDCs. This protects against later forgery of ACDCs in the event the Issuer's signing keys become compromised. A later attempt at forgery requires a new event or new version of an event that includes a new anchoring issuance proof digest seal that makes a cryptographic commitment to the set of newly forged ACDC SAIDS. This new anchoring event of the forgery is therefore detectable. Similarly, to the process of generating a selective disclosure attribute ACDC, the issuance proof digest is an aggregate that is aggregated from all members in the bulk-issued set of ACDCs. The complication of this approach is that it must be done in such a way as to not enable provable correlation by a third party of the actual SAIDS of the bulk-issued set of ACDCs. Therefore the actual SAIDs must not be aggregated but blinded commitments to those SAIDs instead. With blinded commitments, knowledge of any or all members of such a set does not disclose the membership of any SAID unless and until it is unblinded. Recall that the purpose of bulk issuance is to allow the SAID of an ACDC in a bulk issued set to be used publicly without correlating it in an un-permissioned provable way to the SAIDs of the other members. The basic approach is to compute the aggregate denoted, B, as the digest of the concatenation of a set of blinded digests of bulk issued ACDC SAIDS. Each ACDC SAID is first blinded via concatenation to a UUID (salty nonce) and then the digest of that concatenation is concatenated with the other blinded SAID digests. Finally, a digest of that concatenation provides the aggregate. Suppose there are M ACDCs in a bulk issued set. Using zero-based indexing for each member of the bulk issued set of ACDCs, such that index k satisfies k in {0, ..., M-1}, let *d_k denote the top-level SAID of an ACDC in an ordered set of bulk-issued ACDCs. Let v_k denote the UUID (salty nonce) or blinding factor that is used to blind that said. The blinding factor, v_k, is NOT the top-level UUID, u, field of the ACDC itself but an entirely different UUID used to blind the ACDC's SAID for the purpose of aggregation. The derivation path for v_k from the shared secret salt is "k." where k is the index of the bulk-issued ACDC. Let c_k = v_k + d_k, denote the blinding concatenation where + is the infix concatenation operator.
Then the blinded digest, b_k, is given by,
b_k = H(c_k) = H(v_k + d_k),
where H is the digest operator. The aggregation of blinded digests, B, is given by,
B = H(C(b_k for all k in {0, ..., M-1})),
where C is the concatenation operator and H is the digest operator. This aggregate, B, provides the issuance proof digest. The aggregate, B, makes a blinded cryptographic commitment to the ordered elements in the list [b₀, b₁, ...., b_M-1]. A commitment to B is a commitment to all the b_k and hence all the d_k. Given sufficient collision resistance of the digest operator, the digest of an ordered concatenation is not subject to a birthday attack on its concatenated elements . Disclosure of any given b_k element does not expose or disclose any discoverable information detail about either the SAID of its associated ACDC or any other ACDC's SAID. Therefore one may safely disclose the full list of b_k elements without exposing the blinded bulk issued SAID values, d_k. Proof of inclusion in the list of blinded digests consists of checking the list for a matching value. A computationally efficient way to do this is to create a hash table or B-tree of the list and then check for inclusion via lookup in the hash table or B-tree. A proof of inclusion of an ACDC in a bulk-issued set requires disclosure of v_k which is only disclosed after the disclosee has accepted (agreed to) the terms of the rule section. Therefore, in the event the Disclosee declines to accept the terms of disclosure, then a presentation/disclosure of the compact version of the ACDC does not provide any point of correlation to any other SAID of any other ACDC from the bulk set that contributes to the aggregate B. In addition, because the other SAIDs are hidden by each b_k inside the aggregate, B, even a presentation/disclosure of,
[b₀, b₁, ...., b_M-1]
does not provide any point of correlation to the actual bulk-issued ACDC without disclosure of its v_k. Indeed if the Discloser uses a metadata version of the ACDC in its offer then even its SAID is not disclosed until after acceptance of terms in the rule section. To protect against later forgery given a later compromise of the signing keys of the Issuer, the issuer MUST anchor an issuance proof seal to the ACDC in its KEL. This seal binds the signing key state to the issuance. There are two cases. In the first case, an issuance/revocation registry is used. In the second case, an issuance/revocation registry is not used. When the ACDC is registered using an issuance/revocation TEL (Transaction Event Log) then the issuance proof seal digest is the SAID of the issuance (inception) event in the ACDC's TEL entry. The issuance event in the TEL uses the aggregate value, B, as its identifier value. This binds the aggregate, B, to the issuance proof seal in the Issuer's KEL through the TEL. Recall that the usual purpose of a TEL is to provide a verifiable data registry that enables dynamic revocation of an ACDC via a state of the TEL. A verifier checks the state at the time of use to check if the associated ACDC has been revoked. The Issuer controls the state of the TEL. The registry identifier, ri, field is used to identify the public registry which usually provides a unique TEL entry for each ACDC. Typically the identifier of each TEL entry is the SAID of the TEL's inception event which is a digest of the event's contents which include the SAID of the ACDC. In the bulk issuance case, however, the TEL's inception event contents include the aggregate, B, instead of the SAID of a given ACDC. Recall that the goal is to generate an aggregate value that enables an Issuee to selectively disclose one ACDC in a bulk-issued set without leaking the other members of the set to un-permissioned parties (second or third). Using the aggregate, B of blinded ACDC saids as the TEL registry entry identifier allows all members of the bulk-issued set to share the same TEL without any third party being able to discover which TEL any ACDC is using in an un-permissioned provable way. Moreover, a second party may not discover in an un-permissioned way any other ACDCs from the bulk-issued set not specifically disclosed to that second party. In order to prove to which TEL a specific bulk issued ACDC belongs, the full inclusion proof must be disclosed. When the ACDC is not registered using an issuance/revocation TEL then the issuance proof seal digest is the aggregate, B, itself. In either case, this issuance proof seal makes a verifiable binding between the issuance of all the ACDCs in the bulk issued set and the key state of the Issuer at the time of issuance. A Discloser may make a basic provable non-repudiable selective disclosure of a given bulk issued ACDC, at index k by providing to the Disclosee four items of information (proof of inclusion). These are as follows:

• The ACDC in compact form (at index k) where d_k as the value of its top-level SAID, d, field.
• The blinding factor, v_k from which b_k = H(v_k + d_k) may be computed.
• The list of all blinded SAIDs, [b₀, b₁, ...., b_M-1] that includes b_k.
• The signature(s), s_k, of the Issuee on the ACDC's top level SAID, d_k, field.

A Disclosee may then verify the disclosure by:

• computing d_j on the disclosed compact ACDC.
• computing b_k = H(v_k + d_k)
• confirming that the computed b_k appears in the provided list [b₀, b₁, ...., b_M-1].
• computing the aggregate B from the provided list [b₀, b₁, ...., b_M-1]..
• confirming the presence of an issuance seal digest in the Issuer's KEL that makes a commitment to the aggregate, B, either directly or indirectly through a TEL registry entry.
• verifying the provided signature(s), s_k, of the Issuee on the provided top level SAID, d_k, field.

The last 3 steps that culminate with verifying the signature(s) require determining the key state of the Issuer at the time of issuance, this may require additional verification steps as per the KERI, PTEL, and CESR-Proof protocols. The requirement of an anchored issuance proof seal means that the forger Must first successfully publish in the KEL of the issuer an inclusion proof digest seal bound to a set of forged bulk issued ACDCs. This makes any forgery attempt detectable. To elaborate, the only way to successfully publish such a seal is in a subsequent interaction event in a KEL that has not yet changed its key state via a rotation event. Whereas any KEL that has changed its key state via a rotation must be forked before the rotation. This makes the forgery attempt either both detectable and recoverable via rotation in any KEL that has not yet changed its key state or detectable as duplicity in any KEL that has changed its key state. In any event, the issuance proof seal makes any later attempt at forgery using compromised keys detectable.

Inclusion Proof via Merkle Tree The inclusion proof via aggregated list may be somewhat verbose when there are a very large number of bulk issued ACDCs in a given set. A more efficient approach is to create a Merkle tree of the blinded SAID digests, b_k and set the aggregate B value as the Merkle tree root . The Merkle tree needs to have appropriate second-pre-image attack protection of interior branch nodes . The discloser then only needs to provide a subset of digests from the Merkle tree to prove that a given digest, b_k contributed to the Merkle tree root digest. For a small numbered bulk issued set of ACDCs, the added complexity of the Merkle tree approach may not be worth the savings in verbosity.

Bulk Issuance of Private ACDCs with Unique Issuee AIDs One potential point of provable but un-permissioned correlation among any group of colluding malicious Disclosees (Second-Party verifiers) may arise when the same Issuee AID is used for presentation/disclosure to all Disclosees in that group. Recall that the contents of private ACDCs are not disclosed except to permissioned Disclosees (Second-Parties), thus a common Issuee AID would only be a point of correlation for a group of colluding malicious verifiers. But in some cases removing this un-permissioned point of correlation may be desirable. One solution to this problem is for the Issuee to use a unique AID for the copy of a bulk issued ACDC presented to each Disclosee in a given context. This requires that each ACDC copy in the bulk-issued set use a unique Issuee AID. This would enable the Issuee in a given context to minimize provable correlation by malicious Disclosees against any given Issuee AID. In this case, the bulk issuance process may be augmented to include the derivation of a unique Issuee AID in each copy of the bulk-issued ACDC by including in the inception event that defines a given Issuee's self-addressing AID, a digest seal derived from the shared salt and copy index k. The derivation path for the digest seal is "k/0." where k is the index of the ACDC. To clarify "k/0." specifies the path to generate the UUID to be included in the inception event that generates the Issuee AID for the ACDC at index k. This can be generated on-demand by the Issuee. Each unique Issuee AID would also need its own KEL. But generation and publication of the associated KEL can be delayed until the bulk-issued ACDC is actually used. This approach completely isolates a given Issuee AID to a given context with respect to the use of a bulk-issued private ACDC. This protects against even the un-permissioned correlation among a group of malicious Disclosees (Second Parties) via the Issuee AID.

Independent TEL Bulk-Issued ACDCs Recall that the purpose of using the aggregate B for a bulk-issued set from which the TEL identifier is derived is to enable a set of bulk issued ACDCs to share a single public TEL that provides dynamic revocation but without enabling un-permissioned correlation to any other members of the bulk set by virtue of the shared TEL. This enables the issuance/revocation/transfer state of all copies of a set of bulk-issued ACDCs to be provided by a single TEL which minimizes the storage and compute requirements on the TEL registry while providing selective disclosure to prevent un-permissioned correlation via the public TEL. However, in some applications where chain-link confidentiality does not sufficiently deter malicious provable correlation by Disclosees (Second-Party verifiers), an Issuee may benefit from using ACDC with independent TELs but that are still bulk-issued. In this case, the bulk issuance process must be augmented so that each uniquely identified copy of the ACDC gets its own TEL entry in the registry. Each Disclosee (verifier) of a full presentation/disclosure of a given copy of the ACDC only receives proof of one uniquely identified TEL and can NOT provably correlate the TEL state of one presentation to any other presentation because the ACDC SAID, the TEL identifier, and the signature of the issuer on the SAID of a given copy will all be different for each copy. There is therefore no point of provable correlation permissioned or otherwise. The obvious drawbacks of this approach (independent unique TELs for each private ACDC) are that the size of the registry database increases as a multiple of the number of copies of each bulk-issued ACDC and every time an Issuer must change the TEL state of a given set of copies it must change the state of multiple TELs in the registry. This imposes both a storage and computation burden on the registry. The primary advantage of this approach, however, is that each copy of a private ACDC has a uniquely identified TEL. This minimizes un-permissioned Third-Party exploitation via provable correlation of TEL identifiers even with colluding Second-Party verifiers. They are limited to statistical correlation techniques. In this case, the set of private ACDCs may or may not share the same Issuee AID because for all intents and purposes each copy appears to be a different ACDC even when issued to the same Issuee. Nonetheless, using unique Issuee AIDs may further reduce correlation by malicious Disclosees (Second-Party verifiers) beyond using independent TELs. To summarize the main benefit of this approach, in spite of its storage and compute burden, is that in some applications chain-link confidentiality does not sufficiently deter un-permissioned malicious collusion. Therefore completely independent bulk-issued ACDCs may be used.

Appendix: Cryptographic Strength and Security

Cryptographic Strength For crypto-systems with perfect-security, the critical design parameter is the number of bits of entropy needed to resist any practical brute force attack. In other words, when a large random or pseudo-random number from a cryptographic strength pseudo-random number generator (CSPRNG) expressed as a string of characters is used as a seed or private key to a cryptosystem with perfect-security, the critical design parameter is determined by the amount of random entropy in that string needed to withstand a brute force attack. Any subsequent cryptographic operations must preserve that minimum level of cryptographic strength. In information theory the entropy of a message or string of characters is measured in bits. Another way of saying this is that the degree of randomness of a string of characters can be measured by the number of bits of entropy in that string. Assuming conventional non-quantum computers, the convention wisdom is that, for systems with information-theoretic or perfect security, the seed/key needs to have on the order of 128 bits (16 bytes, 32 hex characters) of entropy to practically withstand any brute force attack. A cryptographic quality random or pseudo-random number expressed as a string of characters will have essentially as many bits of entropy as the number of bits in the number. For other crypto-systems such as digital signatures that do not have perfect security, the size of the seed/key may need to be much larger than 128 bits in order to maintain 128 bits of cryptographic strength. An N-bit long base-2 random number has 2^N different possible values. Given that no other information is available to an attacker with perfect security, the attacker may need to try every possible value before finding the correct one. Thus the number of attempts that the attacker would have to try maybe as much as 2^N-1. Given available computing power, one can easily show that 128 is a large enough N to make brute force attack computationally infeasible. Let's suppose that the adversary has access to supercomputers. Current supercomputers can perform on the order of one quadrillion operations per second. Individual CPU cores can only perform about 4 billion operations per second, but a supercomputer will parallelly employ many cores. A quadrillion is approximately 2⁵⁰ = 1,125,899,906,842,624. Suppose somehow an adversary had control over one million (2²⁰ = 1,048,576) supercomputers which could be employed in parallel when mounting a brute force attack. The adversary could then try 2⁵⁰ * 2²⁰ = 2⁷⁰ values per second (assuming very conservatively that each try only took one operation). There are about 3600 * 24 * 365 = 313,536,000 = 2^{log₂313536000}=2^24.91 ~= 2²⁵ seconds in a year. Thus this set of a million super computers could try 2^50+20+25 = 2⁹⁵ values per year. For a 128-bit random number this means that the adversary would need on the order of 2^128-95 = 2³³ = 8,589,934,592 years to find the right value. This assumes that the value of breaking the cryptosystem is worth the expense of that much computing power. Consequently, a cryptosystem with perfect security and 128 bits of cryptographic strength is computationally infeasible to break via brute force attack.

Information Theoretic Security and Perfect Security The highest level of cryptographic security with respect to a cryptographic secret (seed, salt, or private key) is called information-theoretic security . A cryptosystem that has this level of security cannot be broken algorithmically even if the adversary has nearly unlimited computing power including quantum computing. It must be broken by brute force if at all. Brute force means that in order to guarantee success the adversary must search for every combination of key or seed. A special case of information-theoretic security is called perfect-security . Perfect-security means that the ciphertext provides no information about the key. There are two well-known cryptosystems that exhibit perfect security. The first is a one-time-pad (OTP) or Vernum Cipher , the other is secret splitting , a type of secret sharing that uses the same technique as a one-time-pad.

Conventions and Definitions The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 when, and only when, they appear in all capitals, as shown here.

• SAID - Self-Addressing Identifier - any identifier which is deterministaclly generated out of the content, digest of the content

Security Considerations TODO Security

IANA Considerations This document has no IANA actions.

References Normative References ProSapien LLC ProSapien LLC ProSapien LLC ProSapien LLC GLEIF GLEIF GLEIF GLEIF ProSapien LLC n.d. n.d. n.d. n.d. n.d. n.d. n.d. In many standards track documents several words are used to signify the requirements in the specification. These words are often capitalized. This document defines these words as they should be interpreted in IETF documents. This document specifies an Internet Best Current Practices for the Internet Community, and requests discussion and suggestions for improvements. RFC 2119 specifies common key words that may be used in protocol specifications. This document aims to reduce the ambiguity by clarifying that only UPPERCASE usage of the key words have the defined special meanings. Informative References n.d. n.d. n.d. n.d. n.d. ProSapien LLC n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d.

Acknowledgments TODO acknowledge.