Identifier Scheme

An identifier scheme is itself a referent that must be identifiable, scoped and statused before any identifier under it can be interpreted. Its identity comes from an entry in a governing registry (an IANA URN namespace identifier, a URI scheme name, an ISO 6523 code designator, a DID method name), and its scope is the declared class of referents it may name. The namespace expression is the token a consuming model carries next to the value so that the value can be interpreted at all.

Questions

1. Which authoritative registry entry establishes this scheme, and under what key?identity
   Expected answer

   • Governing registry name and URL
   • Registration key as issued (namespace identifier, scheme name, code designator, method name)
   • Registration date and reference document
2. What class of referents may be named under this scheme, and what is explicitly excluded?classification
   Expected answer

   • Referent class statement
   • Explicit exclusions
   • Representative in-scope and out-of-scope examples
3. What namespace expression must accompany a value of this scheme when it is carried inside another model?interoperability
   Expected answer

   • Absolute namespace URI
   • Alternative tokens such as namespace identifier or numeric code designator
   • Statement of which token is canonical for exchange
4. Is the scheme versioned, and does a version change alter the meaning of identifiers already assigned?temporal
   Expected answer

   • Scheme version label
   • Version effective date as an RFC 3339 timestamp
   • Backward-compatibility statement for prior assignments
5. What is the scheme's registration status and when did that status last change?state
   Expected answer

   • Status value such as permanent, provisional, historical, formal or informal
   • Status effective timestamp
   • Reason for the most recent status change

Artifacts

• Scheme registration recordThe registry entry constituting the scheme, holding its key, namespace expression, referent class, status, reference document and change controller.

The decidable definition of a well-formed identifier: a formal grammar, the permitted character repertoire and encoding, length bounds, case treatment, and which substrings are structurally significant. RFC 8141 requires a syntax statement in every namespace registration and constrains namespace identifiers to a bounded alphanumeric pattern; RFC 3986 requires a scheme's own syntax to remain within the generic grammar; DOI names divide at a forward slash into an issuer prefix and a registrant-chosen suffix; the LEI is a fixed twenty-character code whose leading characters denote the issuing unit.

Questions

1. What machine-checkable expression defines a well-formed identifier of this scheme?constraint
   Expected answer

   • Formal grammar or pattern expression
   • Notation used and its version
   • Location of the authoritative expression in the specification
2. Which characters and encodings are permitted, and how is case treated?constraint
   Expected answer

   • Permitted character repertoire
   • Encoding rules including any escaping
   • Case sensitivity statement
3. Which substrings of an identifier are structurally significant, and what do they delimit?composition
   Expected answer

   • Named components and their delimiters
   • Which component denotes the issuing authority
   • Whether component boundaries are fixed-width or delimiter-driven
4. What minimum and maximum lengths apply, and are they enforced at assignment or only at validation?constraint
   Expected answer

   • Minimum and maximum length
   • Enforcement point
   • Behaviour on out-of-range input
5. How are historical values that fail the current grammar handled?exception
   Expected answer

   • Count and examples of non-conforming legacy values
   • Tolerance or quarantine policy
   • Migration or annotation approach

Artifacts

• Identifier syntax specificationThe clause or document holding the authoritative grammar, repertoire, length and structure rules for the scheme.
• Syntax test vector setPaired accept and reject examples that let an implementation prove it applies the grammar correctly, including boundary and legacy cases.

Many schemes carry a self-checking element - LEI check digits, an ARK check character computed over a defined check zone, or the version and variant bits of a UUID - that detects transcription error. This finding separates three distinct claims that are routinely confused: the string is well-formed, the string passes its internal check, and the string has actually been assigned to a referent by the authority. Only the third is evidence of existence.

Questions

1. What check character algorithm applies, and over which part of the string is it computed?measurement
   Expected answer

   • Algorithm name and reference
   • Definition of the check zone
   • Position and width of the check element
2. What does a passing check prove, and what does it not prove?evidence
   Expected answer

   • Detected error classes
   • Explicit statement that a passing check does not prove assignment
   • Residual undetected error rate if published
3. Which authority must be consulted to confirm that a well-formed identifier has actually been assigned?authority
   Expected answer

   • Confirming registry or service
   • Query interface reference
   • Expected latency and coverage of the confirmation
4. If the scheme has no check element, what compensating control detects transcription error?quality
   Expected answer

   • Compensating control description
   • Observed or estimated transcription error rate
   • Point in the workflow where the control is applied

Artifacts

• Validation conformance reportRecord of a validation run over a population of values, reporting well-formedness, check-element failures and unconfirmed assignments separately.

Comparison is not a generic operation. RFC 3986 sets out a ladder from simple string comparison through syntax-based and scheme-based normalization to protocol-based normalization, with each rung raising both true-positive and false-positive rates. RFC 8141 requires case normalization of the scheme token and namespace identifier while forbidding percent-decoding and requiring optional components to be ignored. XML namespace names are compared as identical character sequences with no escaping done or undone. ARK normalization removes the resolver host, lowercases the authority number and treats hyphens as insignificant. A single global normalizer applied across schemes will merge distinct referents.

Questions

1. What is the canonical form of an identifier of this scheme, and how is it derived?definition
   Expected answer

   • Canonical form definition
   • Ordered normalization procedure
   • Worked example from raw input to canonical output
2. Which transformations preserve identity for this scheme and which must never be applied?constraint
   Expected answer

   • Permitted transformations
   • Forbidden transformations with the harm each would cause
   • Handling of optional or trailing components
3. Which comparison level is a consumer entitled to rely on without contacting the authority?constraint
   Expected answer

   • Highest safe comparison level
   • Operations requiring authority contact
   • Known false-positive and false-negative modes
4. Which form is stored and indexed, and is the form as received also retained?quality
   Expected answer

   • Stored form policy
   • Whether the as-received form is preserved
   • Index and collation rules applied
5. Does the scheme permit more than one identifier for the same referent, and how are aliases declared?relationship
   Expected answer

   • Alias permission statement
   • Preferred or canonical identifier designation
   • Mechanism for declaring an alias relationship

Artifacts

• Canonicalization test vector setInput and expected-canonical-output pairs, including pairs that must remain distinct, used to prove a normalizer matches the scheme's rules.

W3C web architecture advises that agents should not infer properties of a referent from its identifier, and ARK strongly encourages semantic opacity so that names survive changes in the world. Yet delegated-prefix schemes deliberately embed the issuing authority in the string, which is exactly what makes decentralised allocation work. This finding records what the scheme permits to be inferred, what must never be inferred, and what the structure discloses to an observer who was never meant to know it.

Questions

1. What may an agent legitimately infer from the structure of an identifier of this scheme?interoperability
   Expected answer

   • Permitted inferences with the component each rests on
   • Specification clause authorising each inference
   • Stability guarantee attached to the inference
2. What must an agent never infer from an identifier value, even when the pattern appears to support it?constraint
   Expected answer

   • Forbidden inferences
   • Concrete failure case for each
   • Whether the specification states the prohibition explicitly
3. What does the identifier disclose to an observer who holds only the string?access
   Expected answer

   • Disclosed attributes such as issuer, sequence position, creation time or category
   • Assessment of whether disclosure is intended
   • Mitigation where disclosure is unintended
4. If an embedded attribute of the referent changes, does the identifier change?lifecycle
   Expected answer

   • Policy on identifier change following referent change
   • Examples of attribute changes that do and do not trigger reassignment
   • Consequence for existing references

Artifacts

• Semantic disclosure statementPublished statement of what the scheme's identifiers do and do not encode, written for consumers who must decide whether to parse them.

A scheme's credibility rests on a named party holding an identifiable mandate. RFC 8141 asks a formal-namespace registrant to demonstrate organizational stability, competence in assignment and a commitment not to reassign. ITU-T X.660 establishes published procedures for object identifier registration authorities. The LEI system separates three roles that are often collapsed: a regulatory oversight committee, a foundation acting as administrator, and accredited local operating units that actually issue. The DOI system likewise separates a single registration authority from a network of registration agencies.

Questions

1. Who holds the mandate for this scheme, and under what instrument was it granted?authority
   Expected answer

   • Mandated party with a reference to the organization record
   • Instrument type such as international standard, statute, charter or registry designation
   • Mandate start date and any expiry
2. Which body oversees the mandate holder, and what powers does it hold?authority
   Expected answer

   • Oversight body reference
   • Enumerated powers including accreditation and withdrawal
   • Reporting cadence
3. How are issuing bodies accredited, and how is accreditation withdrawn?process
   Expected answer

   • Accreditation criteria and process
   • Current accredited issuer list reference
   • Withdrawal grounds and effect on already-issued identifiers
4. What evidence supports the mandate holder's ability to sustain the scheme over its intended horizon?evidence
   Expected answer

   • Stability and competence evidence as filed
   • Stated intended horizon
   • Funding basis
5. What happens to assignments and resolution if the mandate holder ceases to operate?exception
   Expected answer

   • Named successor or escrow arrangement
   • Continuity commitment text
   • Evidence that the arrangement has been tested

Artifacts

• Mandate instrument recordRecord of the instrument conferring scheme authority, its parties, scope, period and oversight arrangements.
• Accredited issuer registerList of bodies currently accredited to issue under the scheme, with accreditation state and effective dates.

Hierarchical allocation lets delegated administrators control portions of a namespace while a higher authority manages the levels above, and RFC 8126 names it as a policy in its own right. The Handle System builds naming authorities as a dotted tree in which global uniqueness follows from a globally unique authority combined with a locally unique name. DOI achieves decentralised allocation the same way. Delegation raises questions the parent scheme must answer: is a delegated prefix exclusive against longer strings sharing its leading characters, who may change an entry once delegated, and what happens to identifiers already issued when a delegation is revoked.

Questions

1. How is allocation authority subdivided, and what is the shape of the delegation tree?composition
   Expected answer

   • Delegation mechanism description
   • Depth and branching rules
   • Current delegation chain for a given identifier
2. Is a delegated namespace portion exclusive, and does it foreclose overlapping allocations?constraint
   Expected answer

   • Exclusivity rule text
   • Treatment of longer strings sharing a delegated prefix
   • Conflict resolution procedure
3. Who is the change controller for a given registry entry or delegated namespace?ownership
   Expected answer

   • Change controller reference per entry
   • Scope of changes the controller may make
   • Procedure for transferring change control
4. On what grounds can a delegation be revoked, and what becomes of identifiers already issued under it?lifecycle
   Expected answer

   • Revocation grounds
   • Effect on outstanding assignments
   • Notice period and migration path
5. May a delegate sub-delegate further, and is the parent bound by the delegate's policy?authority
   Expected answer

   • Sub-delegation permission statement
   • Policy inheritance rule
   • Liability and audit reach of the parent

Artifacts

• Delegation registerRegister of delegated namespace portions with delegate, exclusivity, change controller and effective period.

RFC 8126 supplies the working vocabulary: Private Use and Experimental Use where nothing is recorded centrally, Hierarchical Allocation where administrators control portions, First Come First Served with no substantive review, Expert Review, Specification Required which additionally demands a permanent and readily available public specification, and the progressively stricter document-based policies. It also distinguishes Unassigned from Reserved and introduces Known Unregistered Use as a warning marker. A scheme that cannot name its policy cannot tell a consumer how much review any given identifier has had.

Questions

1. Which allocation policy governs new registrations under this scheme?authority
   Expected answer

   • Named policy value
   • Clause establishing it
   • Whether different ranges carry different policies
2. If review is required, who performs it and against what published criteria?process
   Expected answer

   • Reviewer identity or role
   • Published review criteria
   • Appeal route against a refusal
3. Which value ranges are reserved, private-use, experimental or otherwise withheld from allocation?classification
   Expected answer

   • Range boundaries with status for each
   • Rationale for each reservation
   • Whether reserved values may later be released
4. Can the allocation policy itself be changed, and by what procedure?process
   Expected answer

   • Amendment procedure
   • Effect on registrations made under a prior policy
   • Most recent amendment with its RFC 3339 timestamp
5. How does the scheme record values known to be in use without a registration?exception
   Expected answer

   • Marker used for known unregistered use
   • Discovery and confirmation procedure
   • Whether such values can be regularised

Artifacts

• Allocation policy statementPublished statement naming the allocation policy, the reviewers, the reserved ranges and the amendment procedure.

RFC 8141 makes a namespace registration carry purpose, syntax including equivalence rules, assignment mechanism, security and privacy analysis, interoperability conflicts, and a resolution statement, with a pointer to a stable specification. BCP 35 adds status, applications and protocols, contact and change controller. RFC 8126 further allows a registry to be closed or marked obsolete and individual entries to be deprecated with annotations preserved for history. These filed statements are the primary machine-readable evidence about a scheme and are frequently the only evidence available to a consuming agent.

Questions

1. Which registration template fields were filed, and which mandatory fields are missing or empty?evidence
   Expected answer

   • Field-by-field completeness assessment
   • Missing mandatory fields
   • Date the registration was last revised
2. Where is the permanent and readily available public specification for this scheme?provenance
   Expected answer

   • Specification location and stable reference
   • Availability terms including any paywall
   • Version currently in force
3. What security and privacy analysis was filed with the registration?evidence
   Expected answer

   • Filed comparison, leakage and privacy statements
   • Known gaps in the analysis
   • Date of the analysis
4. How is a registry entry deprecated or obsoleted, and what annotation survives?lifecycle
   Expected answer

   • Deprecation mechanism
   • Annotation retained for historical entries
   • Whether the registry as a whole can be closed

Artifacts

• Registration template submissionThe completed registration template as filed with the governing registry, including purpose, syntax, assignment, security and privacy, interoperability and resolution statements.

The assignment is the atomic fact this model exists to support. The FHIR Identifier datatype gives the minimal shape a consumer needs: a system establishing the namespace, a value unique within it, an optional coded type, a use marker that modifies interpretation, a validity period and an assigner. DataCite shows that a registration agency may impose mandatory descriptive metadata as a condition of assignment. RFC 8141 requires the assignment mechanism and its uniqueness method to be specified. Event time and record time must be kept apart: when the binding took effect is not when the register learned of it.

Questions

1. Which referent is bound by this assignment, and how is that referent itself identified?identity
   Expected answer

   • Referent reference with its own identifier and scheme
   • Referent class
   • Statement of whether the binding is exclusive to one referent
2. Who performed the assignment, under which delegated authority?authority
   Expected answer

   • Assigner reference
   • Delegated namespace portion used
   • Authorisation basis
3. When did the assignment take effect, and when was it recorded?temporal
   Expected answer

   • Assignment effective time in RFC 3339
   • Record or ingestion time in RFC 3339
   • Reason for any gap between the two
4. What evidence supports that the referent is what the registrant claims it to be?evidence
   Expected answer

   • Evidence artefacts referenced
   • Verification level applied
   • Party that performed verification
5. What metadata or conduct obligations does the registrant accept as a condition of assignment?requirement
   Expected answer

   • Mandatory metadata properties
   • Update and renewal duties
   • Consequence of non-compliance
6. What is the intended use of this identifier relative to any others held by the same referent?classification
   Expected answer

   • Use marker such as official, usual, secondary, temporary or old
   • Coded identifier type
   • Validity period

Artifacts

• Assignment registerThe authoritative register of bindings between identifiers and referents, with assigner, effective and record times, evidence and obligatory metadata.
• Registration metadata recordThe descriptive metadata a registration agency requires as a condition of assignment, including any alternate identifiers held by the same referent.

ISO 26324 and ISO 24619 require explicit granularity. ARK qualifiers identify parts and variants. SKOS notations are lexical codes unique within a concept scheme and are not a substitute for URIs. Fragments and f-components identify constituents without changing the assigned name.

Questions

1. At what granularity is the identifier defined, and may finer parts be addressed by qualifiers or fragments rather than new assignments?composition
   Expected answer

   • granularityLevel
   • qualifierSyntax
   • fragmentSemantics
2. If a notation or code is used, is it unique only within a concept scheme, and what URI or URN is the preferred computational identifier?interoperability
   Expected answer

   • notationValue
   • conceptSchemeRef
   • preferredUri
3. Do content-negotiated, language, format or version variants share the assigned name or require distinct identifiers?classification
   Expected answer

   • variantPolicy
   • contentNegotiationFlag
   • versionedIdPolicy

Identifiers are not simply present or absent. The LEI system publishes an explicit registration status vocabulary covering issued, lapsed, merged, retired, annulled, duplicate and transfer states, tied to a renewal obligation. The IANA language subtag registry carries deprecation as a dated field with an optional preferred-value pointing at a successor, while never removing the record. RFC 8126 separates unassigned from reserved. DID methods must define a deactivate operation. A consuming agent needs to know not only the current state but what it is now obliged to do.

Questions

1. What is the complete state vocabulary for an identifier under this scheme?state
   Expected answer

   • Enumerated state values with definitions
   • Which states are terminal
   • Default state on first assignment
2. Which state transitions are permitted, and who may trigger each one?lifecycle
   Expected answer

   • Permitted transition pairs
   • Triggering role for each
   • Evidence required for each transition
3. What must a consumer do when an identifier it holds becomes deprecated or retired?requirement
   Expected answer

   • Required consumer action
   • Successor identifier where one exists
   • Grace period before references are treated as stale
4. Does the scheme impose a renewal or revalidation duty, and what happens if it lapses?requirement
   Expected answer

   • Renewal interval
   • Lapse state and its effects
   • Reinstatement procedure
5. How does a consumer determine the current state of an identifier, and how fresh is that answer?access
   Expected answer

   • State query interface
   • Publication latency
   • Observation timestamp returned with the state

Artifacts

• Identifier state event logAppend-only log of state transitions for identifiers under the scheme, each entry carrying prior state, new state, trigger, actor, effective time and record time.

Non-reuse is a policy choice, not a law of naming, and this is where naive models fail. RFC 8141 asks formal-namespace registrants to demonstrate a commitment not to reassign, and BCP 47 guarantees that subtags are never removed or reassigned. GS1 by contrast operated timed reuse until an all-sector non-reallocation rule took effect after December 2018, and it still permits reuse twelve months after catalogue deletion for items that were never produced, while withdrawn-and-reintroduced items keep their original identifier. Supersession, splits and merges need explicit relation vocabulary of the kind DataCite provides.

Questions

1. Does this scheme permit a retired identifier to be bound to a different referent?constraint
   Expected answer

   • Reuse permission statement with its clause
   • Effective date of the current rule
   • Prior rule and whether values assigned under it are grandfathered
2. If reuse is permitted, what quarantine period and preconditions apply?constraint
   Expected answer

   • Quarantine duration
   • Preconditions such as catalogue deletion or never-produced status
   • Party that verifies the preconditions
3. When custody of the referent transfers, does the identifier travel with it?ownership
   Expected answer

   • Transfer rule
   • Record of the prior and new registrant
   • Effect on resolution targets
4. How are splits and merges of the referent expressed in identifier terms?relationship
   Expected answer

   • Relation vocabulary used
   • Whether a new identifier is minted for each part
   • Retention rule for the superseded identifier
5. How can a consumer detect that an identifier it holds now denotes a different referent?quality
   Expected answer

   • Detection signal available to consumers
   • Historical binding record availability
   • Recommended re-verification interval

Artifacts

• Reuse and reassignment policy statementPublished statement of whether reuse is permitted, the quarantine period, every exception and the effective date of each rule version.

Public identities and bibliographic identifiers are sometimes merged or split. SKOS exactMatch is transitive and maps concepts across schemes without asserting owl:sameAs. A resource may carry several URNs for similar or different purposes, which is aliasing rather than merge.

Questions

1. If two identifiers were merged, which value survives, which are aliases, and is the merge reversible?relationship
   Expected answer

   • survivingId
   • deprecatedIds
   • reversibleFlag
2. If one identifier was split, which new identifiers were minted for which new referents, and what is the parent identifier?event
   Expected answer

   • parentId
   • childIds
   • splitEventTime
3. Are multiple identifiers for one referent true aliases, purpose-specific identifiers, or embeddings of another scheme?classification
   Expected answer

   • aliasType
   • purposeCode
   • embeddedSchemeRef

Artifacts

• Merge or split recordAuthoritative event that rewired identifier-to-referent bindings.

Namespace exhaustion is a foreseeable failure with long lead times and is routinely omitted from scheme documentation. RFC 9562 makes the point in the probabilistic case: implementers must weigh the consequences of collision and choose between local and global uniqueness guarantees. ARK addresses the combinatorial case directly by recommending digit-heavy names with sparing use of a restricted alphabet to raise namespace density. Hierarchical schemes shift the problem to the delegation tree, where a parent can extend capacity by widening or deepening arcs.

Questions

1. What is the theoretical capacity of this namespace under its current grammar?measurement
   Expected answer

   • Capacity figure with the calculation basis
   • Constraining component of the grammar
   • Any capacity already withheld as reserved
2. What is the observed allocation rate, and what is the projected exhaustion date?measurement
   Expected answer

   • Allocation rate with its measurement window
   • Projected exhaustion date
   • Confidence interval or sensitivity of the projection
3. How can capacity be extended without invalidating existing identifiers?decision
   Expected answer

   • Extension mechanism such as longer values, new arcs or a new prefix range
   • Backward-compatibility impact
   • Lead time required
4. For probabilistically generated identifiers, what collision probability is accepted and on what basis?constraint
   Expected answer

   • Accepted collision probability
   • Consequence analysis for a collision
   • Detection and remediation procedure

Artifacts

• Capacity and exhaustion forecast reportPeriodic report of capacity, consumption, projected exhaustion and the extension options with their lead times.

The Handle System illustrates the general architecture: a global registry maps a naming authority to the local service responsible for it, clients cache that mapping, and resolution requests proceed without authentication unless confidential values are involved. RFC 8141 reserves a resolution parameter component but forbids its use before semantics are standardised, and requires a resolution statement in the registration. DID methods must each define a resolve operation. Coverage matters as much as the endpoint: a resolver that covers only part of a namespace produces silent failures indistinguishable from non-existence.

Questions

1. Is resolution intended for this scheme at all, and what did the registration state?definition
   Expected answer

   • Resolution intent statement as filed
   • Whether resolution is mandatory or optional for registrants
   • Clause reference
2. Which services resolve identifiers of this scheme, over which protocols?process
   Expected answer

   • Endpoint references
   • Protocols supported
   • Operator of each endpoint
3. What portion of the namespace does each resolver actually cover?quality
   Expected answer

   • Coverage expressed as namespace portions
   • Known gaps
   • Behaviour on an out-of-coverage request
4. Are resolution parameters or service requests supported, and with what standardised semantics?interoperability
   Expected answer

   • Supported parameter forms
   • Standardisation status of their semantics
   • Whether clients should pass fragment-like components to the resolver
5. What service level, caching guidance and rate limits govern resolution at volume?constraint
   Expected answer

   • Service level terms
   • Caching guidance
   • Rate limits and bulk access route

Artifacts

• Resolver service descriptorDescriptor of each resolution service with endpoint, protocol, coverage, operator and service level.

RFC 3986 states that a URI provides identification only and that access to the resource is neither guaranteed nor implied, and it discourages the strict name-versus-locator partition on which many identifier arguments rest. XML namespace names are explicitly not intended for retrieval. Yet FAIR expects data to be retrievable by identifier over a standardised protocol, and users routinely treat a failed resolution as evidence that the referent does not exist. This finding fixes the interpretation a consuming agent must apply so that resolution failure is not mistaken for non-existence, nor successful resolution for authenticity of the referent.

Questions

1. Does holding a well-formed identifier of this scheme guarantee any access to the referent?constraint
   Expected answer

   • Explicit access guarantee or disclaimer
   • Clause reference
   • Cases where resolution exists but access is restricted
2. What may a consumer conclude when resolution fails?exception
   Expected answer

   • Enumerated failure causes
   • Which causes imply non-existence and which do not
   • Recommended fallback for each
3. Is the scheme's own namespace expression intended to be dereferenceable?interoperability
   Expected answer

   • Dereference intent for the namespace token
   • What is returned if it is dereferenced
   • Whether dereference is required for correct interpretation
4. Does successful resolution establish that the returned resource is the intended referent?evidence
   Expected answer

   • Authenticity guarantee or disclaimer
   • Verification mechanism where one exists
   • Attack scenarios the resolver does not defend against

RFC 8141 r-components are reserved for passing parameters to resolution services and must not affect assigned-name equivalence. DID parameters service, relativeRef, versionId, versionTime and hl distinguish which resource is identified versus how resolution is performed.

Questions

1. Does this scheme define r-component or equivalent resolver parameters, or are they still undefined and therefore not to be used?constraint
   Expected answer

   • rComponentDefined
   • parameterRegistryRef
2. Which DID URL parameters are supported to select service, relative resource, version or hash integrity?interoperability
   Expected answer

   • supportedDidParameters
   • hashlinkSupport
3. Which inputs must be part of the identifier URL and which must be resolver options that are not part of identity?decision
   Expected answer

   • identityParameters
   • resolverOptions

ARK is unusually explicit that persistence is purely a matter of service and that only a current provider can say what commitment it intends; its policy service returns declarations addressing object availability, identifier validity, content invariance and change history. FAIR asks separately that metadata remain accessible even when the data are not. W3C web architecture asks owners to provide representations consistently and predictably. Together these give four separable promises that are routinely bundled into the single unfalsifiable word persistent.

Questions

1. Who makes the persistence commitment for a given identifier, and are they the current provider?ownership
   Expected answer

   • Committing party reference
   • Whether that party currently controls resolution
   • Date the commitment was made
2. Which of identifier validity, object availability, content invariance and change history are actually promised?requirement
   Expected answer

   • Promise held per dimension
   • Exclusions and caveats
   • Wording of the published declaration
3. Over what horizon does the commitment run, and what backs it?temporal
   Expected answer

   • Stated horizon
   • Funding or legal backing
   • Review interval for the commitment
4. What is returned once the referent is withdrawn, and is its metadata retained?exception
   Expected answer

   • Tombstone behaviour
   • Metadata retained after withdrawal
   • Whether the identifier remains resolvable
5. How is the commitment transferred if custody of the referent or the resolver changes?process
   Expected answer

   • Transfer procedure
   • Notification obligations
   • Evidence of prior transfers

Artifacts

• Persistence commitment statementPublished declaration by the current provider stating, per dimension, what is promised, for how long, with what exclusions and what happens on withdrawal or custody transfer.

A handle carries multiple typed and indexed values, each with its own administrative controls, so resolution returns a value set rather than a single location. ARK reserves an inflection that returns descriptive metadata together with the commitment statement. DID resolution returns a document plus resolution metadata. For an operating agent the outcome must be recorded with the identifier as supplied, the identifier as canonicalized, the resolver consulted, the observation time and the outcome class, because the same identifier resolved at two times is two facts.

Questions

1. What does a successful resolution return, and is it a single value or a typed set?definition
   Expected answer

   • Structure of the returned value set
   • Types and indexes present
   • Which value is the default target
2. How does a client request descriptive metadata or a policy statement rather than the referent?process
   Expected answer

   • Request mechanism such as an inflection, parameter or content negotiation
   • Metadata profile returned
   • Whether support is mandatory or optional
3. What outcome classes must a client distinguish, and how are they signalled?exception
   Expected answer

   • Outcome class vocabulary including success, gone, withdrawn, not-covered and unavailable
   • Signalling mechanism per class
   • Retry guidance
4. What must be recorded about a resolution so that the result can be reproduced or disputed later?provenance
   Expected answer

   • Fields recorded per resolution attempt
   • Observation timestamp in RFC 3339
   • Retention period for the record

Artifacts

• Resolution audit logAppend-only log of resolution attempts recording supplied and canonical identifier, resolver consulted, outcome class, returned value summary and observation time.

RFC 8141 contemplates authorised retrieval of representations while others receive only metadata, as in a national legal-deposit collection. Resolution may also be undefined for namespaces that identify protocol features rather than retrievable resources. Failure modes and caching must not equate non-retrieval with non-identity.

Questions

1. Who may obtain a representation versus metadata versus a mere existence check, and under which exception process?access
   Expected answer

   • representationAudience
   • metadataAudience
   • exceptionProcess
2. If the scheme is not intended as a locator, what does software that obtains the identifier do instead of retrieval?exception
   Expected answer

   • nonLocatorBehaviour
   • protocolSlotGuidance
3. How are not-found, gone, unauthorised and resolver-unavailable distinguished in logs without treating them as identifier invalidity?exception
   Expected answer

   • failureCodes
   • cachePolicy
   • invalidVersusUnresolved

Artifacts

• Resolution audit logAuditable record of resolution requests where policy requires it, especially bulk or authorised access.

A crosswalk is an assertion by a party, not a fact of the world. SKOS distinguishes exact, close, broad, narrow and related matches, declares exactMatch transitive and symmetric while deliberately withholding transitivity from closeMatch so that errors do not compound when mappings are chained across schemes, and makes exactMatch disjoint from broadMatch and relatedMatch. DataCite supplies a complementary vocabulary for identity and version relations between registered resources. Every assertion needs an asserter, evidence, a quality judgement and a retirement path.

Questions

1. What relation type does this mapping assert, and is it symmetric or transitive?relationship
   Expected answer

   • Relation type from a named vocabulary
   • Symmetry and transitivity properties
   • Disjointness constraints with other relations
2. Who asserted this mapping, on what evidence, and when?provenance
   Expected answer

   • Asserting party reference
   • Evidence basis such as authority publication, expert review or algorithmic match
   • Assertion time in RFC 3339
3. What is the quality of this mapping and how was it measured?measurement
   Expected answer

   • Quality grade or score
   • Measurement method and sample
   • Known systematic errors
4. May this mapping be chained with others to infer a further correspondence?constraint
   Expected answer

   • Chaining permission with maximum depth
   • Relation types that must not be chained
   • Error accumulation estimate
5. How is a mapping withdrawn when one side is deprecated or reassigned?lifecycle
   Expected answer

   • Withdrawal trigger and procedure
   • Whether withdrawn mappings are retained for history
   • Notification path to consumers
6. What cardinality does the mapping have, and how are one-to-many correspondences represented?composition
   Expected answer

   • Cardinality of the correspondence
   • Representation of splits and merges
   • Rule for selecting a preferred counterpart

Artifacts

• Crosswalk mapping setA published set of mapping assertions between two schemes, each carrying relation type, asserter, evidence, quality and assertion time.

This is the mixin surface. FHIR's Identifier shows the minimum that must travel with a value for it to remain interpretable: a system establishing the namespace, the value unique within it, a coded type, a use marker that modifies meaning, a validity period and an assigner. XML Namespaces warns that a prefix functions only as a local placeholder and that applications should use the namespace name, not the prefix, whenever a name's scope extends beyond its document. ISO 6523 code designators show the same job done with a numeric scheme code instead of a URI. A consuming model that stores a bare string has thrown away the interpretation.

Questions

1. What must accompany an identifier value for a consumer to interpret it without external context?interoperability
   Expected answer

   • Required accompanying fields
   • Canonical exchange representation
   • Minimum viable subset where the full shape is unavailable
2. Are abbreviations or prefixes permitted in carriage, and is their scope local only?constraint
   Expected answer

   • Prefix or abbreviation permission
   • Scope limitation statement
   • Expansion rule to the full namespace token
3. When a referent holds several identifiers, how is the preferred one designated?classification
   Expected answer

   • Use markers available
   • Preference selection rule
   • Handling of old and secondary identifiers
4. How is the period during which an identifier was valid for the referent recorded?temporal
   Expected answer

   • Period representation with RFC 3339 bounds
   • Treatment of open-ended periods
   • Interaction with the identifier state model
5. How is the assigning body recorded alongside the value?provenance
   Expected answer

   • Assigner reference form
   • Whether the assigner is derivable from the value structure
   • Fallback when the assigner is unknown

Artifacts

• Identifier carriage profileSpecification of how identifiers of this scheme are carried in consuming models, naming required fields, canonical exchange form, prefix rules and preference selection.

RFC 8141 allows NSS to encapsulate hierarchical names from non-URN systems. ISO 26324 describes integrating another scheme character string through DOI syntax or system metadata. FHIR says if the value is a full URI then system SHALL be urn:ietf:rfc:3986. Embedding is not the same as mapping.

Questions

1. Does this identifier embed another scheme native string, and what canonicalisation is applied before embedding?composition
   Expected answer

   • embeddedSchemeRef
   • nativeString
   • canonicalisationBeforeEmbed
2. Can the native identifier be recovered losslessly from the encapsulating form, and which form is canonical for matching?interoperability
   Expected answer

   • losslessRoundTrip
   • canonicalForMatching
3. Does the encapsulating scheme claim to replace the native scheme, contrary to ISO 26324 non-replacement rule?constraint
   Expected answer

   • replacementClaimFlag
   • coexistencePolicy

FHIR Identifier shows the operational pattern every consuming model needs: system namespace URI, value unique in that system, optional type, use, period and assigner. Comparison requires knowing the system. Values are case-sensitive unless the system says otherwise.

Questions

1. What absolute namespace URI is the system, and what value is unique within it?identity
   Expected answer

   • systemUri
   • value
2. Is a coded identifier type recorded for when the system is unknown, without treating type as a substitute for system?classification
   Expected answer

   • identifierType
   • typeVsSystemNote
3. What normalisation is allowed before matching, and is case-insensitive comparison actually safe for this system?validation
   Expected answer

   • normalisationSteps
   • caseInsensitiveSafeFlag

W3C web architecture names two failure modes directly: collision, where distinct resources share an identifier, and aliasing, where one resource acquires arbitrarily many. BCP 35 warns against squatting and requires a new scheme to show utility beyond what is already registered. RFC 8126 provides a marker for values known to be in use without registration. BCP 47 shows the hardest case: an upstream source standard withdraws or reassigns a code, and the registry must absorb the change through deprecation and preferred-value mappings rather than by removing the record. Carrying the scheme version alongside every stored value is what makes such drift detectable at all.

Questions

1. What counts as a collision under this scheme, and how is one detected?quality
   Expected answer

   • Collision definition
   • Detection method and cadence
   • Historical collision incidents
2. When two identifiers are found to denote the same referent, which survives and how is the other marked?decision
   Expected answer

   • Survivorship rule
   • Marking applied to the superseded identifier
   • Party empowered to decide
3. How is use of unregistered or squatted values discovered and handled?exception
   Expected answer

   • Discovery mechanism
   • Marker used for known unregistered use
   • Regularisation or enforcement route
4. When an upstream source standard changes a code, how does the scheme absorb the change?process
   Expected answer

   • Absorption procedure
   • Whether records are ever removed
   • Deprecation and preferred-value mechanics
5. What signal lets a consumer detect that stored values no longer match the scheme in force?quality
   Expected answer

   • Stored scheme version and comparison rule
   • Revalidation cadence
   • Escalation path on mismatch

Artifacts

• Collision and drift reportPeriodic report of detected collisions, aliases, unregistered use and divergence from the scheme version in force, with remediation status per item.

External standards are alignments until evidence is produced. A conformance claim must name the standard, the edition and the specific clause, and point at the test or audit that supports it. GLEIF's data quality programme shows that a registry's assertions about its own contents are measurable and reportable rather than declaratory. Mandatory metadata properties give a concrete completeness measure. Where no evidence exists, the honest record is an alignment claim, not conformance.

Questions

1. Which standard, edition and clause does this scheme claim conformance to, and what evidence supports the claim?evidence
   Expected answer

   • Standard, edition and clause
   • Evidence type such as test report, audit or self-declaration
   • Date and issuing party of the evidence
2. What defect rate is observed in the assignment register, and how is it measured?measurement
   Expected answer

   • Defect classes and rates
   • Measurement method and population
   • Trend over the last measured periods
3. Who audits the scheme's operation, how often, and what are the published findings?authority
   Expected answer

   • Auditing party
   • Audit cadence and scope
   • Most recent findings and their status
4. What is the remediation path when a defect or non-conformance is found?process
   Expected answer

   • Remediation procedure and owner
   • Target resolution times by severity
   • Escalation route
5. Which external standards are aligned with but not conformed to, and where do they diverge?interoperability
   Expected answer

   • Aligned standards list
   • Divergence points per standard
   • Reason each divergence is retained

Artifacts

• Conformance and quality evidence dossierCollected conformance claims, alignment statements with divergences, defect measurements and audit findings for the scheme.

GDPR Article 4(1) makes an identification number or online identifier a route to identifiability, Recital 26 keeps re-identifiable pseudonymised data within scope, Recital 30 names device and cookie identifiers explicitly, and Article 87 leaves national identification numbers to Member State law. RFC 8141 requires a filed analysis of comparison issues, information leakage and privacy. Time-based generated identifiers can disclose creation time and generator state. NIST separates a label from a credential: an identifier is never proof of anything, and treating it as an authenticator is a design fault, not a policy preference.

Questions

1. Are values of this scheme personal data, directly or through re-identification?classification
   Expected answer

   • Personal-data classification with reasoning
   • Re-identification pathway if any
   • Applicable legal regime and territory
2. What does a value disclose about its referent or its creation to a party holding only the string?access
   Expected answer

   • Disclosed attributes
   • Whether disclosure is intended
   • Mitigations available
3. Can the namespace be enumerated or harvested, and what controls limit that?access
   Expected answer

   • Enumerability assessment
   • Rate limiting and bulk access controls
   • Monitoring for harvesting
4. Is the value ever accepted as evidence of authorisation or identity, and how is that prevented?constraint
   Expected answer

   • Explicit prohibition statement
   • Known systems that misuse it as a secret
   • Compensating authentication requirement
5. How is an erasure or restriction request reconciled with a persistence commitment?exception
   Expected answer

   • Reconciliation procedure
   • What is redacted versus what is retained
   • Decision record and its approver
6. Does any national or sectoral restriction limit processing of this identifier?authority
   Expected answer

   • Restricting instrument and territory
   • Permitted purposes
   • Consequence of breach

Artifacts

• Identifier privacy classification recordRecord classifying the scheme's values for personal-data status, leakage, enumerability, credential misuse and applicable processing restrictions, with the reasoning and the assessing party.

A scheme's data has terms even when it is free: the LEI system is presented as a public good available free to all users, which is itself a licensing position with consequences for downstream reuse. Registration agencies impose obligations on registrants as a condition of assignment. RFC 8126 requires change control to be settled per registry and, where entries come from many sources, per entry. ARK's observation that only a current provider can state its intended commitment makes succession the central continuity question: if the operator stops, the commitment stops unless someone has already agreed to inherit it.

Questions

1. Under what legal instrument does the scheme operate, and in which jurisdictions is it effective?authority
   Expected answer

   • Instrument type and citation
   • Jurisdictions of effect
   • Any jurisdiction where it is not recognised
2. On what terms may the registry, assignment register and crosswalk data be used and redistributed?access
   Expected answer

   • Licence identifier and terms
   • Distinction between single-lookup and bulk access
   • Attribution and redistribution conditions
3. What ongoing obligations bind a registrant, and what are the fees?requirement
   Expected answer

   • Obligation list including metadata maintenance and renewal
   • Fee structure
   • Consequence of non-payment or non-compliance
4. Who inherits the scheme, its register and its resolution duty if the operator winds down?process
   Expected answer

   • Named successor or escrow arrangement
   • Trigger conditions
   • Whether the arrangement has been tested or exercised
5. Who holds change control over the scheme record and over individual entries?ownership
   Expected answer

   • Change controller at registry level
   • Change controller at entry level where these differ
   • Transfer procedure

Artifacts

• Continuity and licensing recordRecord of the scheme's legal instrument, data licence, registrant obligations, change control and succession arrangement, with evidence of any testing.