Mechanism
The lineage field is one of the seven typed fields that constitute a discovery object's persistent semantic state, alongside the intent, context, memory, policy reference, affective state, and confidence fields. It encodes the ordered sequence of transitions executed during a traversal of the adaptive index. For each transition, the lineage field records the anchor identifier at which the transition occurred, the timestamp of the transition, the semantic state mutation that was applied, the admissibility determination that permitted or refused the transition, and the semantic neighborhood from which the transition was selected. The lineage field is not a result summary appended after the traversal completes. It accumulates as the traversal advances, transition by transition, and it travels with the discovery object as part of the same state that the search, inference, and execution phases read and update at every anchor boundary.
Recording is performed by the execution step, the third phase of the three-in-one traversal step. The execution step writes the admissibility determination into the lineage field regardless of whether the proposed transition is admitted, rejected, or decomposed. A traversal lineage therefore preserves not only the transitions that were taken but the transitions that were evaluated and rejected, the reasons for rejection, and the decomposition paths that were explored. This is what permits a traversal result to be presented with a complete admissibility audit trail rather than the opaque provenance that characterizes conventional search and retrieval, where the path from query to result is discarded once the result is produced.
What a Lineage Entry Records
Each entry in the lineage field is structured, not free-form. It binds together the anchor identifier, the timestamp, the semantic state mutation applied at that step, the admissibility determination, and the semantic neighborhood from which the transition was selected. Because the entry references the neighborhood the anchor advertised at the time, the lineage captures the local rule context under which the decision was made, not merely the decision itself. The semantic state mutation field records how the discovery object's intent, memory, and confidence changed across the step, so that the evolution of the traversing entity is reconstructable from the lineage alone.
Several disclosed traversal events are recorded into the lineage as distinguishable entries. A capability-constrained non-admission, where a transition is refused because the discovery object's capability profile does not satisfy an anchor's advertised capability requirement, is recorded as such and is distinguished from a rejection due to a policy violation, a lineage discontinuity, or an entropy exceedance. A drift event, where the discovery object's current intent has diverged from the intent recorded at initialization beyond the policy-defined drift threshold, is recorded into the lineage rather than silently discarded. A collaborative merge, where two discovery objects with compatible intents exchange accumulated memory at a shared anchor, is recorded in the lineage fields of all participating objects, including the participating identities, the anchor at which the merge occurred, and the memory elements exchanged. An alias resolution that follows a redirect produced by anchor self-organization is recorded as a structural resolution event, distinct from a semantic mutation, so that structural rekeying does not introduce spurious entries that would distort trust evaluation.
Deterministic Verifiability
The admissibility evaluation at each traversal step is deterministic: given the same discovery object state, the same anchor configuration, and the same proposed transition, the evaluation produces the same outcome. This determinism is what gives the lineage field its evidentiary value. Any party with access to the discovery object's lineage field, the anchor's governance configuration at the time of traversal, and the proposed transition can independently verify that each admissibility determination was correct. The verification does not require participation in the original traversal and does not require re-running the inference engine, because the inference engine only proposes transitions while the execution substrate decides them against deterministic governance criteria.
This separation of proposal authority from commitment authority is the structural reason the lineage is auditable. The inference engine, whatever its architecture, produces a preference ordering over candidates. It has no authority to commit a transition. Authority resides in the execution substrate, which evaluates each proposed transition against the policy reference field, the anchor's governance configuration, and the traversal's accumulated lineage. The lineage thereby records decisions made by a deterministic evaluator, which is why a later verifier can reach the same admissibility judgment the traversal reached.
Bounded Recording Overhead
Because admissibility evaluation operates on typed fields, policy identifiers, entropy bounds, lineage hashes, and temporal validity windows, rather than on unstructured natural-language content or high-dimensional probability distributions, the evaluation is a constant-time or near-constant-time operation relative to the size of the semantic state object. The cost of evaluating and recording a step does not scale with the size of the index, the length of the traversal, or the complexity of the inference model. It scales only with the number of governance constraints in the discovery object's policy reference field and the current anchor's governance configuration. This bounded overhead is what makes per-step lineage recording practical even in traversals comprising many steps through a global-scale adaptive index.
Lineage as Presented Provenance
In human search mode, the lineage field is surfaced to the user as part of the result presentation. Each result is accompanied by the sequence of anchor transitions that led to its discovery, the semantic state of the discovery object at each step, and the admissibility determination at each step. This serves a function analogous to a chain of provenance in academic citation: it establishes the epistemic lineage of the result, enabling the user to evaluate the result's trustworthiness based on the quality and governance integrity of the traversal that produced it. The user can see not only what the system found but how it found it, why each intermediate step was taken, and why each transition was deemed admissible.
When a traversal terminates without resolution because the confidence field falls below the termination threshold, the lineage field is returned as part of the termination report. The report includes the traversal path recorded in the lineage, the confidence trajectory over the traversal steps, the anchor at which confidence collapsed, and the conditions that contributed to the collapse. The lineage thus documents failed traversals as faithfully as successful ones, which is a governed outcome in itself: the system states transparently that it could not find a reliable answer rather than presenting a speculative result.
Lineage as Index Evolution Signal
Beyond per-traversal audit, the aggregate of lineage records across many traversals serves as a behavioral signal that drives anchor self-organization. Each completed traversal produces a lineage record encoding which anchor transitions were admitted, which were rejected, what semantic mutations occurred at each step, how much semantic progress was achieved between successive anchors, and whether the traversal reached a resolution state or was abandoned. Read in aggregate, these records describe how effectively the index's current organization serves the discovery objects that traverse it.
The signals are interpreted by the self-organization mechanism. Anchors that frequently reject traversals, indicating that many discovery objects arrive with intents the anchor's neighborhood cannot serve, receive a reorganization signal: the published neighborhood may be misleading or the anchor may be positioned at an inappropriate location in the hierarchy. Anchors whose neighborhoods are frequently bypassed, where discovery objects pass through without selecting any advertised transition, receive a dissolution or merge signal, indicating the neighborhood may be redundant or insufficiently differentiated from adjacent neighborhoods. Anchors through which high-value traversals, those with high resolution rates and high semantic progress per step, frequently pass receive a reinforcement signal indicating the organization is effective and should be preserved. These signals drive the mechanism's decisions about anchor splitting, merging, rebalancing, and neighborhood re-publication, creating a feedback loop between traversal behavior and index structure that lets the index evolve in response to how it is used without centralized index management. The self-organization operations themselves preserve the lineage records and governance history of the objects they reorganize.
Prior-Art Distinctions
Conventional graph traversal systems maintain a visited-node set for cycle detection or result construction, but that set is transient operational state, discarded when the traversal completes. The lineage field persists as part of the discovery object's semantic state and is the substrate against which admissibility is later verified. Conventional search and retrieval systems present results whose provenance is opaque: the path from query to result is not retained, so there is no record of why a result was reached or which alternatives were rejected. The lineage field retains every admitted, rejected, and decomposed transition together with the reason for each determination.
Link-count relevance models such as PageRank score documents by structural popularity and do not encode whether a result satisfies the querier's policy constraints, lineage requirements, temporal validity, or trust evaluation, so governance must be bolted on as a separate post-ranking layer. Here governance is a constituent phase of every traversal step, and the lineage is the record of that governance: a semantic object reached through a governed traversal is, by construction, policy-compliant, lineage-verified, entropy-bounded, and temporally valid, with the lineage field documenting that each of these held at each step. The traversal path itself is the governance record, and the relevance determination and the governance determination are the same computation.
Disclosure Scope
The lineage field of the discovery object, encoding for each traversal transition the anchor identifier, timestamp, semantic state mutation, admissibility determination, and source semantic neighborhood; its recording by the execution step for admitted, rejected, and decomposed transitions alike; its role as a deterministically verifiable audit trail and as presented provenance in human search and termination reporting; and the use of aggregated lineage records as the behavioral signal that drives anchor self-organization, are disclosed in the cognition filing (U.S. Application No. 19/647,395 and its international counterpart), principally at Sections 10.2, 10.3, and 10.20, with the self-organization signals at Section 10.6. This article describes that disclosed mechanism. The scope extends to embodiments in which lineage entries are realized over different anchor, neighborhood, and semantic-state representations, provided each transition is recorded with its anchor context and admissibility determination and the recorded lineage remains independently verifiable against the governance configuration in force at the time of the transition.
