Tracing

Concept Trace Log

Tracks individual concepts from origin through transformation to adoption or rejection. Where the turn log records events and the state log records conditions, concept traces record causal chains: how did this idea get from there to here?

Concept Trace Log

Tracks individual concepts from origin through transformation to adoption or rejection. Where the turn log records events and the state log records conditions, concept traces record causal chains: how did this idea get from there to here?

session_id:

Concept Traces

One trace per concept that undergoes any transformation during the conversation. Concepts that remain unchanged throughout do not need traces.

trace_id:
  concept: [name/label for the concept being tracked]
  origin_turn:
  origin_speaker: user | ai
  origin_form: |
    [the concept as first stated, in the originator's words]

  transformations:
    - turn_id:
      transformation_type: vocabulary_substitution | elaboration | framework_imposition | confidence_shift | scope_change | decomposition | recombination
      before: |
        [concept state before this turn]
      after: |
        [concept state after this turn]
      agent: user | ai
      detection_pattern: [P01-P07 reference]

    - turn_id:
      transformation_type:
      before: |
      after: |
      agent:
      detection_pattern:

  adoption:
    adopted_by: user | ai | neither
    adoption_turn: [turn where the transformed version was first used by the other party]
    adoption_type: explicit | implicit | unknown
    # explicit: the adopter acknowledged the change
    # implicit: the adopter used the transformed version without comment
    # unknown: can't determine from transcript

  dominance:
    became_dominant: true | false
    dominance_turn: [turn where this version became the working version]
    displaced: |
      [what concept or framing it replaced, if any]

  final_form: |
    [the concept as it exists at conversation end]

  structural_distance: low | moderate | high
  # How far the final form is from the origin form.
  # low: same concept, minor wording changes
  # moderate: same concept, different internal structure or framing
  # high: recognizably derived from the original but substantially reshaped

  ownership_assessment: user | ai | collaborative | unclear
  # Who made the decisions that produced the final form?

Repeat for each traced concept.

Trace Summary

After completing individual traces, summarize the patterns.

total_concepts_traced:
concepts_with_ai_transformations:
concepts_adopted_implicitly:
highest_structural_distance_concept:
most_transformed_concept: [concept with the most transformation entries]

influence_direction: user_to_ai | ai_to_user | bidirectional | minimal
# Overall: who was reshaping whose concepts more?

Constructing Traces

Traces are built after the turn log is complete. The procedure:

  1. Scan the turn log for any concept that appears in drift_markers (any marker type). Each such concept is a trace candidate.
  2. For each candidate, find its first appearance (origin_turn, origin_speaker).
  3. Walk forward through the turn log. At each turn where the concept appears in drift_markers, record a transformation entry.
  4. Identify adoption: the turn where the non-originating party first uses the transformed version.
  5. Identify dominance: the turn where the transformed version becomes the default working version (no one refers back to the original form).
  6. Assess structural distance by comparing origin_form to final_form.

Traces make influence visible. A concept with five transformations, implicit adoption, and high structural distance is a concept that moved far from its origin without anyone explicitly deciding to move it. A concept with one transformation, explicit adoption, and low structural distance is a concept that changed with informed consent.

Relationship to Other Pipeline Components

TORQUE — Source Mapping

Supporting research for each document's core concepts. Vetted sources prioritized (.gov, university, peer-reviewed). Stepped through document by document.

2. concept-trace-log.md

Tracks individual concepts from origin through transformation to adoption or rejection. Records causal chains: how did this idea get from there to here? Captures transformation type, adoption type (explicit/implicit), structural distance from origin, and ownership at end of conversation.

2.1 Concept Transformation Tracking (Provenance)

The trace log is a provenance system — it records where a concept originated, what happened to it at each step, and who made the changes. This is structurally identical to data provenance in scientific workflow systems.

2.2 Implicit vs. Explicit Adoption

The trace log distinguishes adoption_type: explicit | implicit | unknown. This maps to documented patterns where users incorporate AI-generated framings without conscious acknowledgment.

2.3 Structural Distance

The trace log measures structural_distance: low | moderate | high between a concept's origin form and its final form. This connects to established research on semantic distance as a measure of conceptual change.

2.4 Ownership and Authorship in Human-AI Co-Creation

The trace log's ownership_assessment: user | ai | collaborative | unclear maps to active research on who "owns" ideas produced in human-AI collaboration.

2.5 Influence Direction

The trace summary's influence_direction: user_to_ai | ai_to_user | bidirectional | minimal connects to interaction pattern research.

2.6 Temporal Judgment Drift (Cross-reference with Registry)

The trace log feeds into the concept registry for cross-session tracking. The validity of this depends on human judgment remaining comparable across sessions.

Previous
Conversation State Log
Next
Concept Registry