FE-012C Capability Role Drift Review

Purpose

This review package supports a focused human assessment of whether FE-012C repeatability drift reflects genuine reasoning-structure change or primitive-level variation around stable capability roles.

This is not a new extraction experiment.

This is not a new theory document.

This is a structured manual review of selected disagreement cases.

Review Scope

The review targets five selected packet/provider cases drawn from the current FE-012C literal and semantic comparison outputs.

The aim is to determine whether comparison should remain at the primitive level or move to a higher abstraction level called Capability Roles.

Datasets

Prior dataset:

• research/experiments/FE-012C-manual-replication/responses/

Repeat dataset:

• research/experiments/FE-012C-repeatability-run-001/responses/

Review Files

• role-mapping.md
• selected-cases.md
• manual-review-template.md
• review-results.md
• comparator-implications.md
• decision-log.md

How To Perform The Review

1. Open selected-cases.md and choose a case.
2. Open the referenced prior and repeat response files.
3. Copy the case into review-results.md or use manual-review-template.md.
4. Convert each primitive sequence into a capability-role sequence using role-mapping.md.
5. Assess whether the capability-role backbone changed materially.
6. Record exactly one classification: stable, elaboration_drift, role_drift, or unclear.
7. Record the comparator implication without assuming that automation is justified.

How To Use The Role Mapping

The role mapping is provisional.

It is intended only to test whether some primitive-level disagreements collapse into stable capability-level agreement.

Reviewers should use the mapping consistently, but should also record when the mapping itself appears ambiguous or weak.

How To Complete Review Results

review-results.md is prefilled only with:

• packet id
• provider
• prior response path
• repeat response path
• primitive sequences

The reviewer must complete:

• capability-role sequences
• backbone review
• classification
• reasoning
• comparator implication
• notes

Decision Relevance

This review determines whether ECP-FE-000033 should be created.

If most reviewed cases classify as stable or elaboration_drift, an automated capability-role comparator may be justified for follow-on work.

If most reviewed cases classify as role_drift or unclear, automation should not be recommended.