Research frontier master

Executive conclusion

The repository's most valuable next move is instrument falsification before theory expansion. The first funded REP should be RFR-2026-001, the negative non-isomorphic control benchmark. It attacks the highest stated validity risk, costs less than a human baseline or matched utility trial, and gates interpretation of ECR-000003, grammar, recognition, and replication evidence.

The repository is strongest in research-process architecture, traceability intent, and generation of testable hypotheses. It is weakest in independent human evidence, calibrated negative controls, causal utility, corpus reliability, and enforcement of its own profile integrity. No current evidence supports calling Framework Engineering a validated or distinct engineering discipline.

Scope and acceptance rule

This analysis inventories the repository as of 2026-07-28. It gives all 1246 publishable catalog records an explicit disposition, analyzes every 17 accepted/current/complete record, and includes 23 additional authoritative dependencies. The 40 analyzed artifacts are grouped into 11 evidence-coherent units rather than falsely treating templates, generated views, repeated provider outputs, or package components as independent findings. Drafts are included only when a current artifact depends on them; their status limits confidence. Superseded and archived work is used only for lineage or contradiction.

The coverage ledger, machine-readable coverage manifest, analysis-unit frontiers, and individual document frontiers make the scope auditable. The 16 semantically deduplicated RFRs are open candidates: they remain mutable until an authorized acceptance decision, after which successor/supersession rules apply.

Frontier score

Each candidate was challenged and rated on six explicit 1–5 dimensions:

raw = knowledge gain × potential impact × cross-project reuse × scientific importance − dependency cost − implementation difficulty

Every RFR records the six inputs and the reproduced calculation. Scores shown here are stabilized ordinal priorities, not measured utility or evidence strength.

Ranked opportunities

Critical contradictions

Dependency strategy

The foundational sequence is:

1. Calibrate false positives (RFR-2026-001).
2. Complete the controlling mechanism-boundary adjudication (RFR-2026-016).
3. Establish human reliability (RFR-2026-002) and corpus reliability (RFR-2026-006).
4. Test causal utility (RFR-2026-003).
5. Only then estimate ROI, field usability, and expansion value.

In parallel, profile integrity (RFR-2026-008) can proceed without changing accepted v1.0 content. Security and reproducibility work depend on that integrity baseline.

Repository-wide themes

• Largest knowledge gap: causal incremental utility over strong adjacent methods.
• Most urgent measurement gap: comparator specificity under negative non-isomorphic controls.
• Most influential uncertain research: ECR-000003 and the primitive/grammar evidence family.
• Over-researched relative to evidence: governance, taxonomy, and process documentation.
• Neglected disciplines: human factors, statistics, economics, accessibility, and security.
• Weak architectural area: the declared ROS–FE contract exceeds executable enforcement.
• Duplicate investigations: human baseline, independent replication, recognition control, grammar separation, and utility comparison recur under multiple names; each is merged here.
• Controlling unresolved gate: EX-FE-0002 remains experimentally inconclusive despite a newer proposed synthesis.

Cross-discipline analysis

Executive recommendations

• If only one REP is funded: RFR-2026-001. It can reveal that the primary comparator is non-specific and therefore prevents costly misinterpretation.
• Most uncertainty reduction: RFR-2026-003, once its prerequisites exist, because it directly tests whether FE adds causal value.
• Largest unlock: RFR-2026-001; it is prerequisite or methodological input to measurement, recognition, accessibility, and human-baseline work.
• Highest ROI: RFR-2026-008 for engineering risk reduction; RFR-2026-001 for scientific information per unit effort.
• Highest risk/highest reward: RFR-2026-005. A defensible grammar boundary could materially advance theory; failure could collapse a major theory branch.
• Resolve the controlling state: RFR-2026-016 after RFR-2026-001; it must precede the utility study unless formally superseded.
• Begin immediately: RFR-2026-001 and RFR-2026-008 in parallel. Neither requires assuming the theory is true.

Specification traceability

Self-critique

This analysis is repository-internal and did not independently reproduce experiments, interview researchers, or re-run every historical tool. The coverage ledger prevents silent omission, but evidence-coherent grouping can still hide a unique claim inside an excluded draft or derivative; such files retain an explicit disposition and can be promoted through a successor review. Frontier scores are transparent structured judgments, not empirical estimates. Document-level frontiers inherit some unit-level synthesis and therefore must not be mistaken for independent readings by different reviewers. Revalidation is required when ECR-000003 review closes, EX-FE-0002 completes or is superseded, the boundary decision is accepted/rejected, profile v1.1 lands, or any top-four RFR completes.