C.22.1:12 - SoTA-Echoing
Claim 1. Current frontier adaptation work judges usable specialization by threshold-crossing under bounded resources, not by terminal score alone.
Practice source, local alignment, and adoption decision. Current QD and agentic-adaptation sources such as A survey on Quality-Diversity optimization: Approaches, applications, and challenges, Swarm and Evolutionary Computation 100:102240 (2026), FactorMiner arXiv:2602.14670v1 (2026-02-16), and SkillOpt arXiv:2605.23904v2 (2026-05-25) repeatedly separate threshold target, budget burn, transfer evidence, reuse evidence, and changed object/version from one final benchmark score. This pattern adopts that practical field set, adapts it through one TaskFamilyRef or TaskSignature-bound adaptation signature, and rejects generic got better narratives that leave threshold and budget semantics implicit.
Claim 2. Current open-ended exploration work treats corridor entry and stepping stones as evidence-bearing novelty signals rather than decorative commentary.
Practice source, local alignment, and adoption decision. Current QD/OEE sources and current FPF C.17, C.18, C.19, G.5, G.9, and G.11 neighbours distinguish real corridor entry from one exotic sample by asking for explicit baseline, stable descriptor shift, reproducible solution class, or an explicit stepping-stone trace. This pattern adopts explicit corridor baseline/evidence discipline, adapts it as declared adaptation-signature fields, and rejects novelty talk that names no baseline, evidence source, or evidence locus.
Claim 3. Current selector and parity practice needs one stable shared field set for specialization claims.
Practice source, local alignment, and adoption decision. Current FPF selector and parity neighbours keep compared candidates reviewable only when candidates reuse the same published field set for threshold, prior exposure, transfer, retention, downside, and corridor-entry field. This pattern adopts that reuse discipline, adapts it by publishing one stable adaptation-signature field set here, and rejects silent downstream field redefinition in G.5 or G.9.
Evidence-source note. Peer-reviewed or archived frontier anchors carry the most direct evidence for threshold, budget, and parity claims. Fast-moving frontier lines remain explicit evidence for corridor-entry and open-ended exploration pressure only when the row names their local contribution; they are not a flattened single evidence status.
| Source-bound anchor family | Source-use relation/currentness | What it disciplines in this pattern |
|---|---|---|
QD / OEE corridor-entry work | Current QD overview plus current FPF OEE/NQD neighbours. | Corridor baseline, descriptor shift, stepping-stone evidence, and whether novelty is reproducible rather than one exotic sample. |
| Agentic adaptation benchmarks | Current narrow source lines such as FactorMiner and SkillOpt when the task family is comparable. | Threshold target, time-to-threshold, budget-to-threshold, prior exposure, and post-threshold efficiency under a declared task-family anchor. |
| Transfer / retention evaluation | Source-use relation/currentness supplied by the applying benchmark or neighbour pattern. | Transfer target, retention window, downside, and reuse evidence so specialization speed is not confused with one isolated threshold crossing. |