C.18:10 - SoTA-Echoing
Source-currentness boundary. This source-use basis was reviewed through 2026-08-26. Mutable preprints are pinned below to the edition actually used; the Qin et al. survey is pinned to its DOI-fixed journal article. Reopen the affected source-use row through G.11 when a cited revision changes the archive, front, generation, or evaluation claim used here; when a newer field survey materially changes the known QD/OEE boundary; or when contrary evidence changes what can be claimed about bounded archives, descriptor generalization, or open-ended evaluation. C.18 records that trigger and the affected row but does not itself perform refresh.
| Source or source family | Adopted FPF move | Rejected overread | Field or boundary changed |
|---|---|---|---|
Lin et al., Quality-Diversity Optimization as Multi-Objective Optimization, arXiv:2602.00478v1 (2026-01-31). | Treat QD and Q-front work through declared Q components, DominanceSet, comparator refs, archive relation, front relation, G.5 selected-set result declaration, separate audience publication, and refresh. | Cell-filling or popularity accounts are the current ontology by default. | FrontRecord@Context must keep dominance grounds, comparator refs, and Q-component refs explicit. |
Qin et al., A survey on Quality-Diversity optimization: Approaches, applications, and challenges, Swarm and Evolutionary Computation 100:102240 (2026), DOI 10.1016/j.swevo.2025.102240, https://www.sciencedirect.com/science/article/pii/S2210650225003979. | Use current survey support for approaches, applications, archive use, diversity use, and challenge framing. | Survey taxonomy replaces FPF relation definitions. | Use C.18 to state and test the ExplorationArchiveRecord@Context, FrontRecord@Context, and OpenEndedVariantGenerationRecord@Project; use G.5 for selected-set result declaration, E.17 for a source-backed publication face and return to source, E.24.PUB for the publication occurrence and audience availability, and G.11 for refresh. |
Batra et al., Quality Diversity for Robot Learning: Limitations and Future Directions, arXiv:2407.17515v1 (2024-07-09). | State retained exploration value, generalization pressure, and limitations when an archive is used beyond current dominance. | Bounded archives or cell occupancy are enough evidence that NQD and OEE are useful. | retainedExplorationValue, retentionPolicyRef, telemetryRefs, and nextGoverningRelation must be filled when the archive is relied on. |
Zhang et al., Darwin Godel Machine, arXiv:2505.22954v3 (2026-03-12). | Keep generated agents, archive lineage, empirically validated changes, method-family use, evaluation, and refresh separate. | OEE is one winner-selection method or source-free self-improvement story. | OpenEndedVariantGenerationRecord@Project records generation and archive or front linkage, while evaluation and refresh move to their subject patterns. |
Novikov et al., AlphaEvolve, arXiv:2506.13131v1 (2025-06-16). | Separate generated method text, method description, evaluator relation, selected set, source-use relation, performed work, and work result. | Generated algorithm text is proof, gate permission, accepted method selection, or performed work. | Use evaluatorOrComparatorRef, lineage, source refs, and nextGoverningRelation to determine whether to use C.18, A.19, G.5, C.11, A.15, or G.11. |
| Cultural-evolution and style-engineering source pressure from the music and dance intake. | Keep generated style or tradition variants as archive or front records until a cultural-evolution case or term bridge is current. | A cultural-style variant is a root cultural kind or a selected set by label. | culturalVariantRefs continue to C.36, F.17, F.18, or F.9; selected-set result declaration continues to G.5, with a stable public identity added only through its conditional UTS branch. |
| Architecture-search and product-family work. | Treat retained structures as candidates for an architecture move only after the architecture claim is named. | An archive of layouts is the architecture or the architecture decision. | Architecture candidates require C.30, C.30.ASV, C.30.AD, or C.32.P2S after C.18 records descriptor, archive or front relation, and telemetry. |
| Taylor, Evolutionary Innovations and Where to Find Them, 2019, read with Di Bona et al., higher-order novelties, 2025, and Kalambokidis et al., diversity and open-ended evolution, 2026. | Distinguish exploratory, expansive, and transformational claims; keep generator, evaluator, building blocks, retention/reproduction, environment, and opportunity structure explicit. | A taxonomy or high novelty/diversity result proves that a transferred non-evolutionary case changed its possibility space. | C.18:4.3a requires an exact earlier/candidate mapping and evidence limit; causal production uses C.28. |