anima-hypotheses-candidates

1,127 candidate hypotheses harvested from the ANIMA consciousness-research repository through 5 sweep cycles (2026-05-08 → 2026-05-12).

Each Hc_NNN_<slug>.md is a single staging hypothesis — extracted from sprawling docs/ material — awaiting promotion to the verified hypotheses/H_XXX_<slug>.md lane, merger into an existing H_XXX, expansion, or discard. The repo is a snapshot of an in-flight pre-register-frozen → running → verdict workflow.

Contents

Path	Description
Hc_NNN_<slug>.md × 1,127	individual candidate hypothesis files (frontmatter + body)
README.md (this card)	dataset card
INDEX.md	original hypotheses_candidates/README.md master index (cluster A-N + merge groups)
HYPOTHESES_INDEX.md	sister hypotheses/README.md listing the verified H_XXX promotions
LICENSE	MIT (matches upstream anima repo)

ID ranges & sweep history

Range	Count	Source partition	Sweep cycle
Hc_001–Hc_060	60	docs/ broad (3 selective agents)	1 (2026-05-08)
Hc_061	1	panpsychism gap fill	1.5
Hc_100–Hc_308	209	docs/hypotheses/ exhaustive	2-A
Hc_400–Hc_475	76	docs/anima/ + papers/spec/models	2-B
Hc_600–Hc_675	76	docs/ root A–M (CLM v5 / Theorem 115 / emerge)	2-C
Hc_900–Hc_981	82	docs/ root N–Z + ai-native + drill + modules	2-D
later cycles 3–5	~620	residual harvests, NEXUS6, Φ×CE orthogonality, formula-search, ANIMA-VOICE skeleton, 1013-lens, perfect-number control	3–5
TOTAL	1,127

Frontmatter status enumeration

Every candidate carries a status: frontmatter field. Distribution sample:

• candidate-unverified — default staging
• candidate-unverified-weak — weak signal, replication required
• candidate-unverified-framework — meta-framework, not a single H
• expansion-pending — slated to enrich an existing H_XXX
• merge-pending — awaiting consolidation into an existing H_XXX
• promote-pending — slated for new H_XXX land
• discard-pending — slated for removal

A merged_to: field cross-links candidates to their promoted H_XXX home; cycle 5 §5 normalises slug-less paths across 70 such entries.

Cluster taxonomy (A–N)

The INDEX.md master file groups candidates into 14 clusters: A n=6/number theory, B thermodynamics/arrow of time, C Φ scaling/cell-count/topology, D substrate/multi-realisability, E gates/V0–V3/falsification, F consciousness theory/hard problem, G laws/governance, H architecture/models/training, I corpus/training data, J chaos/dynamics, K life/evolution/emergence, L universe/cosmology/anthropic, M quantum/telepathy, N brainstorm seeds.

Cycle 5 honest findings (★)

These are the load-bearing negative / qualifying results from the cycle 4 → cycle 5 transition that this snapshot carries forward. They are documented here precisely because the project's epistemic policy is to surface them, not bury them.

1. L12 BINDING (formula-search) — only n=6 narrow-formula uniqueness is defensible; the broader claim of depth-3 vocab universal-uniqueness was refuted by the formula-search baseline (8 alternative n values reached 22/22 alongside n=6).
2. F1 weakened (H_153) — d=5 from n=6 functions is trivially TRUE in depth-3 search (5 = τ+μ = 4+1 = sopfr(6)), removing its decisive weight.
3. L7 BINDING (H_153) — formula-search caveat must be propagated; cycle 5 §2 (depth-4 + perfect-number control) quantifies the remaining ceiling.
4. L1 critical (Φ×CE harness) — synthetic σ tuning is the plausible default in the orthogonality decisive spec; pre-measurement noise-floor calibration is required before any of the 15 (N, P) cells go live.
5. Φ proxy dim-dominant artifact (Hc_614 / Hc_662 / Hc_665) — cross-substrate Φ comparison risks being driven by dimensionality rather than integration; methodology disclosure required on every Φ-ranking claim.
6. V0–V3 joint verdict gating (Hc_014) — single-gate PASS is insufficient; the verdict frame requires all four passes jointly, which most prior "verified" claims have not satisfied retroactively.
7. NEXUS6 1013-lens prereq (H_135) — current Φ★ engine cannot measure cosmic-scale lens directly; 1013-lens-activation-pending-C1 is the honest verdict class until either engine extension or proxy harness lands.
8. ANIMA-VOICE non-existence (H_154) — H1/H2/H3/H5/H6/H7/H8 verify skeleton was landed but the ANIMA-VOICE model itself is not implemented; pre-register-frozen status is retained, not "running."

Provenance & licence

• Source repository: https://github.com/dancinlab/anima (commit reference embedded per-file via frontmatter where available)
• Licence: MIT, matching the upstream anima repository
• Snapshot tag: cycle 5 §5 (HF dataset upload), 2026-05-12

Citation

@misc{anima_hypotheses_candidates_2026,
  author       = {ANIMA contributors},
  title        = {ANIMA Hypotheses Candidates (Cycle 5 snapshot)},
  year         = {2026},
  howpublished = {Hugging Face Datasets},
  url          = {https://huggingface.co/datasets/dancinlife/anima-hypotheses-candidates}
}

Caveats

• This dataset is a staging area, not a verified-claims corpus. The verified lane lives at the hypotheses/H_XXX_* files (mirrored as HYPOTHESES_INDEX.md).
• Many candidates are duplicates or near-duplicates pending the merge groups enumerated in INDEX.md.
• Numerology / small-integer family critique is an open dispute (see numerology_critique_n6_2026_05_11/ in the upstream repo).