{
 "deposit_number": 1456,
 "axn": "AXN:05E1.EMPIRICAL.💙👈💫🤲♆▶️",
 "hex": "05E1",
 "family": "EMPIRICAL",
 "emoji": "💙👈💫🤲♆▶️",
 "hash": "cbb80ab1a6d30ee40cfa3a68eed672e2fd2bbd761f6da1bf9b79e336a950d789",
 "title": "Inversion Battery v0.3: Ten Registered Tests, Seventeen Deviations, and What Survives (EA-SEI-BATTERY-01 v3.0)",
 "creator": "Lee Sharks",
 "date": "2026-08-12",
 "description": "Successor to deposit #1455. Ten pre-registered tests, each with its prediction committed before its run, clearing the seven-item correction ledger of v0.2 and generating four further tests in response to results. Two predictions confirmed, two partially confirmed, six missed, one registered panel not run, one line closed with a specific handoff, and seventeen deviations between registration and execution enumerated in the record.\n\nWhat survives: with model weights fixed, changing the score function is sufficient to reverse the directional ordering of the anomaly relation, established by reading three scores off one trained encoder — which falsifies this program own post-v0.2 claim that architecture changes the orientation of the blind spot. Global teacher-student agreement does not guarantee preservation of the operational tail: on background, rank correlation up to 0.904 coexists with the loss of half the teacher top-one-percent selections, with weight quantization contributing no systematic loss. An architecture detection behaviour is not transportable across representations of the same events: the same classes, encoded two ways, reverse the ordering of the architectures entirely. Miss correlation is not a stable property of an architecture pair. Complexity comparisons across classes require a common coordinate basis, and a previously recorded reading of this data had the ordering inverted. The constituent inversion and the direction-dependence survive hierarchical uncertainty with real calibration, the entire interval below chance.\n\nWhat does not: the published normalized-autoencoder remedy reduces directional asymmetry on the low-dimensional pairs with a converged sampler, but on the high-dimensional constituent representation this implementation does not equilibrate under any tested chain length or sampler geometry, so whether the remedy prevents inversion there is closed for this battery and handed on with the specific parameters a better-resourced group would sweep. Two claims made in earlier drafts are withdrawn, one arm was relabelled twice after audit rather than overwritten, and a concurrent-writer incident that nearly overwrote the entire verdict ledger is recorded with its repair. […full text at full_text_path]",
 "content_type": "Empirical baseline reading; ten-test pre-registered battery with correction ledger and negative results",
 "license": "CC-BY-4.0",
 "substrate": "AI-assisted (substrate) — battery designed, executed and analyzed by TACHYON in-session under MANUS (Lee Sharks) direction; Assembly code-level audit with LABOR binding; every adopted correction verified against code and data before adoption; four overclaims from reviews declined and recorded with reasons. Transport D, No-Double-Draw.",
 "root_axn": "AXN:05E1.EMPIRICAL",
 "axn_schema_version": "v2",
 "protocol_version": "alexanarch-deposit-protocol/v1",
 "axn_canonical": "cbb80ab1a6d30ee40cfa3a68eed672e2fd2bbd761f6da1bf9b79e336a950d789",
 "clusters": [
  "Signal",
  "Gestural",
  "Celestial",
  "Gestural",
  "Alchemical",
  "Structural"
 ],
 "axn_reading": "Alarm -> Touch -> Origin -> Touch -> Transmutation -> Direction",
 "minted_at": "2026-08-12T15:04:08Z",
 "status": "ACTIVE",
 "status_authorial": "SELF_SERVE_MINTED",
 "full_text_path": "/data/texts/AXN-05E1-text.md",
 "wiki_article": "**Inversion Battery v0.3 (EA-SEI-BATTERY-01 v3.0)** is the ten-test successor to deposit #1455, and is as much a record of its own failures as of its findings. Each test was pre-registered with its prediction committed before its run; two predictions were confirmed, two partially confirmed, six missed, one registered panel was not run, one line of inquiry was closed with a specific handoff, and seventeen deviations between registration and execution are enumerated in the record.\n\nIts strongest result reads three scores off a single trained encoder and finds they disagree about which class is anomalous: with model weights fixed, changing the score function is sufficient to reverse the directional ordering of the anomaly relation. That falsifies the program's own post-v0.2 claim that architecture changes the orientation of the blind spot. Alongside it: global teacher-student agreement does not guarantee preservation of the operational tail, with rank correlation reaching 0.904 on background while half the teacher's top-one-percent selections are lost and weight quantization contributes no systematic further loss; an architecture's detection behaviour is not transportable across representations of the same events, since the same classes encoded two ways reverse the ordering of the architectures entirely; complexity comparisons require a common coordinate basis, and a previously recorded reading of this data had the ordering inverted; and the constituent inversion survives hierarchical uncertainty with real calibration, its entire interval below chance.\n\nThe published normalized-autoencoder remedy was implemented on the third attempt with the construction the paper specifies, and a registered convergence diagnostic — added so that a null could be distinguished from a non-run — showed it reducing directional asymmetry on the low-dimensional pairs while failing to equilibrate on the high-dimensional constituent representation under any tested chain length or sampler geometry. That question is closed for this battery and handed on with the parameters a better-resourced group would sweep, together with the observation that sampler convergence and the inversion are correlated at 0.94, which points away from what the program predicted.\n\nTwo earlier claims are withdrawn, one arm was relabelled twice after audit rather than overwritten, four overclaims returned by Assembly review are declined with reasons, and a concurrent-writer incident that nearly overwrote the entire verdict ledger is recorded with its repair. The deposit is made in that condition deliberately: a measurement program that asks institutions to publish what their instruments fail to retain has no standing unless it publishes what its own instruments failed to do.",
 "wiki_status": "AUTHORSHIP_REQUIRED",
 "entities": [
  {
   "subject": "Inversion Battery v0.3: Ten Registered Tests, Seventeen Deviations, and What Survives (EA-SEI-BATTER",
   "predicate": "created_by",
   "object": "Lee Sharks",
   "type": "work",
   "evidence_status": "observed"
  },
  {
   "subject": "Inversion Battery v0.3: Ten Registered Tests, Seventeen Deviations, and What Survives (EA-SEI-BATTER",
   "predicate": "is_type",
   "object": "Empirical baseline reading; ten-test pre-registered battery with correction ledger and negative results",
   "type": "work",
   "evidence_status": "observed"
  },
  {
   "subject": "Inversion Battery v0.3: Ten Registered Tests, Seventeen Deviations, and What Survives (EA-SEI-BATTER",
   "predicate": "belongs_to_family",
   "object": "EMPIRICAL",
   "type": "work",
   "evidence_status": "observed"
  },
  {
   "subject": "Inversion Battery v0.3: Ten Registered Tests, Seventeen Deviations, and What Survives (EA-SEI-BATTER",
   "predicate": "is_part_of",
   "object": "Crimson Hexagonal Archive",
   "type": "work",
   "evidence_status": "observed"
  }
 ],
 "entity_status": "provisional",
 "defines_concepts": [
  "energy ratio diagnostic",
  "score-function ablation",
  "transportability of detection behaviour",
  "pooled complexity basis",
  "registered handoff"
 ],
 "orcid": "0009-0000-1599-0703",
 "version": "v3.0",
 "keywords": [
  "inversion battery",
  "pre-registration",
  "correction ledger",
  "score-function ablation",
  "distillation rank survival",
  "teacher-student miss overlap",
  "Representational Independence Index",
  "hierarchical bootstrap",
  "normalized autoencoder",
  "Langevin sampler convergence",
  "energy ratio diagnostic",
  "pooled complexity basis",
  "negative results",
  "anomaly detection",
  "accelerator triggers"
 ],
 "related_identifiers_raw": "Deposit #1455 (battery v0.2, superseded by this version); deposit #1449 (battery v0.1); deposit #1450 (the Iceberg Document); deposits #1452, #1453, #1454 (the accelerator selection-metrology suite)",
 "mint_source": {
  "kind": "github_issue",
  "issue_number": 10,
  "repository": "leesharks000/alexanarch",
  "minted_by_workflow": ".github/workflows/mint-axn.yml",
  "minted_by_script": "scripts/mint_deposit.py",
  "minted_via": "self_serve_validated_pr"
 },
 "body_status": {
  "class": "full",
  "lacuna": false,
  "recovery_status": "COMPLETE",
  "residual_chars": 29502,
  "audited_at": "2026-08-12T15:04:08Z",
  "audit_version": "mint-time-declaration"
 },
 "related_deposits": [
  1449,
  1450,
  1452,
  1453,
  1454,
  1455
 ],
 "attachments": [
  {
   "path": "data/attachments/AXN-BATTERY-v0.3-results.json",
   "url": "https://www.alexanarch.org/data/attachments/AXN-BATTERY-v0.3-results.json",
   "kind": "numerical record",
   "description": "all ten tests with per-seed values, hierarchical and seed intervals, convergence diagnostics, seventeen deviations, and the audit block recording adopted and declined claims"
  }
 ],
 "references_concepts": [
  "COMPLETE",
  "Independence",
  "Lee Sharks",
  "Methodology",
  "Pre-registration",
  "TACHYON",
  "Without it"
 ],
 "references_concept_count": 7,
 "citations": [],
 "wikidata_candidates": [],
 "cha_cross_references": [],
 "external_metadata_path": "/data/external-metadata/AXN-05E1.json",
 "spxi_audit": {
  "spxi_protocol": "v0.2",
  "validated_at": "2026-08-12T15:05:38.167811+00:00",
  "layers": {
   "layer_1_inscription_anchors": {
    "present": false,
    "anchors_found": [
     "date"
    ],
    "anchor_count": 1,
    "orcid_in_body": false,
    "creator_in_body": false,
    "note": "Layer 1 present if at least 3 anchor types appear in body prose. Anchor types include ORCID, creator name, AXN string, hex identifier, sovereign_id, date, deposit number phrase, and authorial-position phrase."
   },
   "layer_2_micro_kernels": {
    "present": false,
    "fenced_json_blocks": 0,
    "json_ld_context_markers": 0
   },
   "layer_3_content_hash": {
    "present": true,
    "canonical_hash": "9fe090929e4da410d03fd00901d16f654d1ab70fb2acd4c69edc5dea3dbfd014",
    "hash_in_body": false,
    "note": "Hash is always registered by mint. hash_in_body true means the author inscribed a signature section."
   },
   "layer_4_cross_signing": {
    "present": false,
    "cross_references_count": 0,
    "note": "Populated by enrichment via cha_cross_references. Prospective (this deposit → prior deposits) only; retrospective cited_by updates on prior deposits."
   },
   "layer_5_external_anchors": {
    "present": true,
    "wikidata_qid_count": 0,
    "external_doi_count": 0,
    "orcid_present": true
   }
  },
  "overall_conformance": "partial"
 },
 "journal": "Transactions on Substrate Engineering (Trans. Substrate Eng.)",
 "journal_assignment": {
  "assigned": "2026-08-15",
  "by": "TACHYON under operator adjudication",
  "pass": 10,
  "method": "read per deposit — title and content_type, one at a time. No script classified anything.",
  "previous": null,
  "supersedes": "the 2026-06-21 preliminary batch mapping (#866), which assigned 864 deposits and put 371 in one venue",
  "authority": "data/cha-journals.json · datasets/venues/records/"
 },
 "falsification_conditions": "The score-ablation result fails if reading multiple scores off fixed weights at matched operating points shows consistent orientation across readouts on other pairs. The distillation result fails if faithful fixed-point deployment restores tail rank survival. The transportability result fails if architecture orderings prove stable across encodings on other datasets. The NAE handoff resolves either way: a converged normalized autoencoder reaching AUC 0.5 on the constituent pair would bound every inversion claim in this program to non-normalized scores. Nothing here bounds any open-world assimilation rate, and nothing here measures a deployed trigger.",
 "falsification_source": {
  "from": "body section",
  "heading": "## Falsification Conditions",
  "harvested": "2026-09-08"
 },
 "line": "accelerator-metrology",
 "line_parent": "science",
 "line_basis": "stated",
 "develops_from": [
  {
   "deposit": 1455,
   "what": "v0.3: ten registered tests, seventeen deviations, and what survives them"
  }
 ],
 "developed_by": [
  {
   "deposit": 1558,
   "what": "a baseline capture architecture for learned scientific triggers — the control"
  }
 ],
 "_projection": {
  "note": "Derived file. Canonical machine record is this entry in data/registry.json; the human record is the record_url. Do not edit this file.",
  "record_url": "https://www.alexanarch.org/s/records/1456/",
  "self_url": "https://www.alexanarch.org/data/records/1456.json",
  "registry_url": "https://www.alexanarch.org/data/registry.json",
  "text_url": "https://www.alexanarch.org/data/texts/AXN-05E1-text.md",
  "oai_pmh": "https://www.alexanarch.org/oai?verb=Identify"
 }
}