# Raw verifier findings — main run wf_32dc4b47-6cf (2026-07-16)

Collected from the workflow journal for the repair-pass adjudicator. Findings on marker-bank reflect the PRE-repair truncated file (379 lines); the repair completes it, so structural-truncation findings are resolved by construction — re-verify content findings only.


---
## Verifier ac945b1dd48e35a00

Corpus note written to `/Users/fionnenglish/LIFE/specs/waypoint-research/corpus/laukkonen-2023-cessations-nirodha.md`.

DIGEST — Laukkonen et al. 2023, *Cessations of consciousness / nirodha samāpatti* (Progress in Brain Research):
- Paper gives Waypoint a **precise dual definition for the wheel 50–51 markers**: *nirodha* = brief (ms–sec) spontaneous "gap/cut/blip" in consciousness followed by clarity, the culmination of insight (magga/phala) = **small death**; *nirodha samāpatti (NS)* = willful, predetermined-duration absence up to 6–7 days = **big death**.
- **Cessation's necessary-and-sufficient rubric** (marker-ready): (1) total absence of experience, (2) **no retrospective content**, (3) subsequent clarity/openness/vitality. A gap without the after-glow doesn't count.
- **Hard gating ladder for NS** = strong "no-teleporting" prior: mastery of all 8 jhanas + non-returner/Anāgāmi-or-Arahant ethical transformation + both insight-and-samatha. A "7-day cessation" claim without this scaffolding is almost certainly a look-alike → default to claimed-not-corroborated / abstain.
- **Key discrimination for the confusion table**: cessation (absence, no one home) vs. pure-awareness / non-dual (wakeful contentless *presence*, someone home) — "vast open aware presence" is NOT cessation. Plus NS-vs-death, NS-vs-sleep/anesthesia.
- **Richest routine markers = post-cessation trait residue** (footnote i): stable de-reification, low grasping, present-centeredness, reduced self-reference, increased energy/flexibility — distinct from merely *describing* a dramatic event.
- **Methodology to copy**: neurophenomenology (trained-subject grading selects/scores events), necessary-and-sufficient definitional discipline, Nisbett-Wilson self-report skepticism → structure-scoring + behavioral/telemetry corroboration, and **factor-analyzing after-effect descriptors to set marker weights**.
- **Cautions**: nearly all neural data is preliminary n≈1 (treat as UNVERIFIED); "nirodha" is definitionally unstable (path/fruition vs full cessation); heavy reading-ahead contamination risk (the paper's own definitions are the script a well-read faker would recite); adverse-event risk feeds the teacher-flag/dark-night lane.
- **Load-bearing citations** (full list of ~94 refs + 18-item shortlist in file): Visuddhimagga (Buddhaghosa 2020), Sayadaw Progress/Manual of Insight, Nāṇamoli & Bodhi 1995, Laukkonen & Slagter 2021, Dahl et al. 2015, Lutz et al. 2015, Desbordes et al. 2015 (equanimity), Metzinger 2020 + Gamma & Metzinger 2021 (MPE-92M instrument), Lindahl 2017/2020 + Britton 2021 (safety), Nisbett & Wilson 1977, Petitmengin microphenomenology, King 1977.

---
## Verifier a9f130da585b1cd23

Corpus note written to `/Users/fionnenglish/LIFE/specs/waypoint-research/corpus/laukkonen-2025-beautiful-loop.md`

DIGEST:
- Genre: conceptual/formal theory paper, NO empirical data ("No data was used"). Value to Waypoint = a rigorous construct-space + vocabulary, not a validated instrument. Don't cite it as evidence a marker tracks stage.
- Core deliverable: a 3-axis phenomenological state space (abstraction × precision-distribution × epistemic-depth, Figs 5-6) — a ready template for Waypoint's per-pillar profile / position-vs-phase split instead of a scalar.
- Strongest linguistic stage marker: **dereification** — advanced practitioners describe thoughts/emotions as impermanent ownerless constructs ("anger arose and passed") vs naïve-realist reification ("I am anxious"). But the paper explicitly warns this phrasing is learnable — the exact "insight vocabulary vs lived insight" contamination; score structure+behavior, not vocabulary.
- Resolves two spec confusion pairs: (a) absorption-vs-dullness — MPE/absorption = high epistemic depth + gathered precision; dullness = low epistemic depth (same low-content report, opposite clarity-of-knowing); (b) felt-knowing ≠ accurate-knowing — noetic certainty is inducible/false (Grimmer, McGovern), so deflate self-reported insight/expanded-awareness.
- Stage-3 Unbinding markers: cessation/"small death" (transient reality-model collapse; often noticed as a "reset"/shift, under-claimed) with a corroborating after-effect signature (freshness, clarity, flexibility, compassion). Also lucid dreaming / clear-light sleep as advanced sustained-depth proxies.
- Two actual validated instruments named for the top-of-Wheel range → strong calibration-battery candidates: **MPE-92M** (Gamma & Metzinger 2021, pure awareness) and **NADA** (Hanley et al. 2018, nondual traits+states).
- Methodology to copy: orthogonal-axis decomposition; formalize each construct to a generative-model equation before claiming (Table 1); cross-framework concordance table with explicit ↝/✗ marks (Table 2) as a Wheel-band ↔ construct template.
- Load-bearing shortlist (18) leads with Lutz et al. 2015 (phenomenological matrix), Dahl et al. 2015 (3 meditation families), Laukkonen & Slagter 2021 (FA→OM→ND), Sparby & Sacchet 2024 (Jhāna classification), Fleming & Lau 2014 (measuring metacognition) — these seed the psychometrics + statistics lanes. Full ~230-entry reference list extracted in file §4a; Buddhist primary-source editor/translator details marked UNVERIFIED.

---
## Verifier a706f224b26a54e31

DIGEST — Agrawal & Laukkonen, Nothingness in Meditation (Chiron chapter / PsyArXiv 10.31234/osf.io/tygdf)
- INPUT FIXED: staged PDF was truncated at 5/34 pages (ended mid-sentence); full copy found in ~/Zotero/storage/TNYFFBYQ/ and now replaces specs/waypoint-research/inputs/agrawal-laukkonen-nothingness-emptiness-cessation.pdf. OSF downloads were 502ing.
- Core discriminator for the 50-60 band: emptiness (no-thingness, trait insight, de-reification) vs cessation (nothingness, discrete event, de-fabrication) are different marker families — never one "advanced" bucket.
- Most operationalizable line in the paper: non-dual-awareness reports assert awareness as ONTOLOGICAL ground; genuine emptiness/cessation reports are EPISTEMIC process observations undermining any ground (Attwood 2022 framing).
- Cessation transcript signature: gap recognized only retrospectively, no experience during, clarity on emergence; a claim of rich experience "during" points to 7th jhana / ND-awareness / sleep-state near-neighbors instead.
- Gives a 4-level emptiness depth ladder (concepts → self-as-construct → aggregates-fabricated → emptiness-of-emptiness); level 1 is reading-contaminable, level 4 mis-assertions ("void is ultimate reality") are themselves placement signals.
- Reproduces the Mahasi 16-nana map: dukkha nanas (stages 5-10) anchor dark-night-vs-depression; stages 13-15 last moments and are NOT self-distinguishable — expect reconstruction narratives, corroborate via trait deltas.
- Stream-entry = first cessation: a crisp datable milestone; fetter removal is claimed permanent (trait persistence check) but the paper concedes zero empirical evidence of irreversibility.
- Adaptable probe: "retention time / stickiness" — how long a charged topic persists in subsequent turns; appear-then-release = equanimity, never-appears = dissociation candidate, persists = reactivity.
- False-insight lab cluster (Grimmer/Laukkonen) is the strongest scientific footing for structure-scoring over conviction-scoring: Aha feelings are inducible and make false content feel true.
- Cautions: neural cessation evidence is N≤2 and contradictory (alpha-desync vs gamma-sync); sections 5-6 (active inference model, hibernation, AI) are explicit speculation; Gunaratana 2009 cited in text but missing from the reference list (UNVERIFIED which work).
- Full 91-entry reference list extracted + 20-work load-bearing shortlist (Sayadaw Manual of Insight, Britton adverse effects, Lindahl & Britton, Van Gordon 2019, Schmidt & Berkemeyer ASC database, Woods 2022, Alcaraz-Sanchez 2022, etc.) in the notes.
- Output: /Users/fionnenglish/LIFE/specs/waypoint-research/corpus/agrawal-laukkonen-nothingness-emptiness-cessation.md

---
## Verifier ae344b5ac9a98b5a4

The literature dossier is complete: 211 lines, all five required sections, every claim cited, UNVERIFIED flags carried through.

**Output:** `/Users/fionnenglish/LIFE/specs/waypoint-research/deliverables/literature-dossier.md`

DIGEST — Literature Dossier (deliverable 1 of 4):
- **Four independent families triangulated** (traditions / psychometrics / clinical / Laukkonen-PP), with the PP network capped at moderate confidence because lane 6 flags it as *one* author-network, not corroboration.
- **Five findings at very-high confidence (C5):** structure-over-content scoring; state≠station (reproducible+persistent moves the slow latent); a real+large report-only information ceiling (binds above ~41, per canon §10); the map contaminates the report; — these are cross-family, not PP-only.
- **The load-bearing precedent is ego-development sentence-completion scoring** (Loevinger→STAGES) + its LLM replication (Bronlet 2025, κ≈0.78, ≥5–10 item floor) — a near one-to-one map onto Waypoint's LLM-extract→ogive-style-aggregate design; this is the strongest external validation that Waypoint's core bet is sound.
- **Rejected with reasons:** Hawkins muscle-testing (negative control), self-report content scales as estimator spine (FFMQ DIF), PNSE clusters, conviction-as-evidence, v1 neural markers, single-scalar band claims.
- **All 5 spec §10 open questions answered directly:** granularity = band-level (sub-band only for milestone events + per-pillar profile; 100-point resolution overclaims); battery = NADA+MODTAS+Hood(ceiling)+MPE-92M+STAGES-sentence-completion, calibration-only; ñāna/kōan → 5 transferable probe moves; flag mapping = two-tier nowcast, conjunctive, history-neutral, ~5–15% rate.
- **Candor duty (§4):** Wheel AGREES on position-vs-phase, per-pillar, non-linear intensifiable top, small-death milestone. **Six items flagged for Surya (M2):** drop Hawkins numbers as ordinal check; band+interval not sub-band points; re-frame 51–60 Low-X as a recurring phase-cluster not fixed sequence; resolve the "Dissociation@43" label collision with the clinical risk-signature; decide the four-pillars→overall-band aggregation rule (not naïve max/mean); confirm devotional/bliss@46 scored as phase not position.
- **Must-read list:** 25 works, grouped A–G, ranked by builder-relevance, verification flags carried (O'Fallon κ numeric, MPE-92M DOI, Paulhus, Sparby&Sacchet-2025 all flagged confirm-on-pull).

---
## Verifier a9f591dc94419e3a8

Findings (verified against canon and spec; no CRITICALs — no finding breaks safety or build-blocking constraints):

1. **MAJOR — dossier §1.2–1.4 (lines 27–34) vs its own confidence scheme (lines 10–16).** All three claim [C5] (requires ≥3 of the 4 defined families T/P/C/N) but cite only T+N evidence — which the scheme itself caps at [C4]; §1.2 even calls four *lineages* (Sufi/Dzogchen/Patañjali/Theravāda, all family T) "four independent families." Fix: recount families per claim; downgrade §1.2–1.4 to [C4] or add genuine P/C evidence.

2. **MAJOR — dossier §5 item 7 (line 172) vs canon §8.** Describes the Sparby & Sacchet J1–J8 jhāna ladder as "top-of-Wheel (51+) milestone cartography," but the canon's Theravada overlay places jhāna-factor and formless-attainment territory at ~31–50 (Delightful Sensations 31 → No-thingness 49–50), with only cessation at 51. As written it would steer jhāna reports into band 6. Fix: re-anchor the ladder to bands 4–5 (~31–51).

3. **MAJOR — dossier §1.5/§1.9 (lines 36–37, 48–49) vs spec §5 observability.** The anti-contamination rule "weight off-cushion behavioral markers heavily above ~41" / "only when corroborated by structure + behavior" requires data the system cannot observe: spec §5 channels are conversation, probes, telemetry — above ~41 "behavior" collapses into self-report (the same contaminated channel) plus discipline-only telemetry. Fix: restate as claim-vs-telemetry consistency + durable linguistic trait-shift, and name the observability gap explicitly.

4. **MINOR — dossier §1.2 (line 28) misstates spec §6.** Claims "direct empirical backing for the spec §6 HGF rule: only volitionally reproducible + persistent evidence revises wheel position" — spec §6 contains only persistent-runs-of-error; "volitionally reproducible" is the dossier's tradition-derived addition. Fix: present reproducibility-at-will as a recommended amendment, not existing spec content.

5. **MINOR — dossier §4.3 heading + §4.7 item 3 (lines 137, 152) vs canon §3/§7 and the dossier's own body.** Calls it "the 51–60 Low-X cascade"; the cascade is 52–56 (51 = Unbinding onset; 57–60 = Non-Dual Loop/Deconstruction/Void/Luminosity), and the §4.3 body itself says "treat 52–56 as a phase cluster." Fix: rename to 52–56.

6. **MINOR — dossier §3.4 (line 120) internal inconsistency.** "Expected ~5–15% Tier-A rate" vs the same paragraph's miscalibration band "outside ~2–15%." Fix: pick one range and use it in both places.

7. **MINOR — family-tag misassignments (lines 72, 108, 25, 28).** NADA/Hanley 2018 tagged (C) twice though it is psychometrics (P); Nisbett & Wilson 1977 and Cahn & Polich 2006 tagged (N) though outside the defined Laukkonen network — both inflate apparent cross-family independence. Fix: retag Hanley as P; give general-cognitive-science sources a non-N tag or note them as outside the four families.

8. **MINOR — dossier §4.4 (line 141) violates its own verification convention (line 8).** `Lindahl & Britton 2019` appears once, with no [VERIFIED]/[UNVERIFIED] flag, no venue, and no entry in the must-read list. Fix: add full citation + verification flag or drop to "lane 4."

9. **MINOR — dossier §1.3 (line 31) overclaims cited support.** "Nine lineages insist some part of attainment is visible only to a live teacher" but only five are evidenced (Dzogchen, Zen, Neidan, Mahāsi, Christian). Fix: cite lane 1 §11.4 for the full nine or soften to "multiple lineages."

Files: /Users/fionnenglish/LIFE/specs/waypoint-research/deliverables/literature-dossier.md (target), /Users/fionnenglish/LIFE/specs/waypoint-research/corpus/internal-canon.md, /Users/fionnenglish/LIFE/specs/waypoint-practitioner-stage-assessment.md.

---
## Verifier acea43b9ac9c23b3d

All 15 load-bearing citations checked out — existence confirmed via Crossref/arXiv, and each plausibly supports its use. No fabrications or misattributions found. Details below, weakest-support first.

VERIFIED-EXISTENCE-ONLY (support claim not directly re-read, matches paper's known central finding): Belsher et al. 2019, JAMA Psychiatry 76(6):642 (§2.1/§3.4/must-read #20) — paper real; the specific "event-prediction PPV ≤0.01 at low base rates" is the paper's central thesis (systematic review + simulation of suicide prediction) so plausibly supported, but I could not open the full text to confirm the exact figure.
VERIFIED w/ minor note: Laukkonen 2026 "Clear Mind" f-SNR, arXiv:2606.29698 (§1.2/1.6/1.8/1.10) — ID looked anomalous (high sequence no.) but confirmed real, submitted 29 Jun 2026; f-SNR/state-trait-depth content matches usage. Load-bearing and legitimate.
VERIFIED: Mago et al. 2025, arXiv:2511.20990 (§1.10) — real; abstract explicitly reports enhanced frontocentral MMN amplitude in jhāna, supporting "jhāna sharpens deviance-sensitivity."
VERIFIED: O'Fallon, Polissar, Neradilek & Murray 2020, Heliyon 6(3):e03472 (§1.1/2.1/must-read #1) — exact title/authors/year confirmed; supports three-dimension STAGES rubric claim.
VERIFIED: Bronlet 2025, Front. Psychol. 16:1488102 (§2.1/must-read #2) — confirmed; weighted κ=0.779 expert-vs-LLM agreement matches the "κ≈0.78" claim exactly.
VERIFIED: Grimmer, Laukkonen, Tangen & von Hippel 2022, Psychon. Bull. Rev. 29(3):954–970 (§1.1/2.2/#21) — confirmed; abstract confirms false insights are inducible via semantic priming.
VERIFIED: Laukkonen & Slagter 2021, Neurosci. Biobehav. Rev. 128:199–217 (§1.3/1.9/#9) — confirmed title/vol/pages; FA→OM→ND predictive-mind framing supports usage.
VERIFIED: Britton et al. 2021, Clin. Psychol. Sci. 9(6):1185–1204 (§1.7/3.4/#18) — confirmed; abstract confirms dysregulated-arousal as the long-lasting-effect signature.
VERIFIED: Lindahl et al. 2020, "Progress or Pathology?", Front. Psychol. 11:1905 (§1.7/3.4/#17) — confirmed title/authors/journal; supports teacher-criteria / differential claim.
VERIFIED: Lindahl et al. 2017, "Varieties of Contemplative Experience," PLoS ONE 12(5):e0176239 (§3.4/4.3/#16) — confirmed; supports difficulty taxonomy + same-phenomenology-appraised-differently claim.
VERIFIED: Goldberg et al. 2022, Psychotherapy Research 32(3):291–305 (§3.4/#19) — confirmed title/vol/pages; population base-rate use plausible (specific ~10%/1.2% figures not individually re-quoted but consistent).
VERIFIED: Hanley, Nakamura & Garland 2018, Psychological Assessment 30(12):1625–1639 (§2.1/3.2/#13) — confirmed; NADA trait+state instrument supports non-dual-axis usage.
VERIFIED: Gamma & Metzinger 2021, PLoS One 16(7):e0253694 (§2.1/#14) — confirmed (dossier flagged "confirm on pull"; now confirmed); MPE-92M 12-factor pure-awareness profile matches.
VERIFIED: Lutz, Jha, Dunne & Saron 2015, Am. Psychol. 70(7):632–658 (§1.8/#10) — confirmed; multi-axis phenomenological matrix supports per-pillar claim.
VERIFIED: Laukkonen et al. 2023, "Cessations… nirodha samāpatti," Prog. Brain Res. 280:61–87 (§1.5/4.5/#12) — confirmed year/pages/title (Crossref-listed under the PBR series); supports small-death/after-effect claims.
VERIFIED: Van Dam et al. 2009, Personality & Individual Differences 47(5):516–521 (§2.2/#23) — confirmed; FFMQ differential-item-functioning across meditators supports the DIF/reject-as-spine argument.

Note: two searches were safeguard-blocked, so I verified via Crossref/arXiv DOI lookups instead (more reliable for citation metadata). Every DOI in the dossier's must-read list that I resolved was live and correctly attributed; the document's own [UNVERIFIED] flags (WUSCT manuals, Cook-Greuter, Sparby & Sacchet 2025 vol/pages) were not among the 15 and remain appropriately hedged.

---
## Verifier a21585813d4802f4c

Verification complete. Read all three files fully (marker bank is 379 lines — it ends mid-document at M-A4-010). Findings, most severe first:

1. **CRITICAL — whole document (ends line 379, after M-A4-010):** File is truncated/incomplete. Scope (line 6) and spec §10 deliverable 2 promise bands 3–7 × pillars + phase axis + confusion table, but only bands 3–4 exist. Sections the doc load-bearingly cross-references are absent: §4 guard logic, §5 confusion table, §7 source keys — leaving ~20 dangling refs (M-NX-005/006/007/009/012/013, M-PDP-001/005, M-PGL-001, M-T5-003, M-S5-003, CT-1–CT-7, and every source key [L1][L5][AL][BL][MtN][FSNR][NIR][VOH]). Fix: restore or regenerate the missing sections before this counts as deliverable 2.

2. **MAJOR — M-S4-004 (line 181) and M-A4-009 (line 365):** Both key markers to bare "Stage N→N+1" gates, which canon §9-A explicitly calls ambiguous. M-S4-004 anchors the Stage-2→3 hard gate (reverse breathing, the bank's strongest sensations marker) at the band 4→5 boundary (~40); under §9-A's mapping (app Stage 2 ≈ band 5, 41–50) the same gate reads ~50/51 — a full-band displacement. The §6/§9-J reading the bank silently adopted is defensible, but canon requires recording which sense is meant. Fix: add an explicit reconciliation note citing §9-A/§9-J so Surya calibration (M2) adjudicates knowingly.

3. **MINOR — §1.2 Imitability, line 28 (repeated line 323):** "Redmore 1976, PMID 16367343" — that PMID range corresponds to ~2005/2006 publications; a 1976 paper cannot carry it. The citation grounds the entire imitability field. Fix: verify against PubMed and correct or drop the PMID.

4. **MINOR — M-S4-002 (line 168) / M-S4-004:** Gate-position telemetry is keyed to the 7-step CM breathwork decomposition, but canon §9-D marks step counts UNRESOLVED (7 vs 8 vs 9; "trust the day-totals"). A checker keyed to the wrong step count misfires on "teleport" detection. Fix: add the §9-D caveat; key consistency checks to day-totals, not step index.

5. **MINOR — M-X3-002 (line 66):** Anchor "22–28" glossed "(Survival→Civilization)" — canon §3 puts Civilization at 29 (28 = Culture). Fix: either widen anchor to 22–29 or re-gloss as Survival→Culture.

6. **MINOR — M-A4-009 (line 368):** Stage-1 session grammar omits Mobility; canon §4.1 grammar is Posture → Mobility → Opening → … → Closure. Fix: insert Mobility so telemetry matching targets the canonical cascade.

7. **MINOR — M-S4-005 (line 195):** Cites "[CANON] §8 (Theravada overlay rows 31–40)" as support for A&P placement, but those rows read Delightful Sensations / Joy / Contentment — no arising-and-passing content. Overclaims canon support. Fix: drop the canon cite and rest on [AL] §1.4 alone.

8. **MINOR — M-X3-001 (line 59):** Imitability "low" contradicts the field's own up-faking definition for a floor marker the same line admits is trivially faked *down*; the "deflation out-of-model for v0" caveat is buried in one parenthetical though it governs all band-3 floor markers. Fix: promote the caveat to a §1.3 global rule and mark floor-marker imitability "n/a (down-faking)".

No spec §5 observability violations found (all `telemetry` markers use spec-listed channels), and no §6 estimator-principle violations within the extant text (extractor proposes, aggregator integrates and discards evidence-free events). The elicitation-gap figures (3.7% vs 33.2%, Farias 2020) and band/scale-point anchors for bands 3–4 all check out against the canon.

---
## Verifier a9912d000eb4919bb

WORST FIRST — then the rest, one line per checked claim:

1. UNSUPPORTED (structural): file ends at line 380 mid-§2.2; §§3–7 referenced throughout (guard logic, confusion table CT-n, and the §7 source-key table for [L1][L5][AL][BL][MtN][FSNR][NIR][VOH]) do not exist in the file — internal lane citations are unresolvable as delivered.
2. PLAUSIBLE-BUT-UNCONFIRMED: line 135, "Anālayo 2018" anusaya hindsight→live→non-arising ladder — Anālayo 2018 "The Underlying Tendencies" (Insight Journal 44:21–30) exists and covers mindfulness vs latent reactivity, but the classical ladder there is dormant→obsession→transgression (anusaya/pariyuṭṭhāna/vītikkama), not noticing-timing; the hindsight→live framing looks like an interpretive gloss. Correction: cite as interpretation of, not statement in, Anālayo 2018.
3. VERIFIED w/ NUANCE: line 39, Farias et al. 2020 "3.7% vs 33.2% elicitation gap" — real (Acta Psychiatr Scand 142:374–393, PMID 32820538); exact figures confirmed, but the paper's split is experimental (passive monitoring) 3.7% vs observational (active inquiry) 33.2% study designs — "spontaneous vs probe-elicited reports" is a fair but loosened reading.
4. VERIFIED: line 7, Rogers et al. 1993 = Rogers, Bagby & Chakraborty, "Feigning schizophrenic disorders on the MMPI-2: detection of coached simulators," J Pers Assess 60(2), PMID 8473961 — supports detection-strategy-coaching claim.
5. VERIFIED: line 7, Storm & Graham 2000, "Detection of Coached General Malingering on the MMPI-2," Psychological Assessment — directly supports "coaching on detection strategy degrades detection more than symptom/map coaching."
6. VERIFIED: line 28, Redmore 1976, "Susceptibility to faking of a sentence completion test of ego development," J Pers Assess — PMID 16367343 checks out exactly; direction confirmed (fake-down easy, fake-up hard).
7. VERIFIED: line 38, Orne 1962, "On the social psychology of the psychological experiment," American Psychologist 17:776–783 — canonical demand-characteristics source; supports mechanism as used.
8. VERIFIED: line 43, Grossman 2008 (J Psychosom Res) + Grossman 2011 "Defining mindfulness by how poorly I think I pay attention…" (Psychol Assess 23:1034–40, PMID 22122674) — supports shifting-internal-anchor / deflationary self-rating claim.
9. VERIFIED: line 43, Van Dam et al. 2018 "Mind the Hype" (Perspect Psychol Sci 13(1):36–61, PMID 29016274) — supports self-report unreliability claim.
10. VERIFIED: lines 63/320–323, U Paṇḍita 1992 In This Very Life — the three-phase interview formula (occurred → how noted → what happened to it) is verbatim the book's interview-reporting instruction.
11. VERIFIED: line 227, Van Gordon et al. 2019 "Exploring Emptiness…" (Explore 15(4):261–272, PMID 30660506) — emptiness meditation in advanced meditators improved non-attachment AND positive affect; supports co-occurrence exemplar use.
12. VERIFIED: line 339, Lutz, Jha, Dunne & Saron 2015 "phenomenological matrix" (Am Psychol 70:632–658, PMID 26436313) — supports FA/OM bare non-reactive field phrasing.
13. VERIFIED: line 119, Loevinger impulsive/self-protective ego-development ladder (WUSCT) — standard; externalized-blame framing matches Self-Protective stage descriptions.
14. VERIFIED: lines 41/355, Pa-Auk five masteries — the five vasī (adverting, attaining, resolving, emerging, reviewing) are Visuddhimagga doctrine taught at Pa-Auk; supports "volitionally reproducible + persistent" rule.
15. VERIFIED: line 42, bhūmi rule "equipoise identical across bhūmis, only post-meditation qualities differ" — standard Gampopa/Kagyu doctrine (Rigpa Wiki, Jewel Ornament framework). Line 41 Sufi maqām (stable, earned) vs ḥāl (transient) is likewise textbook doctrine (al-Qushayrī); web check for it was blocked by the safety classifier, not by absence of sources.

Net: no fabricated or misattributed citations found among the 15; weakest links are the missing §§3–7 (unresolvable internal keys), the Anālayo ladder gloss, and the loosened Farias framing. The doc's own honesty flag at line 179 (Neidan zhèngyàn lists UNVERIFIED) is accurate and should stay.

Sources: [PubMed 16367343](https://pubmed.ncbi.nlm.nih.gov/16367343/), [PubMed 32820538](https://pubmed.ncbi.nlm.nih.gov/32820538/), [Storm & Graham 2000](https://www.researchgate.net/publication/12430894_Detection_of_Coached_General_Malingering_on_the_MMPI-2), [Coventry Pure — Farias 2020](https://pureportal.coventry.ac.uk/en/publications/adverse-events-in-meditation-practices-and-meditation-based-thera), [Orne 1962 PDF](https://pages.ucsd.edu/~cmckenzie/Orne1962AmPsychologist.pdf), [Grossman 2011](https://pubmed.ncbi.nlm.nih.gov/22122674/), [Van Dam 2018](https://pubmed.ncbi.nlm.nih.gov/29016274/), [In This Very Life PDF](https://people.eecs.berkeley.edu/~alanmi/hamilton/u_pandita/in_this_very_life.pdf), [Anālayo 2018](https://www.buddhistinquiry.org/article/the-underlying-tendencies/), [Van Gordon 2019](https://pubmed.ncbi.nlm.nih.gov/30660506/), [Lutz 2015](https://pubmed.ncbi.nlm.nih.gov/26436313/), [Loevinger stages](https://en.wikipedia.org/wiki/Loevinger's_stages_of_ego_development), [Five masteries](https://www.accesstoinsight.org/lib/authors/gunaratana/wheel351.html), [Ten bhumis](https://www.rigpawiki.org/index.php?title=Ten_bhumis)

---
## Verifier a0dd2968bff519d8c

CRITICAL — target missing: /Users/fionnenglish/LIFE/specs/waypoint-research/deliverables/probe-item-bank-v0.md does not exist. Verification not performed; no findings can be reported against it.

Verified before concluding (not a stale-path guess):
- deliverables/ contains only literature-dossier.md (15:56), marker-bank-v0.md (15:59), estimator-design-memo.md (16:18) — deliverables 1, 2, 4 of the four required by spec §10 ("Probe item bank v0" is deliverable 3, spec lines 32, 134-137).
- `find` across ~/LIFE and $HOME (depth 3) for `*probe-item*`: zero hits; `grep -r "probe-item-bank"` across waypoint-research: zero hits — the file was never written under any path, not merely misplaced.
- No in-progress copy in LIFE worktrees.

Most likely cause: the agent responsible for drafting the probe item bank has not produced it yet (estimator memo landed 16:18; bank may still be in flight from another lane agent), or it was written into a different session's worktree and never landed.

Fix: have the orchestrator (a) confirm the producing agent finished and get the actual output path, then re-invoke this verifier with that path; or (b) if the bank was never drafted, draft it first — spec §137 defines the deliverable: "items with scoring keys + delivery constraints, ready for Surya calibration."

Do not substitute marker-bank-v0.md as the target without explicit instruction: it is a different deliverable (extraction markers, not conversational probe items), and findings against it would not answer the probe-bank checks requested (invisible-delivery violations, item-level safety contraindications, scoring-key observability).

Context read and ready for a fast re-run once the file exists: canon (413 lines) and spec (167 lines) both read; key constraints located — spec §72 (invisible delivery, frequency caps, no probes in flagged-vulnerable moments, never verbatim-repeated, item-id logging, Surya approval), §5 evidence channels (lines 25, 72), §6 binding principles (line 84: LLM-proposes/mechanics-score, partial pooling, horizon-scored vs base-rate null, cross-family extractor/labeler), canon §2 band names, §3 scale-row anchors, §9-H Hawkins LOC non-monotonicity.

---
## Verifier a056b6af75491f612

All 15 load-bearing citations checked (web search + Crossref/arXiv DOI resolution). No fabrications or misattributions found. Worst findings first:

1. VERIFIED-WITH-CAVEAT — Hui & Walter (1980), line 314/414: paper exists (Biometrics 36, DOI 10.2307/2530508) but the original design is 2 tests × 2 populations; the memo's "embryonic 4-test Hui–Walter design" is an extension of the method, not the 1980 paper itself. Supports the use with that caveat; memo already hedges ("embryonic").
2. VERIFIED (upgrade available) — Grove et al. (2000), line 413: memo marks it UNVERIFIED-BY-LIVE-CHECK, but it live-verifies: "Clinical versus mechanical prediction: A meta-analysis," Grove, Zald, Lebow, Snitz & Nelson, *Psychological Assessment* 12(1):19–30, DOI 10.1037/1040-3590.12.1.19. The UNVERIFIED flag can be removed (add that DOI).
3. VERIFIED-MINOR — Corbett & Anderson (1995), lines 69/407: exists (UMUAI 4, 253–278, DOI 10.1007/BF01099821); some sources date it 1994 (vol 4 issue 4); 1995 is the standard citation. Supports BKT guess/slip use.
4. VERIFIED-MINOR — Gosling (2018) SHELF, lines 81/412: chapter exists (Springer ISOR 261, pp. 61–93, DOI confirmed); published online 2017, print 2018. Supports SHELF-style elicitation.
5. VERIFIED — Mathys et al. (2011), *Front. Hum. Neurosci.*, DOI 10.3389/fnhum.2011.00039, lines 54/417: "A Bayesian foundation for individual learning under uncertainty." Supports the hierarchical error-routing principle.
6. VERIFIED — Mathys et al. (2014), DOI 10.3389/fnhum.2014.00825, lines 54/417: "Uncertainty in perception and the Hierarchical Gaussian Filter." Supports "adopt the principle, not the filter."
7. VERIFIED — Gigerenzer & Hoffrage (1995), *Psych. Review* 102(4):684–704, lines 81/411: natural-frequency formats improve Bayesian reasoning. Directly supports the natural-frequency elicitation format.
8. VERIFIED — Jackson (2011), *JSS* 38(8) msm package, lines 129/415: multi-state models for panel data, exp(QΔt) kernels. Supports the continuous-time birth–death movement model.
9. VERIFIED — Liu, Li, Li, Song & Rehg (2015), NeurIPS 2015, lines 129/416: "Efficient Learning of Continuous-Time Hidden Markov Models for Disease Progression." Author list and venue match; supports CT-HMM machinery.
10. VERIFIED — Rabiner (1989), *Proc. IEEE* 77(2):257–286, lines 161/418: HMM tutorial incl. forward–backward. Supports the filter-vs-smoother consolidation design.
11. VERIFIED — Gelman (2006), *Bayesian Analysis* 1(3), lines 202/410: priors for variance parameters in hierarchical models. Supports M6 informative-prior plan.
12. VERIFIED — El-Yaniv & Wiener (2010), *JMLR* 11:1605–1641, lines 218/408: introduces the risk–coverage tradeoff for selective classification. Directly supports the abstention framing.
13. VERIFIED — Chow (1970), *IEEE Trans. IT* 16(1):41–46, DOI matches, lines 218/405: optimum reject rule. Supports abstention-as-first-class-output.
14. VERIFIED — Epstein (1969) *J. Appl. Meteorology* 8(6):985–987 + Murphy (1971) 10:155–156, lines 318/409: original RPS paper + note; RPS is ordinal-aware and proper as claimed. DOIs match.
15. VERIFIED — Cohen (1968), *Psych. Bulletin* 70(4):213–220, lines 320/406: weighted kappa. Supports quadratic-weighted κ for pillar agreement; Angelopoulos & Bates (2021) arXiv:2107.07511 (line 404) also confirmed as the conformal-prediction primer.

Summary: 15/15 exist and plausibly support their uses; zero fabricated/misattributed. Actionable corrections: (a) lift the Grove et al. UNVERIFIED flag and add DOI 10.1037/1040-3590.12.1.19 (line 413); (b) optionally note at line 314 that the 4-test design extends Hui–Walter beyond the original 2-test/2-population formulation. Target file: /Users/fionnenglish/LIFE/specs/waypoint-research/deliverables/estimator-design-memo.md (not edited).

Sources: [Springer UMUAI](https://link.springer.com/article/10.1007/BF01099821), [NeurIPS 2015](https://proceedings.neurips.cc/paper/2015/hash/a591024321c5e2bdbd23ed35f0574dde-Abstract.html), [AMS Epstein 1969](https://journals.ametsoc.org/view/journals/apme/8/6/1520-0450_1969_008_0985_assfpf_2_0_co_2.xml), [JMLR el-yaniv10a](https://www.jmlr.org/papers/volume11/el-yaniv10a/el-yaniv10a.pdf), [ACM Chow 1970](https://dl.acm.org/doi/10.1109/TIT.1970.1054406), [arXiv 2107.07511](https://arxiv.org/abs/2107.07511), plus Crossref records for DOIs 10.2307/2530508, 10.1007/978-3-319-65052-4_4, 10.3389/fnhum.2011.00039, 10.3389/fnhum.2014.00825, 10.1037/0033-295x.102.4.684, 10.18637/jss.v038.i08, 10.1214/06-ba117a, 10.1037/h0026256, 10.1109/5.18626, 10.1037/1040-3590.12.1.19.

---
## Verifier aa1bb4f3260849897

Verification done. I read the memo, canon, and spec in full and hand-checked the toy-example arithmetic (E1/E2/E2b/E4 updates, kernel step, phase update, BKT chain, sufficiency sums, point estimates). Findings, most severe first:

1. **MAJOR — memo §10.5 (line 336):** Adversarial persona "dark-night at 38" contradicts canon: dark night is Unbinding-adjacent, ~52–59 (canon §7; also §9's own dip-gating cites it there). A band-4 dark-night persona tests the depression confusion pair at a canon-wrong position. Fix: place the persona at 52–59, or rename it a low-band "dip" persona and add a true 52–59 dark-night case.
2. **MAJOR — memo §9 (line 267):** "Prior from the day-24 stage posterior: P(reached) = 0.175" is not the day-24 posterior. 0.175 only reproduces from the post-E2 state (0.038, 0.718, 0.229, 0.014), i.e. excluding E2b and E4; the actual day-24 pre-milestone posterior is 0.196 (post-E2b) or 0.115 (post-E4, per the §4.2 step ordering where stage updates before milestones). Contradicts "Numbers are exact." Fix: recompute the milestone chain from the correct state and relabel.
3. **MINOR — memo §7 (line 221):** "or ~4 weak" fails the stated floor: 4 × 0.6 = 2.4 < E_min 2.5. Fix: say 5 weak (3.0 bits) or lower E_min.
4. **MINOR — memo §2.1 (line 62):** "bands 1–2 and 8–10 are structurally ~0" — grid G = {15…70} contains band-2 points 15–20, so band 2 is not *structurally* zero (only 1 and 8–10 are outside G). Fix: "band 1 and 8–10 structurally 0; band 2 prior-negligible."
5. **MINOR — memo header (line 3) vs §14 (line 390):** "before M0 sign-off of the revised spec" conflates milestones — M0 is original spec approval; the spec revision closes M1 (as §14 itself says). Also spec §10 requires Oracle+Codex reviews *attached* to deliverable 4; the memo self-declares "complete" without them. Fix: change to M1 close and mark the deliverable pending-reviews.
6. **MINOR — memo §10.6 (line 342) vs §11 (line 354):** "Synthetic personas are never authored, extracted, or judged by the same family" contradicts §11's "acceptable C = B" (B is the M4–M5 judge family). Fix: either drop the C = B allowance or narrow §10.6 to "never extracted by the authoring family; M3 judging is deterministic."
7. **MINOR — memo §1 (line 20) vs §3.2 (line 109):** "every number signed by a human" is overclaimed: the v0 bootstrap generates all rows mechanically from bank fields until Surya's M2 pass. Fix: qualify as "every number human-signed from M2; v0 bootstrap machine-derived and traceable."
8. **MINOR — memo §9 (lines 255, 284):** E3 "non-dual glimpse" (M-A4-008, band-4 locus) is read as the `first_glimpse` (awakening) milestone claim. Canon separates Non-Dual State (39, a Stage-2 OA attainment) from awakening glimpses (~41, band 5) and §9-J explicitly requires recording which "awakening" sense a marker means. Fix: state which sense M-A4-008 encodes and key `first_glimpse` to the ~41 anchor.

No spec-§5 observability violations found (all marker/telemetry inputs map to the three evidence channels), no invisible-delivery or safety-contraindication breaches (probes pause on flagged-vulnerable; safety path reads raw ungated markers), and all other toy-example numbers verify exactly.

File: /Users/fionnenglish/LIFE/specs/waypoint-research/deliverables/estimator-design-memo.md

---
## Verifier ac2605ae1956c04b9

All 16 checks complete. Final verification summary.

TARGET MISSING: `/Users/fionnenglish/LIFE/specs/waypoint-research/deliverables/probe-item-bank-v0.md` does not exist (waited ~10 min; queried main, no reply). Deliverable 3 of 4 is unwritten. Verified instead the 15 most load-bearing citations in the shared base it will rest on: `deliverables/marker-bank-v0.md` (MB) + `lanes/6-citation-map.md` (CM6) + `lanes/4-safety-adverse-events.md` (L4).
SECOND STRUCTURAL FINDING: marker-bank-v0.md is truncated — 380 lines, ends mid-§2.2 (Band 4, M-A4-010); bands 5–7 and its promised §3–§7 (guard logic, confusion table, source-key legend) are absent despite being referenced in the header.

1. UNSUPPORTED-AS-USED — Farias et al. 2020 "3.7% vs 33.2% elicitation gap" (MB line 39 rule 4; L4 line 138). Paper exists (Acta Psychiatr Scand 142(5):374–393, PMID 32820538) and the numbers are real, but the split is EXPERIMENTAL vs OBSERVATIONAL study design, not spontaneous-vs-probe-elicited reports. L4 §3 (line 38) states it correctly; rule 4 and L4 §7.2 misframe a design gap as a measured elicitation effect. Correction: cite as "experimental vs observational design gap; elicitation interpretation is our inference."
2. VERIFIED-WITH-CAVEAT — Laukkonen (2026) "Clear Mind" [CM]/[FSNR] (MB lines 95, 203, 243, 331) is an arXiv preprint (10.48550/arXiv.2606.29698), not peer-reviewed; several markers rest on author-derived, unreviewed claims (MB already tags §1e items "low weight" — keep that discount).
3. VERIFIED-WITH-CAVEAT — Vohryzek et al. 2025 [VOH] (MB line 347): exists but is a bioRxiv preprint ("Whole-brain models of minimal phenomenal experience... jhāna meditation"), not the published paper CM6 line 6 implies.
4. VERIFIED-WITH-CAVEAT — Agrawal & Laukkonen 2024 [AL] (MB lines 47, 135 etc.): exists as PsyArXiv preprint 10.31234/osf.io/tygdf / Chiron book chapter, not a journal article.
5. VERIFIED — Redmore 1976, "Susceptibility to faking of a sentence completion test of ego development," J Pers Assess 40(6):607–616; PMID 16367343 is correct (MB line 33); supports faking-up-hard/faking-down-easy as used.
6. VERIFIED — Rogers, Bagby & Chakraborty 1993, J Pers Assess 60(2):215–226, PMID 8473961 (MB line 8): strategy-coaching evaded MMPI-2 detection, symptom-coaching didn't — directly supports the internal-only distribution rule.
7. VERIFIED — Storm & Graham 2000, Psychol Assess 12(2):158–165, PMID 10887761 (MB line 8): coaching degrades validity-scale detection; supports as used.
8. VERIFIED — Farias et al. 2020 existence/authors/journal (L4 line 37) — metadata exact.
9. VERIFIED — U Paṇḍita 1992 *In This Very Life* (Wisdom Publications; MB lines 63, 219, 323): "What Occurred, How You Noted It, What Happened to It" is a verbatim section of its interview-instructions chapter; supports M-A4-003 exactly.
10. VERIFIED — Lutz, Jha, Dunne & Saron 2015, Am Psychol 70(7):632–658, DOI 10.1037/a0039585 (CM6 line 18): multidimensional phenomenological matrix; supports rubric use.
11. VERIFIED — Dahl, Lutz & Davidson 2015, TiCS 19(9):515–523, PMID 26231761 (CM6 line 19): attentional/constructive/deconstructive families; supports pillar-typing use.
12. VERIFIED — Laukkonen & Slagter 2021 "From many to (n)one: Meditation and the plasticity of the predictive mind," Neurosci Biobehav Rev, DOI 10.1016/j.neubiorev.2021.06.021 (CM6 line 20 / [MtN]).
13. VERIFIED — Laukkonen, Friston & Chandaria 2025 "A beautiful loop: An active inference theory of consciousness," Neurosci Biobehav Rev, DOI 10.1016/j.neubiorev.2025.106296 (CM6 line 6 / [BL]); epistemic-depth axis supports as used.
14. VERIFIED — Laukkonen et al. 2023 nirodha samāpatti, Prog Brain Res, DOI 10.1016/bs.pbr.2022.12.007 (CM6 line 39 / [NIR]).
15. VERIFIED — Gamma & Metzinger 2021 MPE-92M, PLoS One 16(7):e0253694 (CM6 line 28 — clears its 🔷 flag); 92-item pure-awareness instrument as described.
16. VERIFIED — Sparby & Sacchet 2025, Mindfulness 16(9):2472–2482, DOI 10.1007/s12671-025-02632-6 (clears CM6 line 103 UNVERIFIED flag); and Ehmann et al. 2025, Imaging Neuroscience 3:IMAG.a.82 (clears same flag) — both real, metadata match.

No fabricated citations found; the base is sound. Worst issues: the missing/truncated deliverables themselves, the Farias "elicitation gap" misframing, and three seed papers being preprints cited without that caveat.