Founder brief · source-level correction

The refresh was wrong in 27 places. The hold survived.

CONCEPT STUDY · RESEARCH ONLY · NO HUMAN TESTING · NOT AVAILABLE · NO DEVELOPMENT PROGRAM EXISTS

Three independent source-review lanes checked every one of the 44 pre-Step-7 rows. R2 now preserves the corrected record, its source access limits and its traceability. Fresh QA passed the exact snapshot with no findings.

Decision: retain R2 as the reviewed correction annex. Do not use the original producer package or R1 as the active interpretation. Do not start Step 7 from this brief.

QA R2 passed Decision Hold unchanged Step 7 Not started

QA scope. The pass covers the internal consistency, provenance and correction quality of R2 registry 817b0ec52669d60d49b00eed2375de3add2aa162051e9d0619aecd307c1b4956. It does not accept the annex as decision-bearing evidence or admit it into the Step-7 merge.

Controlling state

The evidence is cleaner. The decision state is identical.

Preserved LIMEN decision state
Field	State	Meaning
Invasive route	`UNKNOWN`	No eligible complete candidate arm.
Removable route	`UNKNOWN`	No eligible optimized removable comparator arm.
Complete comparisons	`0`	Separate component studies cannot be pooled into a fictional comparison.
Bounded search	`OPEN_BOUNDED`	Failure to locate a source remains nonproof.
Science proposal	`PROPOSED_HOLD`	A proposal only; independent review has not adopted it.
Step 7 intake	`HOLD_MISSING_INPUT`	No panel, roster, conflicts, attendance, quorum, votes or dissent exist.

Material corrections

Five corrections change how the evidence must be read.

01 · SCIENCEBurden evidence was wrongly removed.

Full human-implant reports support source-specific observations of connector and contact failure, delamination, encapsulation, difficult retrieval, revision, infection-related explant and retained unpowered hardware. Those observations now remain separated by device, participant, duration and component class. They do not establish a pooled rate, reversibility or a LIMEN conclusion.

02 · DENOMINATORSSmall studies were repeatedly made broader than they are.

Cady’s AM-ULA comparison is N=1, three repetitions per condition, with prosthesis-malfunction trials excluded. Communication and percept observations include N=2, but the controller and performance stages do not. The accepted Step-5 record reports Christensen N=2; the accessible abstract does not, so R2 keeps that count as project-record-only rather than independently source-body verified.

03 · CONSENTThe withdrawal contradiction was manufactured.

ICH distinguishes stopping intervention, leaving a trial, follow-up and accrued-data consequences while also requiring respect for rights and freedom from coercion or interference. Helsinki and ICH therefore create implementation constraints, not a textual contradiction. Participation, physical removal, future use and data rights remain separate unanswered questions.

04 · GOVERNANCEAccepted law and recoverable guidance were described as absent.

Accepted HIPAA and GDPR provisions already covered authorization, access, amendment, lawful basis, special categories, recipients and transfers. Current official materials also recovered the HIPAA retention boundary, California minimization, EDPB model-anonymity analysis, WCAG interface criteria, DoD dependency controls and OECD neurotechnology principles. These texts sharpen questions; they do not establish LIMEN applicability.

05 · REGULATION & IPCurrentness and patent-family claims needed repair.

R2 records current MDR/MDCG and ATMP source structure without classifying LIMEN. The six patent documents are explicitly non-exhaustive: an earlier U.S. publication and issued patent belong in the known family record. California’s AB 1008 / SB 1223 sequence is resolved in current code rather than an open source conflict. Current force, ownership, claim scope, novelty, infringement and FTO remain unverified.

Evidence quality

R2 passed because it records limits instead of hiding them.

All rows accounted for: `GVR-001..044` appear exactly once and remain `OPEN`.
Traceability repaired: all 32 findings resolve, all 28 source-delta records are used, and every R1 source/GVR association is reciprocal.
Access recorded honestly: five formerly blocked GVR rows now have recovered source access; two remain only partially recovered.
Frozen inputs preserved: the accepted Science, Legal and merge registries, the producer package and R1 all reproduce unchanged.
Fresh QA: R2 closed with `P0=0; P1=0; P2=0; P3=0` for this exact correction-layer scope.

Still inaccessible or incomplete. Licensed standards clauses, part of the UNESCO text, official patent wrappers/current legal status, several population and denominator details, and every project-specific LIMEN actor, data flow, intended-use, constituent and operational fact needed for applicability remain unresolved.

Representative source receipts

Primary and official text drove the corrections.

Cady et al. full text — participant-specific denominators, performance stages and implanted-system burden.
Čvančara et al. full text — component condition, encapsulation, explant force and fragment retrieval.
ICH E6(R3) — consent, withdrawal, rights and non-interference safeguards.
HHS FAQ 580 — HIPAA itself does not impose a medical-record retention period.
EDPB Opinion 28/2024 — case-by-case model-anonymity analysis.
California Civil Code §1798.100 — retention notice and necessity/proportionality text.
WCAG 2.2 — interface criteria, not proof of whole-process accessibility.
Regulation (EC) No 1394/2007 — recovered ATMP source structure, without classification.

The complete source register records authority class, access status, inspected scope and nonproof. A link or catalogue record is never treated as inspected normative or claim text.

Next valid decision

R2 is ready to be considered as a future input—not silently merged.

The recovered sources and corrected populations, denominators, burdens, legal-text availability and source versions are changed inputs. Under L2REV-020, admitting any of them requires a new scoped frozen merge and fresh review. This run deliberately stops before that action.

Recommendation: preserve the producer package and R1 for audit history, use R2 as the active correction annex, and authorize a new merge only when you are ready to begin the next governance step. More broad searching is lower value than resolving the exact access and project-fact gaps already named.

Record custody

The reviewed source trail remains in controlled custody.

The R2 package overview, 44-row GVR register, corrected-findings register, source-delta register, access-recovery register, reconciliation, Independent QA verdict and Integrator reproduction remain internal controlled records. They are not exposed here as raw public files.

This public reading copy summarizes the reviewed R2 record. Public reachability does not admit R2 into the Step-7 merge, change its evidence status, or authorize development or release. The superseded candidate Web annex remains internal history.