Falsifiability Sheet Egyptian Inscription Validation Project

One sentence: Test whether an evidence-first review process can keep ancient-inscription claims bounded, auditable, and revisable when evidence is strong, contradictory, uncertain, or incomplete. 1. Problem and Purpose Ancient-script claims can compress object identity, sign recognition, reading, and interpretation into one conclusion. This project evaluates the Falsifiability Sheet as a structured review process that keeps those layers visible. It does not claim a new decipherment of Egyptian hieroglyphs; it tests whether support, contradiction, uncertainty, and missing evidence can be preserved without allowing conclusions to exceed the sources. 2. Why the Project Is Ready • The Falsifiability Sheet and evidence-review workflow already exist. • British Museum EA1006 provides a completed known-baseline control case with a Green/Supported result. • The Nefretiabet and Rekhmire evidence dossiers are completed and locked as Version 1.0 research corpora. • The next step is stronger validation through defined human review, evidence-bound AI review, versioned public records, and cross-case comparison. 3. Validation Series Case Role Evidence state Status EA1006 Control Known supported attribution; completed prior case Complete Nefretiabet — Louvre E 15591 Challenge A Corpus v1.0 locked; submitted Aa1 claim tested against object-specific palaeography and published transcription Review/result sequence preserved and completed Rekhmire / EID-EGY-002 Challenge B Corpus v1.0 locked; dossier does not support submitted Aa1 claim; exact alternative unresolved Existing human record preserved; remaining review/result sequence completed 4. Team and Independence • Michael Louis Grasa — Project Lead and creator of the Falsifiability Sheet. • Mahmoud El Tahan — Egyptian Research Collaborator; Alexandria University master's researcher. His role describes the contribution performed and does not imply university endorsement. • Faculty Process Reviewer — identity withheld pending permission; external methodological/process critique, not presented as journal peer review or institutional endorsement. • R1 Human Reviewer — case-specific human assessment of the bounded claim and frozen evidence. • R2 AI Reviewer — evidence-bound review of the same supplied material; AI output is not evidence or final authority. • Post-Peer Comparison — compares completed reviewer records, preserves disagreement, and applies a public result only to the exact claim and evidence version reviewed. 5. Standard Review Process Bounded claim Evidence gate Frozen packet R1 human Ai R2 evidence-only Ai Post peer G / Y / R / STOP Evidence is admitted independently of the desired result. R2 does not search outside the frozen packet, fill gaps from model memory, or treat model output as scholarly evidence. Missing essential evidence may produce STOP / Not Reviewed. 6. Success Criterion Success will be demonstrated if the same evidence-first process can be applied across multiple Egyptian cases while preserving claim boundaries, reviewer independence, contradictions, uncertainty, evidence versioning, and reproducible public records. 7. What Will Be Measured • Boundary and evidence control — the result stays limited to the exact claim, and reviewers use the same frozen evidence version. • Contradiction and uncertainty — conflicting or missing evidence remains visible. • Reviewer transparency — R1 and R2 remain separate, including disagreement. • AI restraint — R2 stays inside the packet and never functions as evidence. • Revisionability and efficiency — later evidence can update the record without erasing history, while standardized handoff reduces source drift. 8. Outputs and Next Validation Phase • Master Zenodo project protocol and status record. • EA1006 controlled baseline linked as an existing related output. • Nefretiabet and Rekhmire case records: locked dossiers, frozen sheets, reviewer records, and final case results. • Cross-case validation paper comparing performance under different evidence conditions. • Reusable protocol/documentation for later replication in Egyptology and other evidence-limited fields. The core tool and three-case structure already exist. Funding would strengthen independence and scale rather than build from scratch: specialist replication, collaborator time, additional cases, rights-cleared images, open DOI-backed publication, and cross-case analysis. 9. Research Integrity Boundaries • People provide expertise; evidence supports claims. AI-generated statements are not scholarly evidence. • No collaborator, reviewer, university, museum, or institution is presented as endorsing the Falsifiability Sheet unless separately documented. • The faculty process reviewer's identity remains withheld until permission is obtained. • The Nefretiabet and Rekhmire dossiers predate this master record; their findings are not retroactively described as prospectively preregistered. • Revisions are versioned, and third-party image rights remain separate from licenses applied to original project materials. The research trailRelated Works takes them directly to the method, control study, protocols, evidence dossiers, and eventually the final case results.