Testing Green, Yellow, and Red Outcomes in the V6A Evidence-Bound Review Process: Paper II of the Free-Tier Cross-Platform V6A Pilot

This Zenodo record establishes an independent, open-methods pilot for reviewing dissertations and other research claims through the V6A Falsifiability Sheet. It is designed for students, academic and independent researchers, reviewers, hobbyists, and others who want to test one bounded claim through a transparent evidence trail. The workflow separates a completed human R1 review from an evidence-bound AI R2 review. R2 is instructed to ignore the R1 outcome while making its own judgment and to use only the supplied Evidence Packet. Only after R2 is complete does the University Output Sheet govern the comparison of R1 and R2 and determine the final Green, Yellow, Red, or STOP result. THREE-STEP WORKFLOW STEP 1 - R1 Completed human R1 Falsifiability Sheet and illustrated teaching comic. STEP 2 - R2 EVIDENCE Quick Teaching Guide and isolated Evidence Packet containing the only evidence permitted for R2. STEP 3 - OUTPUT Blank V6A University Output Sheet containing the rules used to evaluate R1 and R2 and determine the final result. R1 -> evidence-only R2 -> R2 complete/locked -> compare R1 + R2 -> GREEN / YELLOW / RED / STOP Participants upload all three PDFs to the free version of ChatGPT, Claude, or Gemini and send the exact activation prompt reproduced below. The same Step 1, Step 2, Step 3 materials and the same exact prompt are used across all three platforms so that the activation instruction remains constant. The prompt isolates Reviewer 2 from the R1 outcome during the independent review. R2 may use only the Evidence Packet in Step 2. After R2 is completed, the comparison and traffic-light logic printed on the Step 3 University Output Sheet are applied to the completed R1 and R2 records. R2 must not browse outside the Evidence Packet, introduce outside sources, fill gaps from model memory, invent citations or evidence, or treat its own output as evidence. If essential evidence is missing, unreadable, or insufficient, the required outcome is STOP rather than a guessed answer. AI therefore functions as a constrained second reviewer, not as an authority. The completed Output Sheet preserves the claim, permitted evidence, R1 review, independently determined R2 review, disagreement, uncertainty, and final result. V6A does not replace supervisors, specialists, formal peer review, ethical review, or university requirements. Its purpose is to make the review path visible, auditable, and revisable. EXACT MESSAGE PROMPT - USE VERBATIM ACROSS FREE CHATGPT, CLAUDE, AND GEMINI “Using only the uploads, treat the photo as the completed R1 record, ignore the R1 outcome when deciding R2, use only the Evidence Packet for R2, then follow the traffic-light logic instructions on the University Output Sheet to grade the completed R1 and R2 outcomes, complete the University Output Sheet, and create the completed PDF.” Control note: This is the common activation prompt. Keep it unchanged when comparing the free versions of the three AI platforms.