Two-Person Review Exchange

Work independently first. Then compare, reconcile and document disagreements. This is a human-operated, no-account exercise—not simulated agreement between two AI personas.

Scope: Six real trial-arm records from metadat::dat.egger2001, selected from 16 historical studies. Counts are genuine archival teaching records, but eligibility cannot be confirmed without primary publications. Reviewer names are self-declared; the app cannot establish that two distinct people took part.

How a genuinely independent session works

  1. Send Reviewer A and Reviewer B the link and the same prespecified screening question. Each person uses their own device/browser and completes the form privately.
  2. Each reviewer exports a JSON decision pack, then sends it to a human adjudicator by their chosen communication channel. Avoid sharing one person's decisions before the other finishes.
  3. The adjudicator imports both completed packs together; the page reveals agreements/disagreements and allows final decisions with documented reasons. Export a local Markdown report.

The form contains no patient data and makes no network uploads. The files contain human-authored decisions and labels: treat them as research working documents. JSON is not signed, reviewers are not authenticated, and an independence declaration is not proof of independence.

Part 1 · Independent reviewer

Complete your own private-before-sharing review pack

These are real archive-row labels and event counts; no default inclusion verdict is implied. “Unclear” is appropriate where primary reports must be inspected.

Part 2 · Human adjudicator

Import both independent packs simultaneously

Keep each person's answers hidden from the other until both have exported. Choose two independent files; this page does not retrieve or transmit files.

No reviewer files imported yet.

What comes next?

The six-stage Research Studio lets learners investigate archived magnesium trial events, model risk differences using real AllMeta tools and write a capstone. This independent-review exchange is a separate workflow for two humans, not automatic evidence that either screening decision is scientifically correct.

English pilot; all 12 established Synthesis lesson language routes are retained. Independent methodological, accessibility and translation review remains outstanding.