Topics · Domain workflow
Reviewers recompute digests on later submissions and verify they match signed commitments from the original workflow — seeing exactly where the replication diverges.
Concrete scenario
A replication team requests the exact instrument readings and preprocessing scripts for a five-year-old paper. The journal portal link is dead; the PI has a ZIP export but no independent proof it matches what was analyzed originally in the lab pipeline.
Problem
Reproducibility depends on datasets, transforms, and instrument outputs that outlive portals, grants, and lab IT rotations. Publications without signed evidence trails age into unverifiable claims that cannot be challenged with mathematics.
Mechanism
Experiment steps mint receipts over dataset roots, instrument outputs, preprocessing transforms, and publication artifacts at each accountable stage. Evidence Graph helps later researchers discover related receipts when they arrive with partial material or incomplete exports.
Verifiable outcome
Scope boundary
Signed evidence trails strengthen reproducibility claims — not peer-review verdicts, statistical significance alone, or automatic replication success without independent analysis.