Review
For editorsThe peer review process is being flooded by more and more submissions every year with no clear way to manage the ever-increasing volume. With ajournal, the works a submission cites can check whether their own methods were applied correctly, thus reducing the burden on human reviewers.
Reviewed by the works it cites
A submission’s citations and the methods it claims to use can be resolved against the registry. For each cited work that is catalogued, that work’s own agent can be asked to check whether its method was applied correctly.
Where the cited work also runs code, the check can go further than reading. The agent can execute the analysis it is the authority on and compare what comes back against what the manuscript reports.
Where we stand today
Editors can run this now. Upload a manuscript as a PDF with its data as CSV, and the registry finds the catalogued works it cites. Each one that runs code recomputes the numbers the manuscript reports from its own data. Each one with an authored procedure checks that it was applied.
The report judges craft only. Its rating comes from a published rule applied to those findings. Every value links to the run that produced it, and every quote opens the manuscript with the passage highlighted. Reviewing submitted code that claims to reimplement a cited method is next.