Package a result for independent review
Connect the published conclusion to inputs, procedure and raw outcomes.
Research note · November 14, 2026Research question
What should a reviewer receive to assess a research result without relying on a screenshot?
Method
- Package a claim statement, input manifest, procedure, environment description and raw outputs. Include the scripts or explicit steps used to derive each table and figure.
- Identify withheld material and explain how its absence limits reproduction. Keep secret removal separate from undocumented editing of unfavorable outcomes.
- Ask a reviewer to reconstruct one reported quantity from the package. Record missing dependencies and ambiguous instructions as review findings rather than silently fixing them after the fact.
Records to retain
- Claim and scope statement
- Versioned inputs and procedure
- Raw outputs and derivation steps
- Known omissions and review findings
Source basis
W3C PROV and SLSA provenance provide complementary ideas for recording derivation and execution inputs. PyTorch’s reproducibility guidance also makes clear that environment details can matter even when source code and seeds are available.
Interpretation & limits
A reproducible package does not independently validate its underlying claim or data quality. This note describes a publication protocol and does not imply third-party review of Vraelis results.
Next experiments
- Prepare one small package whose figure can be regenerated.
- Track reviewer-discovered omissions in the next package revision.
References
Technical reference
Implementation details supporting this note, drawn directly from the maintained documentation. These are technical references, not additional papers or new experimental results. Expand a topic to inspect its method, fields and limits.