Replication Review¶
Students inspect a public research artifact or implementation and answer:
- Can the published result be rebuilt?
- Are exact subject-system versions available?
- Are preprocessing and exclusion rules explicit?
- Are random seeds and configurations recorded?
- Can reported tables/plots be regenerated?
- Which result is most sensitive to an undocumented choice?
- What minimum patch would improve reproducibility?
A replication review is a technical audit, not a complaint that software is difficult to run.