Nothing is reduced to 0%. I'd also abstain from differentiating "fraud" and "error" as this is irrelevant to whether the paper is reproducible. What we care a

Nothing is reduced to 0%. I'd also abstain from differentiating "fraud" and "error" as this is irrelevant to whether the paper is reproducible. What we care about is true positives and negatives. But we'll also get false positives and negatives. So replication will have to happen at scale. Individual outcomes don't matter much.