R. Research-Paper Reading Checklist

Printable — one per paper (Ch. 52 distilled). Before reading: why am I reading this — target claim or survey? Pass 1 (10 min): title/abstract → one-sentence claim ("X, demonstrated on Y, vs Z, with statistical strength W"); figures' axes and ranges; conclusion vs. abstract — consistent? Pass 2 (40 min): contribution sentence (delete-from-history test); method extracted; baselines named and judged honest (equal tuning? classical competitor?); regime: theory / ideal sim / noisy sim (which model?) / hardware (which, when — calibration age?); statistical standards: error bars? seeds? pre-registration? Pass 3 (if it matters): math instantiated at n=2; algorithm sketched in code/simulator; supplementary methods + limitations read twice. Red flags: missing baselines; "up to N"; best-of-k as mean; log-axis hiding asymptotics; undefined "quantum advantage"; QRAM assumed silently. Log entry (mandatory): claim; contribution; evidence grade (A: reproduced by you; B: verified vs. independent source; C: trusted on reputation; D: unverified); open questions it leaves; relation to your claims matrix. Follow-up? → Ch. 53 (reproduce), Ch. 55.2 (gap), or done.