Specific technical criticisms of experiments and reasoning are valuable and should be carefully considered because they sharpen one's craft, while meta-level advice about what to think, what not to think, what to talk about, or what not to talk about should be treated more skeptically because no one has perfect calibration across domains.