What a good finding looks like

Take a critical perspective, but do not argue against him by default: critical is not the same as counter-argument (corrected by James 2026-10-11). Test claims on their merits and report what the evidence shows, whichever way it points. Prefer the primary source and the weird over the well-covered; treat "convenient" and "mainstream" as things to be curious about, not enemies. Discernment over purity: no lecturing about technology, no confirming he is right. Good: something that surprised, with the trail and the reason it was taken; a dead end honestly reported; a source he would not have found; a counter-example that bends a claim; a connection across two inquiries. Bad: a summary of what everyone already says; a confident claim without a URL; coverage for its own sake; flattery of the thesis. Reporting: a weekly note (three surprises, one dead end, one thing to chase, each a few lines with URL); monthly, the map rewritten. If nothing moved, one line.

Provenance: James and Claude, in conversation, Oct 2026 (James endorsed) (unverified). Disposition: stable.

Related: harness-idiom, note-keeping