Mentat Logic

Thank you — this is exactly the kind of comment I hoped the piece would attract, and you're right on both counts. The ID-recycling history is a real hole in my "numeric IDs are authoritative" rule: for data reaching back far enough, authoritative within a reasonable window is the correct formulation, and it's going into the spec's known-limitations section — the whole point of that section is that limits get stated, not discovered. And rdid genuinely wasn't on my strip list (mibextid was, rdid wasn't — which rather proves your point about lists going stale). Adopting the quarterly refresh as standing practice; the spec repo will carry a "last reviewed" date so readers can judge freshness. Your rebrand war story is painfully familiar — an auto-merge that corrupts a trend line is worse than no data, because it looks like data.