I'm no expert, but if both Chicago and New York are synching their clocks to NIST(Gaithersburg, MD) or the Naval Observatory (Washington DC), the a message from Washington sent at the same instant, should arrive at the same instant, as measured in local time.

A more 'science' way to make the claim that Nanex is making, would be to compare the same products on other Fed meeting dates.

Any time synchronization protocol I know of takes into account network latency and attempts to compensate. As measured by perfectly synchronized clocks against the same reference source (leaving aside for the moment that there is no such thing), the received timestamps of a message originating from one location and being received by two locations at different lengths of cable from another will reflect the difference in the time it took to get from one to the other.

Real market data uses PTP driven by GPS calibrated clocks.

Correct, most brokers use PTP also.

Local drifts of multiple millis can be excluded. Millis are huge.

I'm also curious about how they prove this. The article assumes the times were in sync.