Comment by djbusby

A good test might include a bunch of domains. And checking the timing on each. Could we demonstrate the delay is on t.co and not on NYT?