chrism - 12:18 pm on Oct 3, 2011 (gmt 0)

Why, then, do we seem to have the largest number of recovery reports at the time of each update?

My belief is that the answer to this is sites are released in bulk because it is the measurement criteria that change with a Panda run, not that sites that are scored against those criteria at the point it is run. If I am correct, this explains why some sites can recover if they address why they do not meet the current criteria between runs of Panda, but sites (even if they have not changed) can escape when the criteria change with each run.