Over the last few days it's dawned on me that my understanding of the 'running of panda' and 'panda dates' seems to be very different to other people's.
My feeling is that when dates occur where 'Panda 2.1' or '2.2' or '2.3' runs, what happens is the factors of measurement change, and this is what 'running Panda' means. In other words, the indicators themselves are 'regenerated'. I believe it is for this reason that we see bulk cases of rises or hits due to Panda on these dates, as the rules are changing, not the sites in the SERPS.
What this also means is sites can escape Panda between 'runs', but only if they specifically address the indicators that are suppressing their rankings. The fact that a wide variety of factors may be involved in suppressing any one site may mean that changing just one aspect of a site may tip it back to 'normal' or improved rankings, or it may require many subtle changes.
As a result of people chasing statements like 'remove ads above the fold' I think it may even be possible that Panda 1 may have heavily weighted advertising, but subsequent versions have a much lower weighting relating to advertising because a significant number of sites initially flagged by Panda have 'improved' their site (in Panda's eyes) so in the greater scheme of things it's a less important indicator (but still an indicator nonetheless).
Of course, all of this is just my feeling as to what is happening, but I really don't buy into the idea that google wants to suppress sites indefinitely until the next 'run' of Panda even if they have addressed the issues flagged.