tedster - 5:31 am on Sep 6, 2012 (gmt 0)
Just think of all the overhead it would take to recalculate PageRank
As I understand it real PageRank [not toolbar PageRank] is already recalculated continually - daily for most pages on most sites. It's been that way for several years, even before the Caffeine infrastructure went live. Once the math for a change in the PR formula is generated, the job is essentially done and the overhead would be minor, IMO.
I don't want to go far off-topic for this thread, however. We're talking about Penguin recoveries and I really doubt that PageRank is a factor there.
One factor that got some discussion for Panda but almost none for Penguin is the keywords the decide which sites get their Penguin factors calculated. With a machine-learning process, it has to be a subset based chosen in some way.
For example, with Panda version 2.0 was reported by Amit Singhal as going further down the long tale. For Penguin, which is reportedly looking for rankings that are "artificially" boosted, we've got to be talking about using a certain selection of keyowrd rankings to choose where the calculations are made.