I've been giving thought to indicators that would identify quality. Life cycle or long term trends seem to be a method that could be used. Understanding how a sites life cycle may appear to an algorithm or plot on a graph could show a defined and measurable difference between a quality and low quality site. Google could manually rate sites within a classification, and see how differ for a given category.
Many quality indicators are often discussed as points in time. Making a change and waiting to see how Google reacts implies that Google Panda is looking at point in time. I don't know if Google views a site quality as a measure that can be turned on or off in Panda. Essentially, Google should have enough data over time to understand how a site was given birth, and how it evolved. Does it have similarities to the quality site or the low quality site?
So what I'm getting at is that, Google may be taking many indicators and not just using them at a snapshot point in time, but applying an algorithm to a site in order to grade it over its life, or as far as they may have data. This may indicate why some sites that may appear similar to others, but suffer from Panda and seemingly can't recover. They may have long term differences that appear a certain way to Google. A life cycle also implies a longer period of time, and sufficient reason that Google would not need to constantly reevaluate, unless they made algorithmic changes to the Panda filter.
We don't know all the indicators that Google is measuring, but there are many. Like trending performance of a server, a single current snapshot does not mean much, but a trend of changes over time tells the story. Google may be using long term trends as a quality measure in Panda. I doubt you can game them or "recover" by making a quick change if that is the case.
