I realized that we need an additional probe that will allow us to answer the question "which ratio of all pages that contain Flash at all only contain SWFs we would identify as ads". I propose a probe with four values: 0: page doesn't contain any Flash content 1: page only contains ad Flash content 2: page contains ad and non-ad Flash content 3: page contains only non-ad Flash content I'm not sure what the state of our heuristics is, but for the moment let's go with just dimensions-based detection - over-counting ads isn't the end of the world if it's not done by a huge margin. Yury, do we have a list of common ad dimensions we could use for this?

Bug 1085044 already has a patch.