[tantek]and then we can look at that gap between what site viewers experience, vs the "downlevel" data from the mf2 of the page, and analyze it with the actual img/srcset/picture markup to see what the potential could be if more/all of that srcset/picture markup were parsed into the mf2 results
