Have heard a theory which sounds plausible to me, but wanted to run it past you lot.
It's this: in order to target content farms and work out their "true" authority, G dialled down the emphasis on internal linking. Therefore sites with good external links to every single page are doing great, but sites that depended on sheer size and internal linking to rank, have tanked.
What makes me think this is plausible is the difference in the hit to Hubpages and Squidoo, which several people have posted. To me, these seem like identical types of site, so I would have thought they would have both dropped by the same proportion. But according to some figures hubpages has dropped by an average 30 positions and Squidoo by just 15 positions, and Quantcast is also showing a 40%+ drop in Hubpages traffic, but circa 15% drop in Squidoo's.
The only real differences between the two sites is the way they are structured. Hubpages was designed by IT folk with brilliant internal linking, and Squidoo was designed by PR people with abysmal internal linking. So Squidoo may have been ranking mainly on external links - and they've dropped less because of this (i.e. they didn't have much of an internal linking advantage to start with).
This might also explain why Mahalo dropped so significantly - nobody links to them naturally, do they, and I doubt they were able to build a bunch of external links themselves either, especially as they scrape and have no users who are looking after their own pages as it were.
It might also explain why Ehow is still ranking - all those writers on ehow, before Demand Media closed it to all comers, used to build links to their own pages, so I imagine each page is pretty well supported externally.
What do people think? Is this way off mark, or is this plausible?