I'm not suggesting people go through and upvote every Less Wrong article; they can distribute their new votes however they want. So if that algorithm is correct, the impact would be for new people that sign up for Stumble Upon to sometimes vote for LW, or people that have dormant SU accounts that can log in and vote for Less Wrong. This is a bit different from an organized ring of people stumbling low quality content -- my intention is to create a culture of people on Less Wrong stumbling articles that they like.

Where did you get info about that algorithm? The last time I was actively stumbling things, for my old biology/genetics blog, it didn't seem to work that way. That algorithm doesn't quite make sense to me, as it implies that sites will start getting less and less traffic as they write more blog posts -- and with that old blog, stumbling new articles as they came up seemed to make a difference.