Randy pretty much always says everything I want to say, before I get the chance to say it. And he's done it again.
What I would add is that the article states:
Think of it this way: a link from a highly trusted web site will still be a highly valuable thing. Similarly, a host of mentions across a range of social media sites, particularly if they are primarily positive or negative, will also be a strong signal.
I think things need to be more granular than that. Just as search engines currently rank quality of Web pages (Pagerank), they may also need to end up ranking quality of members and posts within social networks. For example, within Twitter, they could look at
- who the member follows
- who follows the member
- the quality of the neighbourhoods the member links to
- the quality of the neighbourhoods the members a member is following link to (and so on, recursively)
- the "signal:noise" ratio of a member's posts
and so on. These could give "MemberRank" and "PostRank" equivalents to PageRank.
This may be a lot of work (especially tracking all those tinyURL links
), and will involve quite a deep understanding of the inner machine of a social network. Actually owning the network, or forming a close partnership with it, may be required in order to gain that understanding.