Consensus Web Filters
Kevin Kelly explains at http://www.kk.org/cooltools/archives/001163.php
"Like a lot of people, I find that the web is becoming my main source of news. Some of the sites I read are published by individuals, but I find the most informative sites are those published by groups of writers/editors/correspondents, including those put out by Main Steam Media (MSM). However for the past three months my main source of "what's new" has been a new breed of website that collaboratively votes on the best links.
This genre does not have an official name yet, but each of these sites supplies readers with pointers to news items that are ranked by other readers. None of these sites generates news; they only point to it by filtering the links to newsy items. Using different formulas they rank an ever moving list of links on the web. The velocity of their lists varies by site, but some will have a 100% turnover in a few days. I check them daily.
This new genre fits into a whitespace between already occupied niches of social web sites. In the established center are the group-produced sites such as Slashdot, BoingBoing, WorldChanging, Huffington Post, to name just four very popular ones, where a very small cast of editors (under a dozen or so) collaboratively filter and annotate the links to other sources. A daring and effective extension of this method was devised by the fantastic group at MetaFilter. Here the editors are a very smart mob of 25,000 users. One by one readers recommend the cool new stuff they find. Their filter is simply the emergent one of their collective discretion and taste; no one votes or ranks links. At the other end of the axis of collaborative filtering is the likes of Google and Yahoo News, which use the entire collaboration inherent in the web and many Googleish algorithms to programmatically generate a list of what's new based on who is linking stuff, the most "important" item at the top. No humans explicitly vote on the items.
These new uncategorized sites, which have emerged this year (and reviewed below), fall in between the positions above. They take the smart mob approach of MetaFilter and add the algorithms of search engines. So, readers themselves vote on the importance of linked items suggested by other readers; these votes are then subjected to a complex formula to produce rankings. The sites use various flavors of algorithms to balance and refine the votes and selection of smart mobs. Or they use the action of tagging or bookmarking a site as a type of vote. Each site uses a different algorithm, yielding slightly different mixes of links, and a different personality. The best sites maintain a balance between providing a sense of what everyone is reading (consensus popularity) and some novel items that not everyone is reading (yet)."
See the bottom of the same post by Kevin Kelly, at http://www.kk.org/cooltools/archives/001163.php
Or the review of related Memediggers by Pete Cashmore of Mashable.