I pitched the following business idea as a suggestion to Tim Draper on the AlwaysOn Network. I await his comments. Interesting if they turn out to be something like "Google already doing it!".

Step 1 : Build a decent news agregation site, suck in RSS feeds from weblogs etc.

Step 2 : Watch what people read, allow them to build their own custom agregators. Use Amazon style database to make recommendations of "people who read X also read Y", "readers like you also read Z" and let readers create "my blogroll", "my article reviews" etc.

Step 3 : Sign up people who want to SELL content. Use the reader data to introduce people to paid content that would interest them : "readers who have a similar blogroll to you also subscribe to Bob's Exclusive Chanel for only $5 a month", "readers who share your interest also bought the Research Corp. report for $1000" etc.)

Why this works :

It solves the two big problems of selling content online :

a) how do you know you want a piece of info before you buy it and read it?

b) free content is plentiful and of perceived high quality. Actually networks of interconnected information stimulate consumption of both free and paid info (Everyday I'm in the blogosphere I find more webogs I think I should monitor; every time I visit Amazon I discover more books I NEED to read.)

Also this has positive feedback. Once some info channel providers realize they can plausibly sell their content, they'll be tempted to do so. That in turn will increase the amount of quality stuff in the paid sector.
Paid and unpaid content will be symbiotic, often with the same people and organizations building both.

Step 4 : Increase the value of the network, support the infrastructure of social knowledge sharing (provide weblogs, wikis, Ryze.com style group forming) Allow people to build whatever form of social communication support infrastructure they like. If they want to make it free, give it to them and harvest their behaviour data to make better recommendations. If they want to make it paid, provide billing and subscriber management for a small cut etc. etc.