Successful Web 2.0 applications are often those in which the value of the content of the service increases as users use the service and add more information. For example, a catalog of books might become more valuable as it acquires a better and better model of "People who bought X might also like Y" inferences.

Are such benefits unachievable in a world of Personal Web Services? Would users be isolated and the useful aggregates not exist?

In a world of Personal Web Services, users will be able to do a great deal on-line without being surveilled that, today, they can not do without being tracked.

Here is one simple example: Some Web 2.0 services are experimenting with the general theme of "presence" features: the system can inform a user's friends when they are online or what (within some parameters) they are doing. Apparently some users enjoy these features but they come with a devil's bargain: in exchange for allowing your friends to to "see you on-line" you are also selling that same data into an advertising market (at least!).

If those users, instead of renting the application were leasing Personal Web Servers, they wouldn't have to sell out: just each install a chat program and configure them to communicate with one another.

That shows, first, that not all losses of data aggregation are necessarily bad. Second, it leads to an explanation of how aggregation can work in a world of Personal Servers...

One concept that users are already beginning to understand quite well is the "feed" (as in an "RSS feed").

Another concept familiar to many is the "micro-format" (such as an eCard or an Open ID record).

Federated Data Aggregation is the idea that the bulk of a user's surveillable information on the web should, to the greatest degree practical, be engineered in such a way that the user can reliably control its aggregation by defining rules for generating "feeds of micro-formatted data".

Particularly critical here is that the micro-formats describe data that users have a chance of understanding easily, and that feeds are a natural and easy-to-use interface for specifying privacy policy.

When discussing user freedom, we imagined a group of friends sharing on-line "presence" in privacy, rather than under the eyes of advertisors. Suppose, though, that they do want to share at least some of their data with advertisors, perhaps in exchange for the advertisors paying some of the rent on their Personal Servers. The Federated Data Aggregation concept, privacy managed by feeds of easy to understand data, shows the way.