Joshua Allen and Dare Obasanjo are discussing metacrap.  I think the scenarios Joshua mentioned are, at present, very unlikely, and I don't see much reason (he surely has not given any) to expect things will be much different in the future.

First - people need to have information capturing devices / applications.  Second, those devices / applications have to capture good metadata (time, date, position, etc.).  Third, that information needs to be related to the other data of the user.  Fourth that data needs to be associated with the data of other users.

I think it is a huge leap to even say that a large number of people will get to the first step.  I'm a relatively gadget getting geek, but I don't have a digicam yet.  I use Outlook, my cell, and my PDA like nobody's business, but those don't share data too well.  Yes, there are models that reasonably well, but mine are dumber models.  So, the first step is a huge leap...people need to have smart devices and applications.

The second is also a huge leap.  Many people have cell phones, but how many have smart phones?  How many cell phones will willingly share useage data, address books, call timers, etc. with others .  Even smart phones, so far as I'm aware, will only do address book syncing.

Presupposing that the first and second steps are met, we can depend on the holy grail of Longhorn's WinFS to automatically create relationships between the data (and metadata?) points.  Step four, well, I don't know of plans regarding this.  I suppose p2p networks are a good foundation...so, we'd have secure (?) personal data sharing via p2p networks?

I'm not sure that I buy any of this as close to feasible or conceivable in the next few years.