The main ideas are as follows:

 - Pipe dream for many networking researchers - provide ubiquitous connectivity.

 - There is a wide scale proliferation of mobile handheld devices which come equipped with multiple comm. interfaces and largely unused local bandwidth, fast CPUs, good battery  life, large storage. All these resources are currently untapped.

- Therefore, tap these unused resources to address the problem of ubiquitous network  connectivity, albeit only for apps which can tolerate delays.  

More details can be found shameless plug.

But the key question is - why would anyone want to use this? If one can have blazingly fast 3G 

everywhere  as well as entire cities being blanketed by wi-fi, then what is the point of relying on devices to move your messages across? 

While some answers do exist for this question, after some discussion with people, it would appear the real reason for such an architecture to be useful can be simply : the flat-pricing scheme employed by telcos everywhere.

Flat-pricing means telcos will have more incentive to look at alternate channels to distribute content, specially architectures which would largely ensure that traffic stays local.  Telcos clearly would not want to carry more traffic than necessary.  PSNs can help here me thinks.