Peer to Peer Notes
1) There is a Peer to Peer Working group which got a lot of interest
   http://www.peer-to-peerwg.org
   Curiously to me this includes the "Internet Computing Companies"
   SETI@HOME (United Devices is Commercial version)
   Entropia Parabon
   These are all client Server

2) I think Napster is a neat model -- can think of it as NFS done with
   web technologies. Roughly equivalent to a search engine combined with
   a Web Server on every client (NFS is good old Network File System)

3) Napster is succesful (partly) because MP3 objects are self defining
   i.e. .mp3 file includes music plus metadata
   
4) GXOS is precisely what you need to extend Napster NFS model to any
   files/objects. Information in GXOS is metadata plus location(s) of
   object. Just what you need
   One cannot retrofit general files (e.g. .txt) with metadata. Thus GXOS
   approach (capture metadata in small "control" objects) is only way
   to broadly deploy Napster model
   We would have a nice wizard and users would register their objects
   in an easy fashion to build up powerful distributed object repository