Use a p2p network as a JavaSpaces/Linda - style tuple space.
Create data structures as (possibly metadata-rich) xml packets referring to resource files/urls, and to other xml packets. The data structures from the JavaSpaces book can easily be modelled this way. A weblog or serial could be an open-ended queue, an album, lecture series or split large file could be an array. Use MD5 hashes, domain limits, time limits or other constraints for “security”.
Use the entire internet as a tuple space described in xml packets. Extend the format to allow specifying sub-elements of resources and assemblages of elements/sub-elements for remixes and other derivations.