Gathering metadata is based on domain specific heuristics, since many unrelated types of data have to be processed. A software system should therefore be flexible enough to allow the addition and alteration of algorithms. This can be accomplished by refining the indexing engine presented in the tests. The initial metadata found can be extended by referencing publicly available databases. One example given was the google enabled plug-in. Further automatic refinements may come from crossreferencing items, as presumably used in the Audiogalaxy network. Allowing both manual alteration of metadata as found in KaZaA and moderation techniques as used on Slashdot can to some degree remove the errors created by the automatic process. Refining preliminary results will, however, only be of use when metadata can be made persistent across the network. So far, only the KaZaA and Morpheus applications employ such features.
To allow rapid interoperability between networks, it is imperative that already widespread communication standards are followed. Formatting data as XML or even RDF is a start. The use of a web-based client has a similar advantage over OS specific user interfaces. The lack of roaming metadata features severely reduces the strength of this option for now, unfortunately.