Sorry it took me so long, I have been doing other things like trying
to make a living.
I have written up a problem statement for xml magnet in an attempt to
clarify the goals for the project overall.
I have already thought up a few ways to salve part of the problem and
may soon make a few posts to talk about them with coders and the users
that they would affect after I do some more research.

Once there is a consensus on how to solve the major issues I'll write
the draft and do some testing and hope it'll be worthy of an official
RFC, or do the process over again.

http://1337ware.kicks-ass.net/dev/articles/xml-magnet-problem-statement.txt

The solution I'm working on includes a new way to locate applications
that can resolve magnet URIs and then what will likely be xml/soap
communication over tcp http. I am currently in the middle of
researching 2 local host centric and 2 local network centric ways to
locate applications that can resolve magnet URIs and I have an idea
for an xml namespace that can carry information about entities
described by magnet URIs resolve magnet hyperlink content and be
extensible to allow many different kinds of user interfaces.