> > > Yeah, but with a fully connected graph, you'd need a far greater number of
> > > hops to locate a document than in a small-world graph. Too large for
> > > Freenet to be effective.
> >
> > No, with a fully connected graph you are one hop away from the
> > "epicenter". A fully connected graph is actually ideal in terms of
> > efficiency because the local maxima for closeness is also the global
> > maxima for closeness.
> Sorry, I thought you were talking about the entire Freenet, not a cluster.
In the entire Freenet, path compression works exactly the same except that
nodes in a cluster can only be compression to 2 hops away from nodes
outside of the cluster, instead of 1 hop like public nodes.
> > A clusters don't eliminate small world graphs for the public network. They
> > only mandate that a node in the public network can be a minimum of 2 hops
> > from any node in a cluster instead of 1.
> Which may cause problems for neighboring clusters.
If 99% of users run public nodes then there will be few neighboring
clusters.
_______________________________________________
Freenet-dev mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/mailman/listinfo/freenet-dev
