A 'dense mesh' architecture (my term) intuitively feels to me like a promising way to address a whole lot of concerns.  What I mean by 'dense mesh' is:
  • as many transfer nodes as possible
  • and as many 'active' connections between the nodes as practical
  • connections of as many 'types' as possible.
Among the advantages I see such an architecture providing:
  • messaging flexibility
  • anonymity
  • dos resistance
  • resistance against massive resource and sophisticated network attacks (both denial based and actively subversive.)
I sensed some value in this when I noticed BitEN and even made an initial contribution of financial support although at that time the idea for 'paracoin' had not fully entered my awareness.  It seems to hold a lot of promise for something like the reference implementation certain parts of the software.

n-space characterization:

A central premise of 'para1' is to relate reward to the ability of participants to bring valued diversity to the table.  One of the most valuable aspect would be what I would call 'geo-political diversity.'  Put more simply we want a nice distribution of nodes at all edges of the network.  Geographic situation is, if not the most important, certainly high up there and one of the easier to gauge.  Because there may be more desirable diversities and because we wish to forge them into some managable metric, I'll consider them an n-space construct.

Every node will attempt to locate itself in n-space, and participate in assisting and validating other node's positioning in the same.  As always, we have to assume that nodes will cheat, and in fact that the cheaters may outnumber the honest actors.

What I see is a completely independent layer which is dedicated to dynamically maintaining a workable n-space map of active nodes.  This layer could happily function completely independent of any monetary frameworks, and will probably be the first to be built.  In some ways this is like the primitive heart and long functions wired into higher animals neurological wiring and reliant only on the brain stem 


Kaminisky's 'n00ter' ideas for gauging network level subversion might be useful in node-to-node testing.