I discovered distributed dictionaries and had the idea of using multiple levels.

the bottom level keeps track of 256 first level nodes, each of which keeps track of 256 nodes which on the second level keep track of 256 nodes which on the third level keep track of  nodes... and so on.

you only need 8 levels and the system can keep track of 2^64 nodes and each node only keeps track of 256*8 nodes 

thou for message efferently reason I use 128 nodes in each layer above the first so nodes need to keep track of a minimum of 256 + (128*8) nodes

when looking up a node or forwards it to the node with cover that range of the target's address at the highest level.

This allows the system to take advantage of nodes with greater memory as it can keep track of more node and so it routs in fewer hops.

I'm still working on the latter networking, where nodes keep track of other nodes that are covering the same region of a level