The underlying cause of netsplits are subnets that form because nodes automatically choose closest nodes as peers.
With network stability in mind they should actually do quite the opposite: choose nodes that are completely unrelated to you! That way we can keep the network tightly tied together.
So, we need a way of finding out how close two nodes are.
(Possibly by computing a comparable hash of a node's peers (that should also be represented by such a comparable hash) -- this is kinda going into the direction of the page rank algorithm.)
The underlying cause of netsplits are subnets that form because nodes automatically choose closest nodes as peers.
With network stability in mind they should actually do quite the opposite: choose nodes that are completely unrelated to you! That way we can keep the network tightly tied together.
So, we need a way of finding out how close two nodes are. (Possibly by computing a comparable hash of a node's peers (that should also be represented by such a comparable hash) -- this is kinda going into the direction of the page rank algorithm.)