Harder Drive: Hard drives we didn't want or need

Hilbert curves (or similar) are often used for graphing billions of nodes[1]. However this will not by default show the relationships between nodes in a graph. Depending on your data you may be able to write a function to map from your edge list to a node index that hints at proximity. Note that visualizations are limited by human perception to ~10000 elements, more usefully 1000 elements. You might try a force directed graph, perhaps a hierarchical variant wherein nodes can contain sub-graphs. Unless you have obvious root nodes, this variant would be interesting in that the user could start from an arbitrary set of nodes, giving different insights depending on their starting point. 1 - An excerpt from "Harder Drive", a rather silly implementation of a unix block device using ping latency with any host that will let him. He visualizes the full ipv4 address space in a hilbert curve at this offset: [this video]