Thinking Recursively

Randomly Growing Network

Wm Jess Taylor -

Note: This article was written while conducting research for the Polis Foundation and was first published on their blog here.

This is a video of a randomly growing network using a variation of the preferrential attachment network generation model.

  • The network begins with 5 connected agents and adds 3 agents per time step for 208 steps (626 final agents).

  • Each newly introduced agent adds 2 new attachments randomly selected using the preferrential attachment approach, which weights an attachment towards agents that have more connections.

  • At each time step 2% of existing agents, selected randomly, also add 2 new random attachments using prefferential attachment.

  • Agent size (cirlce diameter) and color is proportional to the number of connections each agent possesses.

There is no strong motivation for the choice of these parameters - but it makes for a pretty video!