Emergent Group Level Selection in a Peer-to-Peer Network
David A. Hales · Complexus · 2006
Many peer-to-peer (P2P) applications benefit from node specialization: for example, the use of supernodes, the semantic clustering of media files or the distribution of different computing tasks among nodes. We describe simulation experiments with a simple selfish re-wiring protocol (SLAC) that can spontaneously self-organize networks into internally specialized groups (or ‘tribes’). Peers within the tribes pool their specialisms, sharing tasks and working altruistically as a team – or ‘tribe’ – even though their individual behaviour is selfish. This approach is scalable, robust and self-organizing. These results have implications and applications in many disciplines and areas beyond P2P systems.