Sociologically Inspired Engineering
By: David Hales and Bruce
Date: 14 July 2004.
CPM Report No.: 04-136.
An article written for the AgentLink (III) newsletter.
It's always hard to predict the kinds
of cross fertilisations that will occur between different areas within
the interdisciplinary hot-house of agent-based research. It is now
widely established that biologically-inspired mechanisms are applicable
as basis for the engineering of distributed and robust systems. For
example, many adaptive agent algorithms draw inspiration from
biological evolution and there has been much recent work on
ant-inspired routing methods to name just two. Here we report on recent
developments in which mechanisms derived from social theories,
particularly computationally expressed mechanisms through agent-based
social simulation, are being applied to tough engineering problems in
The idea of using “social metaphors” for thinking about self-organising
software is not new. Marvin Minsky's classic A.I. Text “Society
Of Mind” (Minsky 1988) explicitly envisages minds as composed of
semi-autonomous entities with coalitions, conflicts and hierarchies.
What is new is the application of techniques from multi-agent-based
social simulation (MABS) to the hard distributed engineering problems
that agent researchers are interested in today.
For well over a decade simulation-friendly social scientists have been
using multi-agent-based simulations to develop, test and communicate
mechanisms of social emergence. This means programming individual
agents with (often sophisticated) individual behavioural and learning
rules and then executing simulations to determine what kinds of
societies and structures emerge. The focus of much of this research has
been on trying to gain a greater understanding of human societies – the
world we actually live in. However, some of the mechanisms that have
been discovered are directly relevant to open engineering issues in