'The Architecture of Complexity' by Herbert A Simon

Reviewed by Pranav Minasandra

27 May 2023

It’s surprising how much connects the dynamics and the statics of all kinds of real world systems. For years now, we have been trying our best to get at why—why do cities, asteroids, and animal groups have the same patterns in the distributions of their sizes? Why do the intervals between earthquakes, neuronal avalanches, and behavioural transitions all follow the same distributions as well? What connects general, universal phenomena like regime shifts, chaos, and synchronisation?

The story of the study of complex systems is the story of science itself. We take knowledge gained from several sources, we tie it all together, and we try to create understanding. I feel we are, right now, in the ‘describing things’ part of science. We first need to describe things properly before we can understand why those things are the way they are.

Herbert Simon’s The Architecture of Complexity is an old book chapter published in 1967, only years after the structure of DNA was discovered, and before man’s landing on the moon. It is astonishing how much this field had already advanced back then. This is just the feeling I got when someone told me the back-propagation algorithm for optimising neural networks was already a thing fifty years ago.

Simon talks about a variety of things in Architecture, but the limelight is given to hierarchy. Most systems in science, such as societies, biological entities, materials, and in a sense, the universe itself, are all hierarchical. A hierarchical system, the way Simon defines it, is something that can recursively be decomposed into groups of entities, wherein interactions within each group exceed the interactions among groups. Such systems, pervasive in nature, have unique properties that Simon begins to explore.

For instance, hierarchical systems are resilient. Human beings don’t fall apart when you disconnect a few atoms from each other. This stability comes from multiple layer’s of fail-safes: A few atoms falling apart might cause a cell to fail, a few tissues need to fail for the organ to malfunction, and lots of terrible things need to happen for a person, or a machine, or a society, or an empire, to cease to function. Simon argues that complex systems tend to be hierarchical. He also explores the link between how much a system can be decomposed into constituents, and its comprehensibility. Standing somewhere between science and philosophy, this 20-page chapter is an easy, mind-blowing read.

I, like many, am tired of the relentless jargon of academic writing. If I want to say that $A$ doesn’t make sense, I need to write “While efforts have been made to understand the dynamics [36] and drivers [37-40] of $A$, literature on explaining $A$ causally are sparse” (or something else that makes me sound like an old rich man in an armchair). In this sea of pretentious writing, Simon’s essay, full of delightful parables, fun thought experiments, and interesting analogies is, maybe not an island, but a beach? Or would it be a grain of sand on that beach? Or an atom of silicon within that grain?…