Worlds Hidden in Plain Sight
“But the way in which complex phenomena are hidden, beyond masking by space and time, is through nonlinearity, randomness, collective dynamics, hierarchy, and emergence—a deck of attributes that have proved ill-suited to our intuitive and augmented abilities to grasp and to comprehend.”
Worlds Hidden in Plain Sight
I’ve always had a curiosity for unexplainable phenomena per conventional facts or models about the world. The curiosity led me to reading Worlds Hidden in Plain Sight1.
The book is a collection of essays or papers written by research members associated with the “Santa Fe Institute”. All of them deal with different topics like biology, economics, politics, physics, ecology, art, etc. What they all have in common is their “complexity science” approach.
Complexity science doesn’t forego the scientific method. It has “science” in its name. Instead, complexity science aims to create digital twins or dynamic models over the experimental method. Complexity scientists build a model of phenomena based on preexisting assumptions. After model construction, they then use the standard scientific method in exploring the model.
Trendy Buzzword?
“And, to add a final touch of spice, all this hoopla often comes wrapped up in language vague enough to warm the heart of any continental philosopher.”
Worlds Hidden in Plain Sight
The words Casti referenced were “adaptive behavior”, “chaotic dynamics”, “parallel computation”, “self-organization”, and “emergence”. All of these have legible meaning, but their definitions are used without formalization. Hence, they “warm the hearts of continental philosophers”.
Learning How to Control Complex Systems
“To control a nonlinear dynamical system, an algorithm must embody a model for the system: that is, the way in which the algorithm processes information must mirror the way that the system processes information.”
Worlds Hidden in Plain Sight
Given the prevalence of modeling, what are the numerical methods for Complexity science? The most popular tool is “Agent-based Modeling”. Agent-based Modeling is a simulation of agents within an environment. As a scientist, you would encode assumptions about the agents and the environment an agent would interact with. Then, for every timestep, you make all the agents perform an action and update the state.
Another standard tooling is “differential equations.” This tool is most well-known in standard use. Differential equations can be used to model many systems and explore their behavior. For example, the weather.
“Genetic Algorithms” are the survival of the fittest, code editions. You write some hyperparameters for a particular agent and then simulate populations on a given task. The goal is to find the optimal hyperparameters. In each population, you slightly alter the hyperparameters for each spawn and only keep the top-performing ones.
-
The full book title is Worlds Hidden in Plain Sight: Worlds Hidden in Plain Sight: The Evolving Idea of Complexity at the Santa Fe Institute, 1984–2019. ↩