Series:
Essay #4:
Synopsis:
Causation
Complexity
Complexity is kryptonite to material causation (mechanistic and probabilistic); complexity requires simple, holistic explanations
This essay is about causation in complex systems. Most everything around us is a complex system: our bodies and minds, our immune system, the environment, forests, grasslands, weather, the internet, economies, cities, ant colonies; also crayfish and armadillos.
What are the primary characteristics of a complex system? (1) A complex system contains large amounts of mutually interacting parts, where most everything is changing most everything else, all the time. (2) A complex system is dynamic because it evolves over time. (3) A complex system is non-linear, meaning that the whole is more than the sum of its parts, which in turn means that the inputs are not proportional to the outputs, for example, a small change in some variable may create an unexpected big change in the system. (4) A complex system is not deterministic, that is, we cannot determine its future states based on its initial or present state.
The most important features of a complex system are the inter-relationships and feedback of its constituent parts. With feedback, the output of a particular process in the system will recycle to become a new input. Feedback occurs throughout a complex system, willy-nilly, across levels of organization, local to global back to local. For example, when the Earth’s atmospheric temperature rises, evaporation increases, which causes an increase in atmospheric water vapor, which traps heat within the atmosphere due to the greenhouse effect, which causes additional rise in temperature (which was step #1 in the loop), and so on to global warming. That’s a classic feedback loop. Luckily, it’s only one among many other loops and relationships in our atmosphere that seem to dampen its amplifying effects, we hope.
Inter-connectivity permits small disturbances in the system to ripple out through the system causing a disproportionate effect (a non-linear effect). In her great book, Complexity, A Guided Tour, Melanie Mitchell calls it “sensitive dependence on initial conditions.” Complex systems show sensitive dependence on initial conditions, meaning that even a little uncertainty in the measurement of initial values can amplify to big errors in long-term prediction.
The first, best example of “sensitive dependence on initial conditions” is the problem of three-body gravity. In classical mechanics, this is the problem of taking the initial positions and velocities of three masses and solving for their subsequent motion according to Newton’s laws of motion and gravity. Think of the sun, Earth and moon as they orbit around and influence each other by gravitational forces. Newton solved the problem of two-body gravity, so you’d think three bodies is no problem. In the 1890’s, Henri Poincare tried and failed. He could not predict the long-term motions of three masses exerting gravitational forces on one another. His problem was this: whatever initial values he put in the problem, he’d reach a point where the calculation became impossible. Now, if in math equations, small variance up-front leads to big variance down the line, what chance do we have in real life? The three-body problem can’t be solved in the real world because the real world doesn’t give perfect input data, and even the slightest, itty-bittiest error in initial locations, masses and velocities will result in huge errors as we predict into the future.
Complexity is a mess. I’m reminded of Tolstoy’s ideas on historical causation, to wit, we know nothing. For example, to develop a causal theory about why one side won a particular battle in the Napoleonic wars, we might focus on Napoleon, and posit that his commands determined the outcome. Which is nonsense. Tolstoy tells us that in the buzzing, overwhelming confusion of war, most commands go unexecuted, and the few commands that find their way into the war usually produce results opposite to that intended. The war and its outcome are the product of the millions of people in the fight, and their horses, and their horses’ fodder, and everything else, and no one can parse out the causal chains in that buzzing maelstrom.
Another way of saying it is that, in a complex system, everything is connected to and depends on everything else, so their causal connections are in constant flux. We can’t isolate a single variable and put a number on it, because it’s changing while we look at it due to input from other variables; further, if we do manage to put a number on something, any slight inaccuracy in the number will lead to exponential inaccuracy in output, due to sensitive dependence on initial conditions.
Complexity overwhelms all materialist attempts at causal explanation, because there’s no physical chain to follow and no simple aggregate on which to derive a probability. So what type of causal theory should we use? Many people, all smarter than me, work on complexity full-time, and no one’s figured it out. My humble contribution is this: each complex system is individual and will require its own explanation. The explanation should be simple and speak to the whole of that particular system from a bird’s eye view. Darwin’s theory of natural selection is the best example, because it explains a complex process, the evolution of life, in terms as simple as, “inherited variations that increase the ability to survive and reproduce.”
I’m thinking about the type of causal theory that might work on a complex system. A marriage is a complex system, so let’s see what type of causal theory can explain outcomes in a marriage. A marriage consists of two sentient beings, each with an intimate understanding of the other, where each moment contains memories and feedback loops from a long history, plus expectations of a future together, preferably with grandkids. Gerd Gigerenzer in his book, Gut Feelings, offers a fun causal rule for marriage. Whether or not the rule is true, I believe its type, as a simple, high-level abstraction, is appropriate for a complex system. Gerd’s rule is modified game theory, and I paraphrase it loosely: act kind to your spouse on the expectation that your spouse will respond in kind. In most cases, your spouse will respond in kind, to which kind response you’ll respond kindly, and so on, building a virtuous cycle with mutual, positive habits of kindness. How wonderful!
Essays in this Series, Causation: