2012/09/02

Chaos and Constitutions

One of the most important insights from chaos theory is that complex systems of interacting actors can exhibit patterns of behavior that seem designed but are not, the details of which are the result of self-organization, and which are not predictable in principle. Patterns like the spots on a leopard, the stripes on a zebra, human fingerprints, a beating heart or intelligent brain, the movements of flocks of birds, schools of fish, or termites building a castle of mud, are all broadly constrained in general by genes, but genes without sufficient information, by typically 11 or more orders of magnitude, to specify the details of swarm or herd behavior by elementary actors of limited intelligence, each responding to its local environment using fairly simple rules.

These insights have implications for public policy and constitutional design. Too often public discourse makes a presumption of determinacy, as though society, the economic system, and our system of government and laws, are fundamentally mechanical, and can be modeled and managed in principle using sufficiently complicated simulations. That is the systems dynamics approach that produced the classic paper, The Counterintuitive Behavior of Social Systems, by Jay Forrester (1970). However, it is not just that human mental models are inadequate. Our computer models, while they might improve our odds of making wise decisions a little, they cannot in principle guarantee desired outcomes. The underlying systems are chaotic, subject like the weather to butterfly effects, and a policy intervention that might have no appreciable result if applied today, can yield mass extinction if applied the day after, and heaven on Earth if applied the day after that.

Electoral candidates and lawmakers win support with appeals to "support me and everything will get better". It is an open question how many people are fooled by such appeals, but it only takes a few to swing an election or a vote on the floor of a legislative body. Most such decisions are made emotionally, not rationally, and we could probably get results as good through a lottery. We must bear in mind that humans evolved for making decisions in a paleolithic hunter-gatherer society grouped into small tribes or villages, not a modern, global, technological civilization. We are barely adequate to live together in a ward republic, much less a global urban world-state.

This has implications for constitutional design. Just as undifferentiated human stem cells might form a heart if they find themselves in a local environment that provokes them to "form a heart here", and the same cells would form a liver or a lung if placed somewhere else, so human individuals can be organized into structures that constrain them to function in a way that serves the health of the system as a whole. Constitutions and laws are attempts to provide such structure, but there are limits to the adequacy of human intelligence, even at its best, to design political and economic systems with an expectation of intended behavior. Nature shows us that such structural designs are generally the result of evolutionary trial-and-error, not intelligent design. When we find a design that works fairly well, it is best to avoid large departures from it, and to stick to small, incremental changes, with some time to observe the consequences.

One of the things we can conclude with some confidence is that humans cannot manage large organizations, even in principle. The best they can do is organize themselves into many small organizations that can then interact in what might be called a marketplace, for which occasional disasters may be unavoidable, but which have enough resiliency to allow some to survive.

The foundations of indeterminacy

The Newtonian view of the physical Universe was as a clockwork, perhaps complex but fundamentally deterministic, in the sense that if one had complete information about the behavior of its components, and sufficient computational power, one could in principle predict its behavior in detail as far into the future as one might wish.

That view was shattered by the emergence in the 20th century of quantum mechanics, which fit observation well but consisted of wave functions that could only be interpreted as probabilities, not as deterministic causation, and were not local but spread over the entire Universe. Some, like Einstein, could not accept this view. He said "God does not play dice with the Universe." They sought "hidden variables" that while they might be forever beyond reach of measurement, would at least explain the Universe as a deterministic system. Further work on this question, however, seems to establish that an underlying determinacy is not consistent with empirical observation. The Universe really is fundamentally probabilistic and indeterminate. If you could run it multiple times from the same initial conditions, it would turn out differently every time.

Yet many large-scale phenomena, like the movement of astronomical objects, seems deterministic to a high degree of precision. How can such phenomena be so predictable when their basic constituents are not? It was the attempt to model astronomical phenomena that led Newton to his physics, by examining the behavior of pairs of masses interacting gravitationally, called 2-body mechanics. Such pairs can be sufficiently replicative in their behavior to make reasonably precise prediction practical, but that only goes so far until the perturbative influences of other bodies becomes significant, and we encounter the n-body problem. There is no general solution to that problem possible, we are left with only approximation methods that may work well enough in restricted situations, by avoiding "singularities" that would defeat computational predictions, but even in those situations are fraught with uncertainties and may require more computing power than can stay ahead of real-time trajectories.

For chaotic systems small perturbations can be significant. It has been estimated that perturbation from the gravitational influence of the dwarf companion of the star Sirius can affect the outcome of a game of billiards, where small changes can have large effects. No matter how skillful the players, there will always be an element of randomness in the course of the game. And while the orbits of planets of the Solar System may have been fairly stable for the last 4 billion years, we can computationally predict that the system is also chaotic over a longer time span, and that eventually the Earth or other planets may be flung out of their current orbits, perhaps out of the Solar System or into the Sun.

In biology we have come to the realization that the amount of information carried in the coding of our genes falls short of being enough to specify the details of our bodies or our minds. We are all randomly self-organized systems, for which genes and environmental influences may have had some impact, but which are fundamentally indeterminate, even in principle. Our genes may make it likely that we will have fingerprints, but they do not specify the patterns in detail. Genes may make it likely that a leopard will have spots, or a zebra will have strips, but identical twins will not have the same fingerprints, the same spots, the same stripes.

Within our bodies, our hearts are chaotic systems. Our genes may constrain the self-organizing of stem cells into a heart that beats, albeit somewhat irregularly, and can respond to increased demand for it to pump faster, but without a master control mechanism like the pacemakers we install when the function begins to falter. We have brains, but evolution has not attempted to enable our brains to command every detail of our bodily functions. Social insects function using simple rules for each member of the colony with no command structure. Evolution has produced designs that allow for leadership but not command, and there is a deep reason for that. It is not just that command management of complex systems is unnecessary or inefficient. It is that such command management is impossible in principle. We will never be able to redesign our genomes by computationally predicting the effects of genetic changes on the chaotic structures and behaviors that unfold. We can borrow genes with known effects and apply them elsewhere, but we are doomed to having to rely on trial and error for real innovations. We can redesign ourselves as a species, but we must accept there will be many bad outcomes.

Some unsettling insights


The Universe is rational only to first approximation.
Roland's First Corollary to Finagle's Law.

From all this one can come to understand that all large scale phenomena are chaotic systems. They may seem predictable under certain circumstances, which we may call islands of stability, and we may even be able to so structure complex phenomena that they self-organize into replicative patterns, but we must expect the unexpected when we push the boundaries of those islands of stability, and we can never be certain how any design change will work out as the self-organizing system emerges.

But it is not just large scale phenomena. How does it seem that some quantum systems are "entangled" and some are not, when theory suggests the entire Universe is entangled? The answer that now seems apparent is that those subsystems that seem more entangled are actually islands of stability in chaotic processes, that, like beating hearts, self-organize into predictable patterns for a while, until they are perturbed and the patterns dissipate.

For constitutional design, what does not work, except for a few things like going to war, is command management. The impulse to resort to hierarchical command systems arises from the mental tools we evolved for leading men into combat, but as Helmuth von Moltke said, "no plan survives contact with the enemy", and the outcomes of combat depend less on command management than on the ability of troops to self-organize in real time. Yet the instinctive impulse persists and leads to authoritarian attempts to do things like manage human behavior and the economy in ways that are fundamentally beyond the possibility of such control, even if it were desirable. We will never become able to prevent all human depravities or ward off all economic collapses, any more than we can do so for storms, earthquakes, or volcanic eruptions (although we might for asteroid impacts). All we can do is to try to prepare ourselves for surviving the calamities and emerging in some order after they have subsided.

However, we don't have to wait for authoritarian methods to fail before we abandon them. That is the wisdom of the libertarian impulse, understood by our constitutional founders. A constitution can and must structure how we self-organize, even if we cannot be sure what clauses will work or how, or how to change them to produce better outcomes. the same design change that has no effect if applied today might cause mass extinction of the human race if applied tomorrow, and heaven on Earth if applied the day after.

So the wisest rule is likely to be a conservative one: Avoid changes that are simple, direct, obvious, and large, because they are almost certainly a bad idea. That cannot be avoided for a new constitution, but sound design of further changes should proceed carefully, and make no move toward micromanaging society and the economy. The best that can work is structuring self-organization, and for that we can learn from experience to make some outcomes more likely, while swimming in a sea of intrinsic unpredictability.


Notes


See also




Donate Now!

No comments:

Translate

Search this and affiliated sites

Blog Archive