Notes from Under Sky

An Orderly Design

Does the search for an intelligent designer start and end with humanity?

A week ago as I write this (in November 2005), the Kansas State Board of Education rewrote the state science standards, starting a hubbub of criticism. One of the standards they voted to add was the criticism of evolution by natural selection.

This is perfectly all right, of course. I want to emphasize that I think that natural selection is all that is needed to explain the undeniable evolution that has taken place in the development of life on the Earth. Just the same, it's wholly cricket to examine the evidence for natural selection, to see if it all really holds water.

However, the Board of Education also removed two words from the science standards: "natural explanations." In other words, the Board doesn't want to restrict explanations of natural phenomena to natural causes. They want to include supernatural causes, too, and it's only, ahem, natural that they should want to include only their supernatural cause. This used to be called creationism, but that has left such a bad taste in people's mouths that they have a new name for it, "intelligent design." Nonetheless, there's no confusion as to whose intelligence is being proposed here.

These two changes are entirely related. I'm all for teaching intelligent design, or creationism, or whatever they want to call it, within the science curriculum—provided they really teach it as a science. That is, they must state it as precisely as evolution by natural selection is stated, and they must subject it to as rigorous an examination.

However, they don't want to do that, because there really isn't any concrete evidence for an intelligent designer. What little evidence proponents of intelligent design are really putting forth is (they claim) against natural selection. By allowing criticism of natural selection, which is perfectly legitimate, and permitting the inclusion of supernatural explanations, they've laid the groundwork for teaching intelligent design—but teaching it uncritically.

Never mind that their criticisms of natural selection are woefully misguided. One frequent claim is that evolution by natural selection violates the second law of thermodynamics. I suppose that if one gets one's science education from the Discovery Channel, one might get the idea that the second law of thermodynamics is about order tending to disorder. Well, obviously, natural selection purports to create order out of disorder (the human being, of course, being the pinnacle of order thus far)—a clear violation of the second law. Imagine those uppity scientists not seeing this contradiction under their very noses. It just reveals their lack of imagination.

Except that the second law says nothing about order and disorder. If people actually bother to look up the second law, they will see that it's about—surprise!—thermodynamics. It says that in any spontaneous reaction, the change in entropy is positive, where the change in entropy is defined as the change in energy at a given temperature, divided by that temperature.

For instance, suppose you have two identical metal plates, except that one is at a temperature of 300 kelvins, and the other is at 600 kelvins, and suppose you bring the plates together. They can begin to exchange energy by conduction, and thereby change their temperatures. Which way does the temperature change? Well, in order to increase a plate's temperature by 1 kelvin, you have to add some amount of heat; let's suppose it's 6 joules. In order to decrease the temperature by 1 kelvin, you have to take away 6 joules.

By the first law of thermodynamics, which says that energy is conserved, whatever energy one plate gains, the other must lose. So at some point, unless the plates are already at the same temperature and therefore at equilibrium, either the cooler plate gives up 6 joules to the hotter plate, or vice versa. For the cooler plate to give up 6 joules, it has to lose an entropy of 6 joules, divided by 300 kelvins, or 0.02 joules per kelvin. In exchange, the hotter plate gains an entropy of 6 joules, divided by 600 kelvins, or 0.01 joules per kelvin. The net change in entropy is therefore –0.01 joules per kelvin.

In contrast, for the hotter plate to give up 6 joules to the cooler plate, it has to lose an entropy of 0.01 joules per kelvin, whereas the cooler plate gains an entropy of 0.02 joules per kelvin, resulting in a net gain in entropy of 0.01 joules per kelvin. As small as that is, it's positive, and the other is negative, which is why (as we would have expected) the cooler plate heats up as the hotter plate cools down. That seems like a lot of work to go through to demonstrate what we already knew, but it's a simple case. The benefit of the second law comes in situations that are not so clear-cut.

The important thing, however, is that all these situations concern thermodynamics, not perceived order. OK, so a few people, judging (perhaps tactically) that they can't attack natural selection that way, have decided that there's another law that says much the same thing, that entropy increases, but applies it to information theory instead. And there is such a law, based on the information theory developed by Claude Shannon.

What about that, then? Doesn't evolution by natural selection contradict that version of the law? Ahh, but that conclusion neglects a proviso of the law of increasing entropy—one that is present in both the thermodynamic and information-theoretic versions—and that is that the system you're observing must be closed. Nothing can enter or exit the system; otherwise, the law can trivially be observed not to be true. In the thermodynamic version, this amounts to external sources of energy. The Sun is one such source. You can get systems to heat unevenly by not illuminating part of it with the Sun. It is the Sun's energy that drives the engine of the Earth's natural systems, and prevents the second law of thermodynamics from holding on the Earth.

In the case of the information-theoretic version, what is the system under observation? It is the the information content of the molecules on the Earth, progressing from single atoms on one hand to complex organisms on the other. This version of the law of increasing entropy states that a closed system cannot proceed spontaneously from simple to complex. Now, what does that mean—"spontaneously"? It means that the change is not particularly likely, based on the dynamics of the system.

That's a hand-wavy sort of way to put it, but it can be made precise in mathematics, using notions of state space (how many different conditions the system can be in, roughly speaking) and probability, and so forth. For instance, if you typed a 34-letter phrase at random, expecting to get the phrase "WHATDOYOUGETIFYOUMULTIPLYSIXBYNINE"—well, you're not very likely to be satisfied. It's not a spontaneous result; what comes out is much more likely to be complete garbage. In fact, if you could type 34 characters in 5 seconds, say, it would still take you over 10 to the 40th years of trying it over and over again to get it completely right.

But a simple thought experiment shows that, again, it's important that the system be closed. Suppose that after you type your 34 random letters, you go through and mark the ones that are correct. It's quite likely, actually—it happens almost three times out of four that you try it that you get at least one letter correct. And you leave the ones that are correct in place, and only white out and replace the wrong ones. Each time, then, you "improve" the phrase, until finally the last correct letter falls into place. If it takes you 5 seconds per pass, how long will it take? Do you suppose that it will take you years, or weeks, or perhaps even only days?

No: Even supposing that it takes you 5 seconds to type the last few passes, down to just one or two wrong letters, it will take you just eight minutes on average to get the entire phrase correct, from beginning to end. What is responsible for this remarkable change in time scale? Your input—the information input that holds the right letters fixed, and the wrong ones changeable.

But what is the information input in the case of the Earth? Surely not the Sun, which may be a source of light and heat but is hardly a source of information. But it is! Certainly the presence or absence of the Sun is of vital importance to many organisms, and the selective application of sunshine is very much a source of selective information. And if we consider individual species as information systems, then predators and food sources alike also become sources of information, all applying their selective influence on the genetic signature of those organisms. None of these systems are closed—not any individual species or ecosystem, not even the entire Earth. So the information-theoretic version of the law of increasing entropy doesn't hold, either, any more than we would expect it to hold when we make selective corrections on the typewriter.

In short, no version of the second law of thermodynamics can be used as a convincing and consistent counter-argument to evolution by natural selection, no matter how much proponents of intelligent design might insist that it can. (Incidentally, have you ever noticed one of these folks admitting that their version of intelligent design violates the first law of thermodynamics? I haven't. Let there be light, indeed.)

The problem is that humans are very good at seeing intent—even when the intent is entirely illusory. We don't win (or more likely, lose) the lottery because of pure chance, it's because we did something good or bad the week before. Sure, a week doesn't go by that we don't do both good and bad things, but who remembers events that go against karma? We desperately want evidence that we have some control over what happens in the universe, and the notion of karma gives us that in spades.

That same ability to see intent underlies the watchmaker paradox. Intelligent design must be true because no undirected, mindless process like natural selection could give us such exquisitely assembled organisms as the hummingbird, say. But obviously physical processes as crystallization are capable of producing startlingly complex structures. Before we can proceed along the intelligent design line of reasoning, shouldn't we gain a proper appreciation for undirected, mindless processes?

But all this is really beside the point, which is that science is by definition not about supernatural explanations for natural phenomena. It is about selection, an artificial selection of ideas, in which we critically examine hypotheses to see which ones are consistent and explanatory, and which ones are contradicted by past observations or are unable to predict future ones. It's impossible to do that with supernatural explanations, where one can always claim that some process is or is not the way that God works, according to one's beliefs.

In other words, the battle is not really about religion vs science. It's about explaining one natural phenomenon in terms of another. We always want to know ultimtely why something must be so, but although it may be intuitively satisfying to think that some intelligent being is at the root of it, and all will eventually be made clear, it isn't science. What science is about is finding that two things you can see or touch or smell, which you thought weren't connected, really are connected, like water and steam, electricity and magnetism, probability and biology. Rooting all of these in an intelligent creator, which cannot be seen, or touched, or smelled, is qualitatively different. It ought therefore to be taught differently.

But no! The Kansas State Board of Education wants the imprimatur of science on intelligent design, and they'll do what they must to shoehorn it into science classes. If they succeed, I can only hope that some teachers will be bold enough to subject it to the kind of scientific rigor it so richly deserves.

Copyright (c) 2005 Brian Tung