Cheaper than printing it out: buy the paperback book.

Out of Control
Chapter 21: RISING FLOW

While we can agree that evolutionary life exhibits no progress, perhaps it has a general direction?

In a quick survey of textbooks on evolution, I couldn't find a single one with the word "trends" or "direction" in the index. In the heated zeal to eradicate the notion of progress in evolution, many neodarwinians have banned any notions of trends or direction in evolution whatsoever. Steven Jay Gould, one of the most outspoken naysayers about evolutionary trends, is actually one of the few biologists who even discusses the idea.

The central metaphor in Wonderful Life, Gould's entertaining book about the reinterpretation of the Burgess Shale fossils, is that the history of life can be thought of as a video tape. One can imagine rewinding life, and by some divine miracle, changing a pivotal scene at the beginning, and then rerunning life again from that point. This time-honored literary technique reached its apex in the all-American classic Christmas movie It's a Wonderful Life, from which Gould adapted his title. In this nearly archetypal drama, Jimmy Stewart's guardian angel replays Stewart's life without him.

If we could replay the epic story of biological life unfolding on Earth, would it progress in a similar story as the one we know? Would life recapitulate any of its familiar stages, or would it stun us with contrary alternatives? Gould spins a masterful narrative of why he thinks we would not recognize life on Earth if evolution could be run again.

But since we have this magical tape of life mounted in our machine, there are further, and perhaps more interesting, things to do with it. If we turned out the lights, flipped the cassette at random, and then played it, would a visitor from another universe be able to tell if the tape was running properly forward or unconventionally backward?

What would the screen show if we played the epic Wonderful Life in reverse? Let's dim the lights and see. The story opens with a glorious, bluish Earth wrapped in a very thin film of living things, some mobile, some rooted. The cast of character types totals in the millions, half of them insects. In the opening scenes, not much happens. Plants morph into endless shapes. Some larger, very agile mammal things dissolve into similar, but smaller mammal things. Lots of insects melt into other insects, while some wholly new insect creatures appear. They too gradually merge into others. If we inspect any single character and follow it in slow motion, it's difficult to discern much sensible change going either forward or reverse. To speed the show, we fast-forward (fast-backward to us).

The screen shows life becoming sparse on the planet. Many, but not all, of the animal creatures begin to shrink in size. The total number of kinds of things decreases. The plot slows down. Living creatures inhabit fewer roles, and the roles change less and less as the tape proceeds. Life steadily collapses in scope and size until it becomes small, bland, and naked to the elements. In a very boring ending, the last variety of animated things disappear as they melt into a single tiny amorphous blob.

To review: a wide, complex, convoluted web of diverse forms just collapsed into a relatively simple, unitary speck of protein that mostly just copies itself.

What do you think, friend from Thor? Is the speck the alpha or the omega?

Life surely has a direction of time, but beyond that, neodarwinists would argue, nothing is sure. Since there are no directional trends in organic evolution, nothing about life's future can be forecast. Therefore the unpredictable nature of evolution is one of the few predictions we can make about it. Neodarwinists count on evolution being unpredictable. Who could have guessed while the fishes leaped in the oceans -- the "pinnacle" of life and complexity at the time -- that the really momentous long-range work was being done by some ugly freaks in dried up mud pools near land? Land, what's that?

The postdarwinists on the other hand keep bringing up the word "inevitable." In 1952, engineer Ross Ashby wrote in his influential book Design for a Brain, "The development of life on earth must not be seen as something remarkable. On the contrary, it was inevitable. It was inevitable in the sense that if a system as large as the surface of the earth, basically polystable, is kept gently simmering dynamically for five thousand million years, then nothing short of a miracle could keep the system away from those states in which the variables are aggregated into intensely self-preserved forms,"

Real biologists cringe when "inevitable" is used in the same sentence as evolution. I believe the reflex is a vestigial response from the time when inevitable meant "God." But one of the few legitimate uses for artificial evolution -- that even orthodox biologists will grant it -- is as a test-bed for directional trends in evolution.

Might there be some fundamental constraints in the physical universe that channels life along a certain grain? Gould addresses this concern by comparing the possibility-space of life to the metaphor of "a very broad, low and uniform slope." Water dropped randomly onto this slope trickles down, eroding a chaotic path of microcanals. Newly hatched channels are reinforced as more water flows down, quickly carving out small valleys and permanently setting the location of succeeding larger canyons.

In Gould's metaphor, each tiny groove represents the historical timeline of a species. The initial groove sets the course for succeeding forms of genus, family and taxa. In the beginning, where the groove meanders is totally random, but once established, the course of the following canyons are fixed. Even though he admits his metaphor has an initial slope that "does impart a preferred direction to the water dropping on top," Gould insists that nothing disrupts the sure uncertain course of evolution. In his favorite refrain, if you replay this experiment over and over again, starting with a blank slope each time, you would get a vastly different landscape of valleys and peaks each run.

The curious thing is that if you actually set up Gould's thought experiment as a real test in a sand box, the results suggest an alternative view. First thing you notice as you repeat the experiment over and over again, as I have, is that the landscape formations are a very limited subset of all possible forms. Many landforms we are familiar with -- rolling hills, volcano cones, arches, hanging valleys -- will never appear. Thus one can safely predict what general structure the valleys and subsequent canyons will take: gentle gullies.

Second, while the starting groove begins at random in response to a random falling drop, the shape of further channel erosion follows a very homogeneous course. The canyon unfolds in an inevitable sequence. Continuing Gould's analogy, the initial drop is the first species on the scene; it might be any unexpected organism. Although its traits cannot be predicted, the sandbox analogy says that its descendants unfold somewhat predictably, according to trends inherent in the makeup of sand. So while there are points in evolution where results are sensitive to initial conditions (the birth of the Cambrian explosion could be one) this by no means rules out the influences of large trends.

Evolutionary trends were once promoted by prestigious biologists at the turn of the last century. One version is known as orthogenesis. Orthogenic (straight) life advanced in a direct line, from organism A through the alphabet of life to organism Z. A few orthogenesists in the past really thought evolution proceeded without branching: imagine a ladder climbing upwards, each species stationed on a rung, and every rung closer to heavenly perfection.

But even those orthogenesists who weren't so linear were often supernaturalists. They felt that evolution had direction because it was directed. The directing forces were supernatural purpose or some mysterious vital force that infused living things, or God himself. These notions were clearly outside the ken of science, so what little attraction the idea had to scientists was poisoned by its attraction to the mystical and new-agey.

But in the last several decades, godless engineers have made machines that set their own goals and seem to have their own purpose. One of the first to discover self-direction within machines was Norbert Wiener, the original cybernetic man. Wiener writes in 1950: "Not only can we build purpose into machines, but in an overwhelming majority of cases a machine designed to avoid certain pitfalls of breakdown will look for purposes which it can fulfill." Wiener implied that at a certain threshold of complexity of mechanical design, emergent purpose was inevitable.

Our own minds are a society of mindless agents; purpose emerges from that mix in exactly the same way purpose emerges from other nonintentional vivisystems. In a very real sense, a lowly thermostat has a purpose and a direction -- to find the set temperature and hold it there. Astoundingly purposeful behavior can emerge from purposeless subbehaviors cultivated in software. Rod Brooks's MIT mobots built with bottom-up designs perform complicated tasks based on decisions and goals which percolate up from simple goal-less circuits. Genghis the robot insectoid wants to climb over phonebooks.

When evolutionists shook off God from evolution, they believed they had shaken off any trace of purpose and direction. Evolution was a machine without a designer, a watch made by a blind watchmaker.

Yet when we actually construct very complex machines, and when we dabble with synthetic evolution, we find that both run by themselves and acquire a sliver of their own agenda. Is the self-organizing order-for-free that Stuart Kauffman sees in adaptive systems, and the teleological goals that Rod Brooks can grow in machines, enough to suggest that evolution -- however it came about -- might have also evolved some goals and directions of its own?

If we look we may find that direction and goals can emerge in biological evolution from a mob of directionless and goal-less parts, without invoking vitalistic or supernatural explanations. Experiments in computational evolution confirm this inherent teleogism, this self-produced "trend." Two complexity theorists, Mark Bedau and Norman Packard, have measured a number of evolutionary systems and concluded, "Just as recent studies of chaos have shown that deterministic systems could be unpredictible, we claim that deterministic systems may be teleological." For those with an ear that burns at the combined sound of "goal and evolution," it helps to consider this trait less as a conscious goal, plan, or willful purpose, and more as an "urge" or "tendency."

In the following list I suggest possible large-scale, self-generated tendencies in evolution. Tendencies, as I'm using the word here, are general and provide for exceptions. Not every lineage in a category will follow that trend.

As an example, take Cope's Law, a principle often found in textbooks. Cope was a swashbuckling bone collector in the 1920s who put dinosaurs on the map in more ways than one. He was a pioneer dinosaur surveyor and a tireless promoter of these exotic creatures. Cope noticed that, overall, mammals and dinosaurs seemed to increase in size over time. When studied carefully by later paleontologists, though, his observation applies to only about two-thirds of the cases on record; one can find plenty of exceptions to his rule even in the species lines he had in mind. If Cope's law was without exceptions then the largest living things on Earth would not be "primitive" fungi as large as city blocks hiding under the forest floors. Still, there is definitely a long-term trend in evolution that small things such as bacteria have preceded big ones such as whales.