Friday, October 23, 2009

Making All Things Make Themselves

I have been reading the new book by Iain McCalman, Darwin's Armada -- Four Voyages and the Battle for the Theory of Evolution. This book has put me in mind of an article I wrote for SIGGRAPH some years back, a version of which was included in Thinking Like Einstein. I think the issues and conceptions presented are still very much alive today -- and are given a new aspect with links to the newest visual technologies.

An article on science and religion in Science magazine noted: "The reaction to the Darwinian theory was . . . diverse when it first it exploded onto the Victorian scene. . . . For Charles Kingsley, a deity who could make all things make themselves was far wiser than one who simply made all things."

Computer graphics has always been about making things, whether making 3D images of real objects, making images of imagined visions or making models of the “unseen” in visualizations of scientific data.

During the Renaissance, the goal of art was to imitate nature, to make true and accurate images of the real world. Even in the poetry of the period, the “prime aim … was to make an imitation” in order to “grasp the essential meaning and value.”

Although this goal fell away some time ago, we can now see that our newest tools make it possible to imitate nature once again - but this time at much deeper levels. The imitation can be not only of images of surfaces, but also imitation of nature’s growth, its physical motion, its processes, its inner workings, its unfolding instructions according to an simple code -- imitating the immense complexity generated by variation of a simple code within a context of meaningful selection. And it is already becoming clear that part of this imitation is developing some understanding of things that “make themselves.”

Virtual Creatures

Years ago, when I first saw the short silent movies of Karl Sims’ block creatures -- walking, swimming, seeking light and guarding their food -- I was thunderstruck. I felt immediately that something very powerful indeed was afoot. And I imagined that what I was seeing was really only the very beginning.

Of course, Sims’ work is in many ways mainly a highly sophisticated extension of Richard Dawkins’ “Blind Watchmaker” program developed for the early Macintosh. With this program (included with Dawkins’ book of the same name), two-dimensional, static, monochrome stick figures varied their form in random ways. This allowed the player to select among an array of mutations for different kinds of shapes and traits, play after play, generation after generation -- until a new and wonderful centipede, starfish or complex crystal was produced.

This piece of software showed the potential of a modest interactive game to reveal a deeper concept, a fundamental process of nature. However, for me, although the stick figures were fascinating, it took the more life-like motions and behavior of Sims’ virtual creatures to drive the point home.

What Works

Accordingly, I think our new tools and our new toys are bringing us face to face with a deeper understanding of how nature’s system really works -- how truly dazzling adaptations to a particular environment can be generated with a relatively simple mechanism, constantly interacting with a particular and changing environment.

This deeper understanding also belies the idea of fixed superiority, since so much depends upon one form of environmental context for any particular form of superiority. The system may evolve an amazingly superior swimmer, but this does not make for a superior walker. (Or, as friend has recently pointed out, a good basketball player would probably make a poor jockey.)

With such models, we can see what an effective mechanism evolution is, partly because of this intense and constant mutual interaction between genetic code and outside environment. Indeed, we might reflect that even if there were overt intention by a maker, this might not be a good thing.

We humans are constantly designing things by intention. But often we come up with the “unintended consequences” that are so often observed and lamented. It becomes clear that making things that make themselves may be a much more practical strategy in an ever-changing world.

But there is a rub here -- for many people, an enormous change of metaphor. We now have to view the maker’s role not as a craftsman or designer or engineer, but as a maker of things over which the maker has only limited control. Over time, the vast complexity generates a freedom of form and function well beyond the power of the maker. And often, I would argue, this is a good thing.

Early Discoveries

One of Sims’ creatures in particular stands out in my mind. In his design specifications, “walking” was defined as forward motion. Most creatures found ways to “walk,” more or less as expected. However, to the surprise of the programmer, one creature found that by simply building itself very tall (with an appropriately placed top-heavy weight), it could fall and tumble heel over head so that it could generate ample forward motion and thus satisfy the original selection criteria. Sims observed that this was a strategy that had never occurred to him -- the programmer, the arms-length designer.

What is no less remarkable is that Sims also found that other creatures took advantage of the bugs and programmer’s mistakes in specifying the physical world. Indeed, Sims noted that such a process was so reliable that it might be seen as a “lazy” way of finding such bugs.

Thus he argued that “it is important that the physical simulation be reasonably accurate when optimizing for creatures that can move within it. Any bugs that allow energy leaks from non-conservation, or even round-off errors, will inevitably be discovered and exploited by the evolving creatures. Although this can be a lazy and often amusing approach for debugging a physical modeling system, it is not necessarily the most practical.”

Unexpected innovation and relentless exploitation of tiny areas of possible advantage -- these are impressive indeed and show the amazing power of nature’s engine of adaptation. If these striking innovations are so easy to observe in a synthetic universe, in only a handful of generations, how much more should we expect in the real world of nature?

Of course, when viewed in this way, ample evidence can be seen in every direction. A recent article in Science News, for example, noted that a graduate student has just found a kind of bacteria living on a certain kind of ant that proved to be highly beneficial to the ants. The ants “farm” a certain kind of fungus for their own food.

However, this fungus would be easily killed by another pest fungus - if it were not for the antibiotic secretions of the ants’ bacteria ride-along buddies. Thus the graduate student “who once mused about funny-looking patches” on the ants found that the insects have “a microscopic partner species overlooked despite about a century of study.”

“What’s interesting from an evolutionary perspective is that once again the ants hit on something before we did. Ants beat humans in developing agriculture by some 50 million years. Now, says [one scientist], it looks as if the same ants came in ahead on bacterial antibiotics by millions of years.”

Learning from the Lowly

Such stories take us to a place where we should be prepared to have much higher respect for apparently ordinary and humble creatures. What we are not sufficiently aware of is that these creatures are the beneficiaries of a system capable of innovation far beyond our own small observations or imaginings.

Perhaps it is time (especially now in an era of climate change) to reconsider centuries of human self-congratulation on our own cleverness and begin to look at the cleverness -- indeed the wisdom -- of humble creatures.

These creatures have used an elegant process to not only produce clever innovations, but to produce clever innovations that are proven to work, again and again, over millions of years. Such accomplishments should teach us a great deal about the possibilities for “sustainable development.”

We have built an elaborate maze of justifications and explanations. Modern culture has sophisticated arguments to explain why human beings are at the top the pecking order and why human language is a supreme achievement. Accordingly, it is very hard for us to fully appreciate the accomplishments of the humble ant, the unsavory fungus and the insignificant microbe.

Yet, our newest technologies -- as the microscope and telescope did centuries ago -- are slowly opening up these worlds so that we may see clearly once again what we should have known all along.

However, learning from such lowly creatures is not easy. Our whole culture, for most of us, has trained us not to think such thoughts. Certainly there is very little in the Modern Western tradition to cultivate such an appreciation.

Generally, there has been virtually no respect for the intelligence of animals. Intelligence is not seen as manifest in nature. Rather, it comes in spoken words and in writing. Respecting the intelligence of animals and creatures that make themselves would seem to be more consistent with the worldview of our very distant ancestors.

It would indeed be curious if our new visualization technologies and computer simulations were to take us back in time to a place where we can admire -- once again, as our distant ancestors did -- the high intelligence of lowly creatures.

These creatures learn and innovate not from the words and logic but from relentless experimentation and selection over long periods of time, thus showing the deep wisdom of making “all things make themselves.”

Based on chapter 12 of Thinking Like Einstein. References: Brooke, John. “Science and Religion: Lessons from History?” Science, Dec. 11, 1998, pp. 1985-1986. Dawkins, Richard. The Blind Watchmaker-Why the Evidence of Evolution Reveals a Universe Without Design, W.W. Norton, 1987. Dean, Leonard. Renaissance Poetry, Prentice-Hall, 1960, p. 1. Milus, S. “Farmer Ants Have Bacterial Farmhands,” Science News, vol. 155, April 24, 1999, p. 261. Sims, Karl. “Evolving Virtual Creatures,” Computer Graphics Proceedings, Annual Conference Series, 1994, pp. 15-22. More information about Karl Sims work can be found at: