Computer Viruses, Artificial Life and Evolution

Mark Ludwig
American Eagle Publications, Inc.
ISBN 0-929408-07-1
1993

Computer Viruses, Artificial Life and Evolution
The Little Black Book of Computer Viruses Volume II

Mark A. Ludwig

American Eagle Publications, Inc.
Post Office Box 41401
Tucson, Arizona 85717

1993

© 1993 by Mark A. Ludwig.
All rights reserved.
Artwork by the world's most famous virus writer, Michaelangelo
ISBN 0-929408-07-1

Preface
- 1. Introduction
- 2. Are Viruses Alive?
Part I: The Mechanics of Life
Part II: The Philosophy of Life
Part III: The Genesis and Evolution of Life
Appendix A: Introduction to Cellular Automata
Appendix B: Some Basic Biochemistry
Appendix C: The First International Virus Writing Contest
Appendix D: Solving Differential Equations
Appendix E: Stochastic Population Equations
Appendix F: The Darwinian Genetic Mutation Engine
Selected References
Index

Raphael: ecfe

Plato and Aristotle debating the nature of reality.

Preface

In 1831, Charles Darwin climbed aboard a ship named The Beagle and headed toward the nethermost parts of the world. As a naturalist, the observations Darwin made on that journey would have a profound impact on his life and on the whole world.

Today I would like to invite you to take a similar journey with me. A journey clean out of this world, and into the world of bits and bytes in which a computer virus operates. This journey has already revolutionized my understanding of life and evolution. In researching the material presented here, I've had to rethink and rewrite this book several times. Computer viruses just did not fit into any of the usual categories people had for them.

Now I know some people have already decided I'm crazy for writing this book. At least one has even said so in print, months before even a word of it was made public. And certainly the people who fight computer viruses day in and day out may have some misgivings about a book that contains viruses and discussions that could teach people how to make them. I understand that. Yet it seems foolish to try to hide your head in a hole and remain ignorant of viruses. They are here. We may as well learn to live with them, because we get to, like it or not.

As far as science goes, computer viruses may be a small part of the big picture that the broad discipline of Artificial Life gives us. Yet I think they are important because they are the only artificial life-form that has become a phenomenon, rather than just a laboratory toy. And though some particular phenomenon may be a small part of the big picture, the scientist can often make great gains by staring hard at it. Certainly the animals of the Galapagos are a small part of the big picture of life on earth. Short of a few television documentaries, most of us would never know it if they were swallowed up by the ocean. Yet these animals worked mightily for Darwin

Therefore I do not apologize for making use of viruses here. If - as Alexander Solzhenitsyn put it - freedom is "to fill people's mailboxes, ears and brains with commercial rubbish" and "for adolescents of 14 to 18 to immerse themselves in idleness and pleasure instead of intensive study and spiritual growth" then we're all dead. I exercise my freedom to write this book with an eye only for what is true and good. Believe me, it would have been easy to play the demagogue and give the people what they wanted to hear, putting what is true in the back seat I could have filled my pockets with gold for it too. I'm not much of a politician though, and I couldn't ever hope to live with myself if I said "Read my lips" while lying through my teeth. So in a sense, I write not what I wish, but what I must.

I'm only sorry to see that it is getting difficult to say what I have to say in many of the so-called "free" nations of this world. But then, it should be no cause for wonder in a world which increasingly denies the possibility of spiritual growth and sees everything in terms of economics. In such a world, our destiny is not to learn to cherish truth, but to learn to eat from that pile of rotting commercial rubbish and be satisfied.

Mark Ludwig

September, 1993

Introduction

In this volume I want to discuss the relevance of computer viruses to modern science, and specifically to life and evolution.

There has been a debate going on as long as computer viruses have been around, as to whether or not they can ever be beneficial. Usually this debate degrades to the level of "show me one," and then an argument as to whether or not some particular programming application is best accomplished in viral form or not.

Here I want to step back from that fray a bit and look at the bigger picture. I am not here concerned about the economic advantages or menaces of viruses, or the pros and cons of a particular virus, but whether, in studying them, we might learn anything of the world we live in.

Imagine with me for a moment the scene of a modem office, filled with PC's. It is the birthday of some unknown person halfway around the world. People come into the office in the morning, flipping on their computers, only to be surprised by the order to type in "Happy Birthday, Joshi!" The office is filled with commotion. Experts are called in. For days afterward, people are talking excitedly about the incident and wondering what will happen when they turn their computers on the next time.

Certainly if you've experienced this, it's no joke. But should our only response be anger and fear? We respond almost as if it were an invasion from another planet - the terror of the unknown menace - and our aboriginal instinct is to kill it before it kills us. And certain elements of the media - under the influence of the pundits of anti-viral software-dom have tried to encourage us to respond like that.

But shouldn't there be at least a little wonder.....?

No science fiction writer fifty years ago ever imagined something so bizarre. Invaders from other planets, sure. Warrior robots, sure. Genius machines, sure. But computer programs that move around and reproduce like living organisms, and attack other programs? We're not talking about some weird laboratory experiment either. This is a real-world phenomenon that the average person is becoming more and more used to.

So let's wonder a bit.

These things have properties similar to living organisms. Are they alive in some sense? If so, can they teach us anything about carbon-based life - or can our knowledge of carbon based life teach us anything about computer viruses and what we might do with them? Matters like evolution and the beginning of life obviously come to mind, as well as ideas like consciousness and intelligence.

Perhaps we can use viruses to better understand what life is in a more abstract sense. A lot of people are interested in finding life somewhere else in the universe. But how would they even know they found something living, unless it was carbon-based like us? It might be that you could stare a life-form in the face and never know it without a sufficiently abstract concept of what life is. Or perhaps you could be conquered by it, and never know it until the conquest was long complete. You can find lots of books on the subject of Extra Terrestrial Intelligence, and most barely touch on this question.

Recently I brought up the possibility that viruses might be worth studying for scientific reasons among a room full of anti-virus types. A number of people in the audience vigorously shook their heads without a second's thought Poor, closed-minded souls they are. I said it in the last volume, and I'll say it here: be willing to listen to different ideas at the risk of offense. If you find yourself irritated by what I say, at least consider the possibility that you might be wrong. I promise you, I will do the same. If I didn't, I'd never grow, I'd never learn anything or expand my horizons, and I certainly don't want that, and I don't think you do either. Believe me, I'm not writing this book because I have all the answers. Quite to the contrary, I have lots more questions than answers!

This book will offend people for two reasons: Firstly, it defends the idea that viruses can be good and useful. I'd like to think that it would forever close the door in the faces of those who want to make all free and open discussion of viruses illegal. But that would be a little proud on my part, and a little ignorant of human nature. Some people will never be convinced - by any amount of reason - that computer viruses are anything but totally evil.

Secondly, people will be offended by this book because of its approach to science. Here we get into some deep things that have caused me a lot of trouble over the past several years. Yet I have embraced that trouble willingly, knowing that it is a means to an end that otherwise could not be had.

Let me explain: Once I was a scientist of scientists. Born in the age of Sputnik, and raised in the home of a chemist, I was enthralled with science as a child. If I wasn't dissolving pennies in acid, I was winding an electromagnet, or playing with a power transistor, or doing a cryogenics experiment - like freezing ants - with liquid propane. When I went to MIT for college I finally got my chance to totally immerse myself in my first love. I did rather well at it too, finishing my undergraduate work in two years and going on to study elementary particle physics under Nobel laureates at Caltech. Yet by the time I got my doctorate, the spell was forever broken. As a young student I learned of the great men of science and their noble contributions to humanity. However, as I advanced, I saw less and less of the noble scientist, and more and more of the self-satisfied expert. I saw less and less of the great contribution to humanity, and more and more of the ignored exposition. I began to understand the difference between the science of the textbooks - where hundreds of years are compressed into a few pages of text by admirers of the discipline - and real science done by real men.

One of the beauties of science that attracted me to it was that it seemed to have something to do with absolute Truth. Sitting in a classroom learning of the glories of our government in the early seventies, it just wasn't too hard to imagine that Soviet children were learning - just as convincingly - of the glories of their government too. Or if I wrote an essay for my English Literature class and my teacher did not like it, what made his opinion better than mine? This world seemed full of ambiguities. But science - ahh, science - here was something I could lay my hands on. Who could argue with the motion of a falling object? That was precise and mathematical. It did not depend on what country you lived in, or which century, or on the opinion of someone who didn't like the way you look. Science was True.

My love of science was born out of a love for Truth, and my studies nurtured that love. Yet as a graduate student I came to realize that, practically speaking, it was impossible to make an important contribution to science without being a master politician. Scientists are not different from other people. They have jobs and egos to protect. They have likes and dislikes. And there are fads and fashions in science just as surely as there are fads and fashions in clothing or music. In order to do something really new in science, you must become a trend-setter - kind of like a fashion designer. The alternative is to simply pander to existing trends. Truth with a capital T - for all practical purposes - takes a back seat.

I do not say these things to get down on science or scientists per se. I greatly respect some scientists even when I disagree with them. And I suspect a lot of them hate this predicament as much as I do. ... And all of it is very understandable. If you try to say something really new in science, it's like speaking a different language. For someone else to understand you, they have to do a whole lot of work. Then, of course, it is a question of why they should spend so much energy to understand you. Will it be worth the trouble once they do? Unless you are famous - the Fashion Designer - they probably won't think so. If you were handed a book written in ancient Egyptian hieroglyphics, and you didn't have a clue what it was about you probably wouldn't go to the trouble to learn the language and translate it. Now if it was a really great book, and you knew it you might go to the trouble. Might. That's what doing something new in science is like. And most scientists are so inundated just trying to keep up with the trends (which, after all, put food on the table) that they have little motivation to look in any other direction, unless that direction promises to unlock the answer to some homing question. And of course, that burning question is usually defined by the current trends.

Anyhow, my quest for Truth was in ruins, at least as far as science could take me. I had sought the treasure at the end of this rainbow, only to find that to be a scientist, I must be a slave to the whims, passions and opinions of my peers. And I knew I wasn't the charismatic politician who could seize the day and bring all the world to my feet Therefore I had no real chance of doing the kind of pioneering science I wanted to. I knew I'd end up wheedling out the details of some remote corner of the universe that nobody even really cared about, or I'd end up wasting my whole life pandering to some fad which was only going to pass away. I could not have the life of black-and-white Truth I had hoped to find in science. That was tremendously unsettling. So, by choice, I turned toward a more technological life, and got involved in computers. I know a lot of other people got hooked on computers the way I did. It was a whole new world, ripe for exploration. With so much to do, and so few people to do it (relatively speaking), this new world was not yet choked to death with envy, as the old world of pure science had been. One was relatively free to stake out his claim and do what he could. That suited me just fine.

Yet, despite my new career, I never lost my eye for pure science. I spent a lot of time in the wilderness - if you will - far away from scientific circles, trying to understand better what I had seen and lived. With this book I am making a sort of a return to that polite society. But I am at heart a New World ruffian who does not come back to please the court. I have no intention of writing a book popularizing science or Artificial Life or anything of the sort And I have no intention of playing by the rules of the polite society, because I have no need for them or it.

Instead of seeking to please those in power, I come to recommend some changes. Scientists are often people who are very adept at understanding complex equations, and sorting through facts. But they're usually weak on understanding the subjective side of the universe. They consider it unimportant or even unreal. However, that subjective side often deeply influences scientific research. Often pure, blind faith in some totally non-scientific idea motivates vast numbers of scientists over centuries of time. Many of those scientists completely fail to understand these ideas and consider their validity and their value. They simply assume such ideas to be givens and press forward.

An example of such an idea is the "unified field theory," a concept developed by Einstein, which has been the Mt. Olympus for the particle physicist to climb for the past half-century. It is simply the suggestion that all the forces of nature can be described by a single equation which is, in some sense, simple. Einstein had good reason to believe a unified field theory existed. He had wonderfully applied the basic concepts of electromagnetism - already known in his day - to gravitation, resulting in the theory of General Relativity. In so doing, he took a big stride toward unifying these two forces, which were the only ones known in the 1920's.

Yet there is no scientific reason to believe that all the forces of nature should be unifiable into one simple master force. Why not three? or sixteen? Secondly, one must question the value of a unified field theory and its relative importance in a larger scheme of things. Untolled capital and human effort has gone into discovering such a theory, with only very limited success. While scientists today are clamoring for hundred-billion-dollar particle accelerators to further divine the mysteries of this elusive idea, there are fascinating mysteries about elementary particles right under their noses which might yield an understanding of nature as revolutionary as anything that has been. Such mysteries are ignored, however, because they do not particularly fit in to the philosophical framework. Perhaps the unified field theory is nothing more than a decoy, which will bury several generations of scientists in the sands of time.

To be a good scientist, one must also be somewhat of a philosopher. A few centuries ago, scientists were called natural philosophers. There was wisdom in that. When a scientist fails to be a philosopher, he tends to be blinded by philosophy. Then he becomes a slave to what he believes, rather than its master. Practically, he turns into a sort of religious fanatic whose goal is to bend all the world to what he believes, rather than to seek truth and himself grow up.

This kind of blindness is unfortunately rampant among scientists. Few people understand the philosophical presuppositions they make, and few understand the philosophical consequences of what they do. Right now, the whole scientific "system" seems to promote such blindness. Most scientific research and institutions which carry out scientific research are heavily government funded. That means they are government controlled. Research must then develop technology which the government wants, or promote a philosophical world view which panders to the government Too often that means a philosophical materialism which elevates the government to the status of God, by default. ¹ The whole "peer review" system is inherently conservative in a way not unlike religious conservatism. It tends to resist beneficial change because it excludes those whose philosophy doesn't conform to the standard. Yet unlike the religious conservative, who openly reasons from theological doctrines, the scientist's doctrines are often subliminal and unspoken. That makes him a slave. The solution is to become part philosopher. Then, at least one knows more about what assumptions he is making, and he can consciously think about the validity of those assumptions. Then he is master. ²

The bottom line of all of this is that I am intent on avoiding such blind spots. The biological sciences are not different than any other kind of science. So I have every intention of ingeniously discussing the philosophical issues that undergird them. The reason for doing that are threefold: Firstly, I don't believe I could do justice to what computer viruses have to offer to science without discussing philosophical issues. Our understanding of life has always had deep roots in philosophy and religion. So if we're going to talk about viruses as being somehow alive, I simply must speak to these non-scientific understandings. Secondly, I think I would be cheating you, my reader, if I did not discuss philosophical issues. The barbarians who do not understand the philosophy behind what they're doing invariably become false prophets. When faced with a point of contention, they naturally defend their philosophical assumptions as if they were divinely revealed Truth. And the expert who, by virtue of his imputed expertise, gains ignorant followers, can easily sacrifice honesty and depth to become a demagogue and a sophist. I can't stand that attitude in others, so I don't want to take it with you. While it is not my intention to offend anyone, I know I cannot avoid it if I discuss the subject at hand honestly and openly. Thirdly, I think there is a real need for such a discussion for science's sake. From the point of view of a stalwart physical scientist, theoretical biology - and especially evolution - looks a lot like voo-doo. It is very effective if you believe in it But when you dig into it, it looks more like magic than hard science. Frankly speaking, the biological sciences are being choked by philosophical dogmas. Artificial life is new enough that it could challenge those dogmas and bring about some needed and tremendously worthwhile change. Unfortunately, it seems that the general tenor of AL work is not just to buy into the same philosophy that has marred biology but to take the lead in it, to be its prophet, and to push it to its logical conclusion. ³ That only confirms and furthers the errors. I think that is a shame in view of the potential AL has to put biology and evolution on a more solid footing.

So with that in mind, I'd like to again invite you to come exploring with me. I do not make myself out to be an expert guide in these waters, but only a fellow explorer and adventurer. As far as I can tell, they have never been explored before. So let us go and see some wondrous sights, and play together on sandy sun-drenched beaches where no man has walked before. Let us stand in awe of the might of the deep, and the silence of the stars, and humbly know our weakness. And let us not forget to laugh heartily at the fools who knew we'd fall off the edge of the world.

1 Much of modern history and politics can be understood in the light of Hegelian philosophy, in which the government is made into "God walking on earth."

2 Philosophy cannot be put on the throne either, though. Even in the philosophy of science, you can find some pretty dumb ideas. And some philosophers are active, describing how science should be, while others are more passive, describing how it is, etc., etc.

3 I think the reason for this attitude is that AL researchers tend to be eager to gain the acceptance of mainline biologists.

Are Viruses Alive?

Most of this book will focus on the similarity between a living organism and a computer virus. From a purely naïve point of view, computer viruses seem to be alive. They seem to have a will of their own, doing things in a computer quite independent of what the operator wants: they reproduce, and some have proven quite adept at moving around from computer to computer. They come in many varieties, some colorful, some secretive, some dangerous, some innocuous, some extremely active, some sluggish.

But are they really alive?

Or are they just good imitations, like a little mechanical barking puppy in a toy store - maybe a bit more sophisticated, but nothing more than a cute (or not-so-cute) machine after all?

Perhaps a better question is, in what sense are viruses alive? Certainly they are not carbon-based organisms such as we are. However in this scientific age of ours, it seems a little foolish and narrow-minded to call only carbon-based organisms life. When we are exploring the limits of our universe for alien forms of life, both by direct observation such as the Mariner expedition to Mars, and by indirect attempts to receive intelligible radio signals, we have to expand our horizons. We have to dig deeper and ask the question "What makes life life?' Only when we know what life is can we weigh some entity in our understanding to determine whether or not it is alive, and thereby properly recognize it as life.

The whole phenomenon of computer viruses brings this question close to home. They are here, now. We can contain them and experiment with them. And, like it or not, we have to deal with them in our day-to-day lives. Alien life forms are not so readily accessible.

I, for one, think it is very important to take advantage of the opportunity which viruses offer us to broaden our understanding of what life is. Firstly, I believe we can only stand to gain a better understanding of life as it is by studying life as it could be. Modern man is an arrogant creature, who usually thinks he knows a lot more about this universe than he really does. This boastful pride seems particularly strong in relation to extremely complex systems like living organisms. The truth is, we know very little about life today. We know lots more than we did a century ago, but we still shouldn't deceive ourselves into thinking we've somehow arrived at the final word. For example, the difficulties we face in synthesizing carbon-based organisms severely limit our ability to perform experiments to better understand how the DNA coding (genotype) affects the physical characteristics (phenotype) of an organism. Even if we could synthesize DNA strands at will, and build the complex machinery that goes with them to create living organisms, we might not want to do so, for fear of unleashing a monster that would make the Bubonic Plague look like Chicken Pox.¹ Computer simulated life - or artificial life - may offer a reasonable way to safely study this genotype/phenotype connection.

Secondly, if the day should ever come that we do discover life on an extra terrestrial body, shouldn't we be ready to recognize it as life and act accordingly? At present, if we discovered a life-form that was not carbon-based we probably would not recognize it Then what damage would we do? Would we completely obliterate it and never know it? Or would we so offend it that it will take up the goal of obliterating us from the bowels of the universe? Some people think we will just naturally recognize alien life-forms and respect them (or go to war with them) when we find them That is naive science-fiction. Just a century and a half ago our ancestors held black slaves, and sometimes beat them and killed them, thinking they did no wrong because negroes weren't human. Some people really believed that. We may look back in disbelief, but we're no better off (probably a whole lot more ignorant, in fact) when it comes to exploring the universe for life. That needs to change - or else we'd just better stay home and shut up.

Any time we have ever gone out exploring our world, we have had to work hard to come to grips with it The best and brightest of mankind have labored all their lives to understand a little piece here and there. Yet even they have fallen short of all-knowing comprehension. That little piece of understanding has often come only with great trouble: intense hours of labor, searching, trying and failing to see, giving up, returning to the chase, and only then, insight Often rejection and persecution follow for daring to share that little bit of knowledge with mankind because it forces men to change their philosophical presumptions about themselves and their relationship to the world.

So why should a deeper understanding of life be any different? Certainly, what we know of life on our planet has only come with great difficulty. If we try to understand life in a more abstract way, we will again be stretched. We will again have to wrestle with difficult ideas and stubborn facts. But wouldn't it be responsible to do that now, rather than only after we've made a serious blunder in destroying a whole civilization for want of even knowing it was there? or destroying our own civilization by stepping on somebody's toes?

As soon as we begin to ask the question "What is life?" we come to the deep realization that our very concept of life is woefully inadequate for any scientific purpose. "What is life?" is a difficult question for which there is no crisp, clean answer. In a sense, life is a metaphysical concept familiar to us in experience, but difficult to cast into a scientific mold.

At present it is fashionable in scientific circles to try to jettison all metaphysical considerations when studying life. The assertion is made that a living organism is simply a highly complex machine built of organic molecules, and that it is not fundamentally different from any other system of molecules. Once such an assertion is accepted, the question of what life is becomes merely a question of function. Design a machine with the proper functions, and it will be alive. This has largely been the philosophy and objective of the AL community.

Of course, such an assertion, until proven, is little more than a metaphysical consideration in and of itself. And to take it as a given is to avoid the question of what life is, not to confront it Indeed, it may be theoretically impossible to prove that a living organism is a highly complex machine which can be understood using only the presently known laws of physics. The only way such a proof could be reasonably accomplished would be to "solve the equation" of a complex living organism, and successfully predict its behavior. There are a number of formidable obstacles to doing that:

There is every reason to believe that the most intractable form of catastrophe theory (the idea of how a butterfly flapping its wings in Japan could cause a tornado in Kansas) must be involved in determining the behavior of living organisms.² Thinking in terms of basic physics, why do one sequence of vibrations in the eardrum of a man result in a smile and a handshake, while another sequence (the same words being spoken, just a different tone of voice) results in a fist fight? If such results are purely due to the known laws of physics, a staggering degree of accuracy will be needed in any calculation that could produce accurate end results.
If any result becomes too sensitive to the initial conditions and the inputs, relativistic quantum field theory must come into play. To make sufficiently accurate calculations, one must take it into account. However, we aren't even sure what the laws of physics are at that level And if we were, quantum uncertainties alone could bar the way to obtaining a decisive answer.
The sheer magnitude of a calculation (to a given level of accuracy) for even a single-celled organism preclude the possibility of any computer modeling it. There is simply too much information involved, and any real computer has a finite memory capacity.

Thus the idea that a living organism is no more than a complex machine could well remain a metaphysical concept which will be argued - pro and con - until the end of human history. Certainly, this idea of life as a machine is nothing new: it has been argued, and waxed and waned in favor, since the days of ancient Greece. More on all of this later....

In view of these complications, I do not want to take the easy way out of dealing with the question of what life is by defining a living organism as a little machine, and nothing more. To do so is simply not intellectually sound, even if it is a very common and acceptable thing to do at present. Of course, neither do I want to adopt a purely metaphysical definition of life (e.g. "Something is alive if it has a spirit"), and settle the question that way. Function is certainly important in any discussion of life. Any candidate for the label "life" must perform certain functions which we normally associate with a living organism. However function must be viewed as a component of the larger framework of our metaphysical and philosophical understanding of life, and not as the whole framework.

Here I am consciously making a break with the AL community. I think AL'ers have gone at the question of what life is with the traditional naïevity of scientists. They assert that life is nothing but atoms and physics, define life purely in terms of function, and then proceed to build models with the proper functions. These models are then cautiously suggested to be alive. The danger here is that you will exalt yourself into a "creator-god" and trivialize life to match your creative powers. The idea that you have somehow become a creator of life is intoxicating - but when you find yourself making statements like "we see that a candle flame is a life form,"³ you'd do well to start considering yourself intoxicated, because certainly others will. Naïve pride often finds its end in foolishness.

In the end, we will find two important results: Firstly, our metaphysical understanding of life always has a direct bearing on our scientific understanding of it The two cannot be separated. Secondly, computer viruses are important not just because of their functional aspects, but for philosophical reasons as well. They can force us to confront metaphysical issues - and maybe even resolve them - if we choose to admit such issues exist (and they do, whether or not we admit it).

So I would like to ask the question "What is life?" and view it first from the physical, mechanical angle, and then from the philosophical angle. At the same time I want to apply some of the answers we get to computer viruses, and see if they are alive. Then, given an understanding of where our viruses fit into the grand scheme of things, I'd like to use them to look at some of the real-world problems which life presents to the scientist

1 Some researchers believe AIDS got its start in just this kind of experimentation. See the video, The Strecker Memorandum, (The Strecker Group, 1501 Colorado Blvd., Eagle Rock, CA 90041:1989). Dr. Strecker demonstrates that a virus like AIDS was predicted as early as 1966, its development suggested in a 1972 World Health Organization bulletin, and it was spread by human agency in smallpox vaccines used in Africa.

2 In feet, catastrophe theory is ideally suited to biology. See P.T. Saunders, An Introduction to Catastrophe Theory, (Cambridge University Press, New York:1980), pp.x,98, 127.

3 Edward Rietman, Creating Artificial Life (McGraw Hill, New York: 1993) p. xvi.

Part I The Mechanics of Life

Mechanical Properties of Life

It seems reasonable to suggest that there might be a certain set of functions which any physical system ought to be capable of performing if it is to be classified as alive. That is, a living organism ought to be able to do certain things.

Unfortunately, defining such a list of functions proves to be an almost intractable mess, even when merely dealing with carbon-based organisms. For example, we have a general idea that living organisms ought to be able to reproduce. Yet exceptions can be found. Mules are not capable of establishing themselves as an autonomous race, but they are still very much alive. On the other hand, a Sodium Chloride crystal in a saturated solution of Sodium Chloride does grow - the structure of the crystal reproduces, yet we do not commonly believe it is alive. In short, no matter what kind of a list of functions we can come up with, one can almost always find an exception - either something which common sense would suggest is alive, but doesn't perform a required function, or something which performs the function that we wouldn't quickly call alive.

This problem is particularly acute when dealing with macroscopic functions. If one focuses down on the microscopic details of how carbon-based organisms work, one can draw the line between life and non-life much more closely. Then we are turning away from the quest for an abstract understanding of life, though, and focusing on how life as we know it works.

To define life in the abstract, one must stay away from the microscopic details which characterize specific systems and focus on abstract properties.

The only realistic way to avoid being caught in a web of propositions and counter-examples is to take a step backwards and give up the idea of a set of functions which act as a dividing line between life and non-life. We admit that, even functionally speaking, we really don't know what life is and how to define it - but we do know something about how living organisms behave. As such, we can look at any system in the physical world, or in the memory of a computer, and ask if it performs functions similar to those of living organisms. If so, then it has a certain claim to life. That claim is stronger or weaker, depending on how closely its functions compare to those functions we consider essential to life. Such is the approach that researchers interested in the concept of Artificial Life have taken, and we will adopt it here. In doing that, I don't want to wholly abandon the idea of a set of functions that would be necessary for a system to be alive. Rather, we take the attitude that, due to the newness of this field, and our ignorance of it, we cannot yet begin to formulate such a list of functions. With that attitude, we understand that others may attach a different relative importance to the various functions than we do, and we invite free and open discussion and even argument.

Perhaps one of the simplest examples of a computerized simulation of life which exhibits a function usually attributed to living organisms is John Conway's game of Life, which dates back to 1970.¹ (A copy of this game is included on the Program Disk for this book.) This program simulates population dynamics of living organisms in a rather rudimentary way. It consists of a logical cellular array, initiated in some arbitrary fashion so that each cell is either on (populated) or off (unpopulated). The array is then time-evolved according to the following rules:

If 3 of the neighboring 8 cells of any given cell are on, then that cell is turned on.
If 2 of the neighboring 8 cells of any given cell are on, then that cell is left in. its current state.
If 0, 1 or 4 to 8 of the neighboring 8 cells of any given cell are on, then that cell is turned off.

These rules allow for colony growth, as well as death due to over- or under-population. Though very simple, the rules allow for complex population dynamics similar to the behavior of colonies of living organisms.

Of course, no one would seriously suggest that Conway's individual cells are really alive,² but they do simulate the behavior

Fig. 3.1: The rules of Life.

of a population, and one of the functions of a colony of living organisms is its population dynamics. So such a model is of interest in artificial life research.

What we are really interested in here are individual organisms, rather than populations, so we will concentrate on those. In this realm, artificial life researchers seem to put a great deal of emphasis on several functional aspects of life:

The ability to reproduce and the method of reproduction.
The concept of emergent behavior.
The possession of a metabolism.
The ability to function under perturbations of the environment and interact with the environment
The ability to evolve.

I would like to discuss each of these aspects of life and artificial life in some detail, so I will devote a chapter to each. However I am not ready to buy into all of these ideas completely or even suggest that they constitute a "good idea" of what life is, functionally. For centuries men believed that a living organism could be set apart by its ability to move, be it locomotion, or, in the case of plants, growth and generation. This concept led clock makers of the 17th and 18th century to take up the goal of reproducing life mechanically. By the mid 1700's Jacques de Vaucanson built a mechanical duck which could stand up, sit down, flap its wings, look around, eat and relieve itself.³ People seemed ready to claim it was alive. By today's standards, though, we would not call such a contraption alive in any sense. However it is unclear whether AL's standards are really any better. That is, to some extent, a philosophical issue which we'll leave for later. Yet we should not be too quick to assume that we even know what's important yet, any more than the clock makers. Indeed, our very studies may bring important new functions to light.

1 The story of the development of this game is recorded in Steven Levy, Artificial Life: The Quest for a New Creation (Pantheon, New York:1992) pp. 49-58.

2 Although Conway's rules support universal computation, and therefore presumably a logical equivalent of any artificial organisms we may devise. See Elwyn Berlekamp, John Conway, Richard Guy, Winning Ways for Your Mathematical Plays, Vol. 2 (Academic Press, New York:1982) pp. 817-850.

3 A. Chapuis and E. Droz, Automata: A Historical and Technological Study (B.T. Batsford, Ltd., London: 1958), translated by A. Reid.

Self-Reproduction

Living organisms, in general, are able to reproduce. Although specific individuals may not be able to, either due to accidental circumstances, their stage in the life cycle, or an unusual genetic combination, members of the population as a whole must have the ability to make copies or near-copies of themselves. Barring that, the population will simply not be around for long.

The abstract concept of self-reproduction has been studied almost wholly within the domain of computer programs. Real-world self-reproducing machines have never been constructed - primarily due to their complexity - although they have been proposed.¹ John von Neumann is usually called the father of the self-reproducing machine. He developed an abstract theory of self-reproduction in the 1940's and 1950's, and described a very complex self-reproducing automaton (machine) as an example. His work was left incomplete at his death in 1957. A student of his, Arthur Burks, organized and finished it, and published it posthumously in 1966 as The Theory of Self-Reproducing Automata.²

One of the first problems one must face in discussing self-reproduction in the abstract is to differentiate between reproduction which mimics living organisms, and trivial reproduction, which might be more like the growth of a crystal. The latter is driven by "obvious" and relatively simple physics, whereas the former is less than obvious, and has to do with the detailed structure (e.g. information content) of the system. For example, in Conway's game of Life, if the cells were considered to be individuals, then the game would exhibit the trivial type of self-reproduction, where cells reproduce obviously as a result of the rules we defined for the system. Most random configurations of the array will result in reproduction somewhere in the array as it is time-stepped.

Von Neumann and Burks were primarily interested in proving the existence of a non-trivial self-reproducing automaton. They developed their ideas from within the framework of the idea of a Turing machine. The Turing machine is simply a generalized computer. It consists of a finite number of internal logical states or rules, and a tape, which contains a (possibly unlimited) number of instructions to be executed. A universal Turing machine is a Turing machine capable of carrying out any finite definable calculational procedure (an algorithm, or program). In 1936, Alan Turing defined such a universal machine.³ Today close approximations to them are on desktops in the form of general purpose computers.⁴ Von Neumann took this concept of a calculating machine and extended it to the concept of a constructing machine, or Constructor. Von Neumann's Constructor was similar to a Turing machine in that it had an input tape of possibly infinite length, and a finite number of internal states. However, instead of using the tape to perform an algorithmic computation, the Constructor used the information on the tape to construct another object A Universal Constructor - in analogy to a universal Turing machine - was a Constructor which was capable of constructing any object (which can be constructed out of a given, finite set of materials) which might be specified on a tape.

Von Neumann and Burks were able to demonstrate the existence of a Universal Constructor within the framework of cellular automata. (Cellular automata are a very important part of AL research, and we will assume throughout that the reader is familiar with the concept For those who are not, a short introduction is provided in Appendix A) In The Theory of Self-Reproducing Automata, they describe a cellular automaton in a system with 29 possible states, and perhaps a half-million cells. (See Fig. 4.1) Such an automaton may not be even close to a minimal configuration, but the important point is that it was a Universal Constructor in the cellular system.

Once Von Neumann and Burks had proven the existence of a Universal Constructor, they had also proven the existence of a self-reproducing automaton. One need only feed the Universal Constructor a tape which contained the instructions for constructing another Universal Constructor, complete with a new tape, and the constructor would make a copy of itself. Hence it was a self-reproducing machine.

Von Neumann's self-reproducing automaton did not fit into the category of simple physics-driven reproduction. It was much more like a living organism in that it relied on a detailed piece of information - the tape - and its detailed design to drive the reproduction process. Such self-reproduction was strikingly similar to that achieved by living organisms. They, too, rely on information - DNA - coupled with a mechanism to interpret that information and do something with it However, Von Neumann's machine was far too complicated to do any serious modeling and study with.

You might notice that Von Neumann's automaton was actually greatly over-specified if all you are interested in is self-reproduction as relates to living organisms. Clearly, no living organism is a Universal Constructor in any sense. It is capable of constructing a copy of itself, and limited variations of itself - but that is all. One cannot simply hand an organism any arbitrary strand of DNA and watch it construct the beast which would result from that strand. Von Neumann and Burks had proven the possibility of such a self-reproducing automaton, but now - given the possibility - would it be possible to construct something much simpler than what they had proposed?

Fig. 4.1: Von Neumann's self-reproducing automaton.

In 1968 E. F. Codd⁵ was able to demonstrate a much simpler Universal Constructor using only eight states, but it was still a Universal Constructor, and it was still tremendously complex. It was not until 1984 that Christopher Langton⁶ jettisoned the idea of a Universal Constructor in favor of a specialized one. Langton argued that an automaton would be capable of life-like self-reproduction if it used information both actively - as interpreted instructions to execute - and passively - as uninterpreted data which is merely copied. Such an automaton would avoid trivial physics-driven reproduction by forcing the construction of the copy to be actively directed by the automaton itself, rather than passively, by the transition rules. (At the same time, the automaton could take advantage of the transition rules to facilitate its job.) Langton suggested that such restrictions were sufficient to differentiate self-reproduction from physics-driven replication.

0000000000000000000000000000000000
0000000000000000000000000000000000
0000000000022222222000000000000000
0000000000270140140200000000000000
0000000000212222221200000000000000
0000000000202000021200000000000000
0000000000272000021200000000000000
0000000000212000021200000000000000
0000000000202000021200000000000000
0000000000272222227222220000000000
0000000000210710710711112000000000
0000000000022222222222220000000000
0000000000000000000000000000000000
0000000000000000000000000000000000
0000000000000000000000000000000000
0000000000000000000000000000000000

Fig 4.2: Langton's automaton.

Using his new rules, Langton was able to demonstrate a vastly simpler self-reproducing automaton, which consisted of 94 cells in a 10 by 15 array, with eight states per cell. Such a structure was actually simple enough to model on a PC and study its behavior! Figure 4.2 shows the detailed structure of Langton's automaton in its cellular array. Figure 4.3 shows its time-evolution and how reproduction occurs. The Program Disk for this book also includes a program, SRALAB, which demonstrates the Langton automaton on a PC and allows the user to experiment with different configurations and transition rules.

By 1989 John Byl⁷ had demonstrated a number of automata much simpler than Langton's which were also capable of self-reproduction.

           22222222
	  27 14 14 2
	  2122222212
	  2 2    212
	  272    212
	  2 2    212
	  27222222722222
	  21 71 71 711112
	   2222222222222
	   
	   22222223
	  27 17 14 2
	  2122222212         2
	  2 2    242        212
	  272    2 2        212
	  212    212        272
	  2 2    212        2 2
	  272222221222222222212
	  21 71 711111 41 41 72
	   2222222222222222222
	  
	   22222223   22222222
	  27 14 14 2 2 17 17 12
	  2122222212 2722222272
	  2 2    212 212    2 2
	  272    212  2     212
	  212    212        2 2
	  272222227222222222212
	  21 71 71 711111 41 42
	   2222222222222222222
	   
	   22222222   22222222
	  2 17 17 12 211117 172
	  2722222272 21222222 2
	  212    2 5 2 2    212
	  2 2    212 242    272
	  272    242 212    2 2
	  212    2 2 2 2    212
	  2711111 42 21 71 71 6
	   22222222   22222222
	   
	          2
		 212
		 212
		 272
		 2 2
		 212
		 272
		 2 2
	   222222212  22222222
	  217 17 172 21117 17 2	
	  2 222222 2 2122222212
	  272    212 212    272
	  212    242 2 2    2 2
	  2 2    2 2 242    212
	  272    212 212    272
	  2122222242 2 222222 2222222
	  2 711111 3 241 71 71 71 7112
	   22222222   222222222222222

Fig. 4.3: Time-evolution of Langton's automaton.

One, for example, was an automaton consisting of 12 cells of six different possible states.⁸ What a far cry from Von Neumann's giant! (See Figure 4.4)

Fig. 4.4: Byl's automaton.

The problem with taking self-reproduction to the limits which Byl did is simply that the distinction between information-driven and purely physics-driven reproduction begins to get seriously blurred. Let's take a look at this: Langton's automaton contains a "tape" of a minimum of 28 cells. It is this tape which contains the information which the artificial organism uses to reproduce itself. Disallowing the sheath state (2), these cells can have seven possible states, so there are some $7^{28} = 4.6 \times 10^{23}$ possible tapes which could be inserted into the automaton. In all likelihood only a few (it's hard to tell) will effect self-reproduction in Langton's automaton. If a computer could check a thousand combinations per second, it would take 14 trillion years to check them all. That's a thousand times the age of the universe. The rarity of a useful tape suggests that the reproduction of Langton's automaton is a highly information-driven reproduction scheme. It is entirely unreasonable to suggest that simply throwing a random tape into Langton's automaton will result in self-reproduction. On the other hand, using Byl's 12 cell, 6 state automaton, the "tape" is only five cells long, each with 5 possible states. Thus, only $5^5 = 3125$ possible tapes might be constructed. All of these can easily be checked by computer, and only 3 of them result in self-reproduction.⁹ The chances of a random tape yielding self-reproduction is thus 1 in 1042. While the chances are small, are they small enough? What if the chances were one in 100? one in 10? Where does one start saying that reproduction is physics-driven and not information-driven?

One must be careful not to be betrayed by the visual appeal in this game. Certainly Byl's automaton has the "feel" of a reproducing cell. It is a round glob which, over a period of time, succeeds in producing a second round glob right next to it. However one can easily construct an automaton with exactly the same information content dependence that effects "self-reproduction" - only it looks like crystal growth - the automaton grows in one direction with a complex but repeating pattern which reminds one of the growth of a one-dimensional crystal lattice. Such an automaton may look very different, but the information content and the reproduction are identical.¹⁰ (See Fig. 4.5)

So does Byl's automaton mimic a living organism or a crystal of moderate complexity? This is a very difficult question to answer, and it brings us right up against some of the philosophical questions which we need to look at hard in order to understand life. At least for now we have some idea of what self-reproduction entails as modern AL researchers see it - and we can cite some examples of it - even if we are not completely clear on where to draw the line between information-driven self-reproduction and physics driven reproduction.

Fig 4.5: The CRYSTAL automaton.

Viruses and Self-Reproduction

Does a virus both use information as instructions to execute and as uninterpreted data to copy? Most certainly it does! For example, consider the INTRUDER virus discussed in The Little Black Book of Computer Viruses. It obviously executes the code of which it consists. Yet at the same time, in the INFECT routine, it takes all of that code and appends it to the EXE file it is infecting. At that point INTRUDER is using its code as uninterpreted data. This is a very common method of operation for viruses. Thus a computer virus does indeed effect self-reproduction according to Langton's definition.

In fact, computer viruses are commonly set apart from ordinary programs because they reproduce. Normal programs do not reproduce, but, by definition, viruses do. From their construction, we can see that their reproduction is not very different from other self-reproducing automata.

1 Robert Freitas, Jr. & William Gilbrealh, Advanced Automation for Space Missions, (National Technical Information Service, Springfield VA:1982) NASA Conference Publication 2255 contains a discussion of self-reproducing factories in the context of space exploration.

2 John Von Neumann & Arthur Burks, Theory of Self-Reproducing Automata, (University of Illinois Press, Urbana:1966).

3 A. M. Turing, "On Computable Numbers, with an Application to the Entscheidungsproblem" Proceedings of the London Mathematical Society (2) 42 (1936) pp. 230-265, and 43 (1937) pp. 544-546.

4 The "approximation" is simply that a real computer does not have an infinite amount of storage.

5 E. F. Codd, Cellular Automata, (Academic Press, New York:1968).

6 Christopher Langton, "Self-Reproduction in Cellular Automata," Physica D 10, pp. 135-144.

7 John Byl, "Self-Reproduction in Small Cellular Automata," Physica D 34, pp. 295-299.

8 See the configuration files BYL for SRA_LAB on the Program Disk.

9 The programs CHECKB1 and CHECKB2 on the Program Disk allow you to check every tape. They give the following results: 3 different tapes give Byl-like automata (tape 53341, the original, and also 51334 and 54133). If we allow the sheath state (2) in our calculation, we get two other interesting tapes, 10205 and 10222 that result in very different reproducing automata.

10 This automaton is also on the Program Disk. The configuration files for it are called CRYSTAL, and should be used with SRA_LAB.

Emergent Behavior

In a biological organism, the DNA determines the physical characteristics of an organism. It is essentially the medium in which the "program of life" is written. However, the relationship between that program in its raw form - the DNA - and the manifest expression of it in a living organism is extremely obtuse. It is the obtuseness of this relationship that forms the core of the idea of emergent behavior. To understand this concept better, let's look at it first within the context of biological organisms, and then we'll take a look at how artificial life fits into the picture.

Let's start by looking at how DNA works. The DNA molecule could properly be described as a one dimensional crystal, or fiber. However, unlike the crystals we normally think of, DNA is an aperiodic crystal. Rather than being composed of a single molecule (called a nucleotide) repeated ad infinitum, DNA consists of four different nucleotides - called bases - which are all functionally identical with respect to the structure of the DNA itself. These bases may be substituted for one another in the crystal cells without altering the physical structure of the DNA molecule. Therefore, one could conceivably construct a DNA molecule to encode any kind of information desired. For example, a byte could be encoded by a string of four nucleotides; a megabyte of data could be encoded into a DNA molecule four million nucleotides long, which is about the size of the DNA molecule in a simple one-celled bacterium (A pretty compact storage mechanism!)

In fact, the DNA molecule stores information necessary to construct proteins. Proteins are complex chains of amino acids which perform all the functions of cell metabolism, including reproduction. In all there are 20 different amino acids which living organisms use to build proteins with. Each amino acid is encoded in the DNA by a three-nucleotide chain. A protein, which may consist of a hundred or so amino acids, is represented in the DNA by a sequence of three-nucleotide chains called a gene. Each strand of DNA in a cell, which consists of many hundreds of genes, is called a chromosome. See Appendix B for more information about the chemistry of life.

The genes in DNA might be thought of as complex "instructions" which are "executed" by proteins which perform every kind of cell function, ranging from digestion to making hair to reproduction. The encoded DNA instructions for an organism, taken as a whole, are referred to as its genotype (from the word gene). In the case of a living organism, the "execution of the instructions" consists of the proper functioning of the various proteins in a cell to produce life. I say "proper" because an imbalance results in the various proteins attacking each other, and the DNA, resulting in death and decomposition instead of life. This extremely complex "execution of the instructions" is called the phenotype (from the word phenomenon). The phenotype is the outward appearance of the organism, its life span, its dietary preferences, its instincts, etc.

One of the great questions of biology is how the genotype translates into a phenotype. That is, how does a particular genetic code result in a given feature? Until forty years ago this question was purely academic, because our understanding of biochemistry was so limited. Now we have a vast amount of data about the detailed genetic structure of a wide variety of organisms. And the question is still barely tractable due to its sheer complexity.

One might view a single cell as a package of little machines (proteins) each performing its own specialized function within the cell. The DNA specifies how each of those machines is to be built - and what machines are to be built - but it is not in any sense the director of these machines. It does not control or coordinate their activities once they are built Quite to the contrary, the machines manage and maintain the DNA In fact, apart from invoking some kind of vitalism, there appears to be no "director" of these machines - no centralized coordinator of how the machines should work together. The biological organism is more like a complex parallel-executing distributed processing environment than a serial computer.

The concept of emergent behavior revolves around this question of how the genotype - the description of what machines are to be built - determines the phenotype - the coordinated operation of those machines. In short, "emergent behavior" is simply the idea that the behavior of a complex living organism is not directed by a centralized program. Rather, behavior emerges from the complex interaction of the individual parts. The concept of emergent behavior also suggests that the behavior of the organism as a whole exhibits features which could not be deduced from the behavior of the individual parts.

In essence, emergent behavior is what makes life interesting.

When it comes to constructing artificial organisms, the idea of emergent behavior can also be employed, by designing a modular organism with locally interacting parts, but no centralized controller. The behavior of the parts as a whole is not specified. Since the interaction of the parts is normally non-linear, the behavior of the whole organism will not necessarily be an obvious result of the behavior of the parts. The whole is thus, in a sense, greater than the parts.

In this way, computer programs can model the genotype-to-phenotype emergence phenomenon observed in living organisms. The "genotype" for an artificial organism is the specification of the individual machines. The "phenotype" is the behavior of these machines working together.

The emphasis of emergent behavior in AL work has tended to steer researchers in the direction of highly parallel systems, and particularly toward modeling artificial life using cellular automata. The serial computing environment does not seem to lend itself well to the idea of emergent behavior, because serial programs tend toward centralized rather than distributed control.

However, one cannot simply dismiss serial programs. Any computation done on any computer, no matter how massively parallel, can also be carried out on a simple serial universal Turing machine. That is a mathematically proven fact.¹ And certainly one wouldn't want to say one implementation of an algorithm exhibits emergent behavior, and therefore qualifies as life while another implementation of the exact same algorithm does not! In fact, much AL work using cellular automata is actually modeled on a serial computer. So at one level the automaton looks parallel. At another level, though, it is completely serial. So we can't jump to the conclusion that a serial program is necessarily defective in AL research. We might wonder, though, to what degree emergent behavior is representation dependent, more a matter of appearance than of any quantifiable property of a system.²

In addition, the whole idea of emergent behavior might be just a relic of the kind of science people did before they had computers. To suggest that phenomena will arise in a system that could not be deduced from the behavior of the parts suggest some formal method of deduction. If that formal method is analytically solving analytic equations, then maybe we are calling something emergent that really isn't - but we're just not smart enough to find the right equations or their solutions. On the other hand, perhaps we should scrap that game altogether, and call anything we can work out on a computer a valid deduction. After all, couldn't I work the same thing out with pencil and paper - at least in theory?

If emergent behavior is representation dependent, and dependent on some formal concept of deduction, then we have to wonder whether it is even real. Or is it just the perception of our limited minds. Once again, we find ourselves getting into somewhat of a philosophical muddle. This we'll take up more seriously in a few chapters. For now, let's just accept the usual AL wisdom about emergent behavior, and apply it to viruses.

Viruses and Emergent Behavior

Viruses apparently fall short of the ideal of emergent behavior. They tend to follow the paradigm of serial, centralized control, rather than that of a parallel, distributed organization involving a complex relationship between many small parts which interact to produce the phenotype. Viruses were originally developed on serial, single-user computer systems by programmers used to designing centralized control structures. Thus it is hardly surprising that they are what they are. They were not originally attempts to model living organisms. Still, if we are to suggest that viruses are alive in some sense, we ought to deal with emergent behavior carefully.

There are two problems we face. One is purely technological, and one is deeply philosophical. The latter - which involves trying to understand the true nature of emergent behavior - we have only hinted at, and we will take it up more fully later. The technological problem with viruses is that they don't even appear to exhibit emergent behavior. The relationship between genotype and phenotype seems somewhat trivial in a centralized, serial program. The phenotype is the genotype executed by the processor. And if you have the assembly language listing of the virus, you can study it and understand the phenotype. Of course, if one were to look at a file full of the miscellaneous bytes called executable machine language instructions, one would have a hard time imagining what the execution of those bytes would really look like. So we can't say that the relationship between genotype and phenotype is entirely trivial.

Viruses are normally designed from the top down. A given behavior is imagined by the author, and a program is designed to produce this behavior. The fewer "surprises" the better. Yet emergence suggests exactly that - interesting "surprises." So serial virus programs don't do very well at even giving the appearance of emergence. Instead, they give the appearance of careful centralized control. The idea of emergent behavior moves us away from centralized control toward localized, distributed control.

On a single-user operating system like DOS, one is pretty much stuck with serial programming. In that environment, viruses may never appear to exhibit anything like emergent behavior. However, a multi-threaded preemptive multitasking environment like OS/2 provides some fascinating possibilities. We could design a virus which looks much more like a living organism, with different parts that perform different functions, working independently and yet working together. Such a virus could exhibit at least the appearance of emergent behavior - as much of an appearance as anything AL research has produced.

Now, as I said at the start of this section, we aren't going to use a set of hard and fast rules to determine what is alive and what is not. We just don't understand life well enough to do that yet Just because viruses don't look very "emergent" we aren't going to conclude they're not alive. All the more so since we have some unanswered questions about emergence itself. In fact, the apparent lack of emergent behavior will be valuable in the next part of this book, when we dig into evolution. That's because the relatively trivial genotype to phenotype connection facilitates our analysis of evolution, while evolution itself is not particularly dependent on emergence to work.

1 Roger Penrose, The Emperor's New Mind, (Oxford University Press, New York: 1990) pp. 30-73.

2 It seems as if the AL community has at least a subconscious appreciation of this problem, as "emergent behavior" is often used as a convenient label for any unexpected, unpredicted results, not just those arising from the complex dynamics of distributed processing. Thus, one might find it being used to describe genetic algorithms, etc., which are not necessarily parallel in any sense.

Metabolism and Adaptability

Now I want to discuss two different mechanical aspects of life: metabolism and adaptability. Although different phenomena, I lump them together because in our world of bits and bytes, they are quite useful at helping us to distinguish between self-reproduction and physics-driven replication. Beyond that, they become rather nebulous, and often misleading as to what life is and is not.

Metabolism

Concisely put, most living organisms use energy to maintain themselves in a state of low entropy and carry out their activities, including self-reproduction. Biological organisms either make use of direct energy from the sun (plants/photosynthesis), or they convert energy stored in complex organic molecules into forms they can use (animals/digestion). To understand this process, let's take a look at the second law of thermodynamics.

The second law is one of the most universally applicable laws of physics known. It applies to microscopic dust particles as well as to galaxies and even the whole universe. Simply put, it states that in any isolated system, entropy must stay the same or increase with time. It cannot decrease. Entropy is a measure of the "orderliness" of a system - the greater the entropy, the less the order.¹ Thus, the second law simply states that real-world systems, left to themselves, proceed from states of greater order to states of less order - they decay.

This second law applies to everything, including living organisms. If a living organism is placed in a closed environment, it will soon die and decay. The complex organic molecules it is made of will attack each other and break down into simpler ones. Living organisms can only avoid the immediate consequences of second law decay because they are not closed systems. They continually utilize matter and energy, which flow through them, to keep their entropy low. This flow of energy moves them away from the high-entropy equilibrium point so they can sustain life. (See Figure 6.1) If for any reason the energy flow stops, they die and decay. This was illustrated by the Biosphere II, a pop-ecological project near where I live, in which a few people were enclosed for 2 years in a "closed" ecosystem. Of course, it was closed only to the transfer of matter. Our gas company sent fliers out boasting of how they were selected to provide energy for the project, which consumed as much energy as some 4000 ordinary homes. Without that energy flow - and the tremendous energy from the sun - the inhabitants would have been doomed.

The process by which a living organism uses matter and energy is its metabolism. Presumably artificial organisms should have a metabolism in some sense - an ability to convert matter and energy into vital processes which the organism uses to locally reduce entropy. For real-world organisms, these concepts can be mathematically quantified.² Of course, defining just what terms like matter and energy mean in the realm of cellular automata is rather difficult, unless we go back to the physical machine which is running the automaton program. (Entropy is a different story, since it can be mathematically defined.) Such concepts are closely wedded to real-world laws of physics, and they tend to lose all significance when applied to a completely abstract system.

The requirement for a metabolism can become fuzzy even in real-world biology, though. For example, biological viruses do not have a metabolism of their own. They consist of little more than a strand of DNA and some protective material. This structure is inert until it takes control of another cell's metabolism and uses it to accomplish the virus' plan instead of the cell's.

Fig. 6.1: Metabolism uses energy to lower entropy.

Adaptability

Individual living organisms can generally both interact with and modify their environment, as well as adapt to small changes in that environment. This makes an organism flexible, and stable in the face of change. They can adapt to heat and cold, or a low food supply, or even the presence of other organisms.

Of course, the ability of an organism to adapt to its environment is limited. When faced with an extreme change, e.g., a fish out of water, the organism may not be capable of adapting, in which case it dies.

Artificial organisms should also be capable of interacting with their environment and adapting to it in a limited way. Of course, one should not expect an artificial organism to be capable of adapting to any arbitrary change in its environment It can adapt to some changes, while others might completely throw it off balance, just as in the case of biological organisms. The change that short-circuits an artificial organism does not have to somehow seem big, either. Minute quantities of Arsenic in a pond are just as capable of killing the fish as draining the pond.

Self-Reproduction, Metabolism and Adaptability

The real value of metabolism and adaptability lie in their ability to help us distinguish between self-reproduction and physics-driven reproduction. Think for a moment about the growth of a crystal of Sodium Chloride in a saturated salt solution. As the saturation level of Sodium Chloride increases beyond a certain point, a seed crystal introduced into the solution will grow. The order of the crystal lattice is replicated by the Sodium and Chlorine atoms in the solution. This is a classic case of physics-driven reproduction. In this situation, the environment (the solution) must be carefully controlled in order to make the reproduction work. A slight change in temperature, saturation, or impurities, and the process will reverse itself, causing the seed crystal to dissolve entirely. The adaptability of such a system is almost nil. Likewise, there is no real metabolism involved in this example. There is no machine working to make energy flow through it and pull it away from equilibrium - just simple physics driving the whole system toward equilibrium.

Viral Metabolism

As I mentioned above, it is rather difficult to define concepts like matter and energy entirely within an abstract world of bits and bytes, short of designing a system which simply mirrors our world. The alternative is to look at real programs which run on computers that exist in the real world. In that context, computer viruses certainly do use matter and energy to manipulate entropy. That is, they use the physical components of the computer system and the energy which the computer consumes in order to maintain their existence and replicate. Replication consists of making a copy of a certain sequence of magnetic domains on a disk drive. As the virus continues copying itself, this sequence of domains repeats itself again and again on the drive, increasing its organization (mathematically speaking - if you're the one being infected, you may not see it that way) and lowering its entropy. As such, a computer virus does have a metabolism.

Of course, one might say that just about any computer program can use energy to impose its order on the disk. More often, though, programs are concerned with allowing the user to impose his own order on the disk. They are conduits, rather than being self-sufficient Your word processor allows you to write letters, etc. Those letters represent order on your computer's disk, but the word processor did not create that order itself. It needed your intelligent input to create it

Viral Adaptability

Certainly most computer viruses are capable of adapting to the environment they find themselves in. For example, they can detect the presence of another copy of themselves and adjust their behavior to avoid double-infecting a file, and to stop infecting files altogether once a disk is fully infected. Likewise, they are capable of adapting to hostile programs, such as anti-virus utilities, and remaining quiet while such utilities are in place, or taking measures to escape detection. Likewise, viruses are capable of modifying their environment to promote their own welfare. For example, the INTRUDER virus in The Little Black Book of Computer Viruses can change an EXE file's attribute from read only to read/write so that it can infect it

Computer viruses have also shown a phenomenal ability to adapt to changes in programming techniques and environments. For example, it is amazing that the Jerusalem virus is still capable of infecting a wide variety of executable files and function properly five years after it was released. Most of the programs it infects today were not even written when it was first released. All kinds of new programming techniques, compilers and operating environments have been infected - yet Jerusalem still works very effectively. That is not to say it does not have its troubles. For example, Jerusalem uses an interrupt which conflicts with a Novel network, so it will not function in that environment very well.

Thus viruses can be more or less adaptable to their environment They can interact with their surroundings, modify them as necessary to promote their survival, and they can adapt to their environment.

Again we can say the same thing of ordinary programs, though. Any common word processor will run in a variety of environments - the more the better. So adaptability is not something unique to living organisms or computer viruses.

1 Note that the concept of order can be mathematically quantified. We'll discuss that more latter.

2 James P. Wesley, Ecophysics (Charles Thomas, Springfield, Illinois: 1974) pp. 36 ff.

Evolution

Most biological organisms seem to have at least some capacity for Darwinian-style evolution. Any viable organism has a progeny - if it does not, it will soon be extinct Generally, the genetic makeup of that progeny is not quite the same as the genetic makeup of the parents. Thus, the genetic composition of a population can change over time. In fact, the genetic composition of a population can be influenced by external factors, as one gene proves to have more survival value than another.

As far as the mechanical properties of life go, I am going to treat evolution as a second order phenomenon. I am doing that because of the abysmal state of current experimental and theoretical evolutionary biology. Given an individual, or a population, there is no way (at present) to determine what it will evolve into, or even whether it has the capacity to evolve into something else at all. This statement may sound somewhat heretical in the ears of the typical modern scientist - and it is, intentionally so.

Certainly, reading most popular literature on evolution, one gets the impression that evolutionary processes are infinitely powerful - that any organism can mutate into any other organism, given a reasonable amount of time and the proper environmental pressures. The fact that no predictions can be had stands in strange counterpoint to such omnipotence.

One might suspect that some genetic codings result in organisms that are much more capable of evolving than others. For example, in the extreme, if every one-nucleotide substitution of a particular coding was immediately lethal, we might expect that organism to have a harder time evolving than one which had 1000 neutral one-nucleotide substitutions and 50 potentially beneficial ones.

When extended to AL, one might expect to find a similar variability - perhaps on an even wider scale. Some artificial organisms may not be realistically capable of any evolution. Others may be capable of far more than what the real-world can support.

Against this backdrop, I think the approach of refusing to call something alive unless it can evolve is rather blind, though this school of thought has strong support among the AL community. It could exclude a wide variety of potentially interesting phenomena, including some life on earth.

Rather, I imagine some sort of evolvability coefficient, $\varepsilon$ , that could be assigned to any self-reproducing automaton, where an $\varepsilon=0$ would mean evolution could not occur, and a large $\varepsilon$ would mean lots of evolution could occur very fast. And I would prefer not to use $\varepsilon$ to determine whether something is alive or not, but to study evolution itself. Although this $\varepsilon$ is obviously very naïve, it helps us to see evolution as a secondary phenomenon. In the abstract, evolution is little more than a study of genetic change. Thus, wherever you have self-reproducing automata, you have evolution. Of course, $\varepsilon=10^{-6}$ is just as much evolution as $\varepsilon=10^6$ is, in this broad sense of the term.

If ordinary evolutionary biology were advanced enough to quantify evolution, we might be able to determine some minimal value of $\varepsilon$ to qualify an artificial organism as "living". Since evolutionary biology is not that advanced, though, it seems rather absurd to use an unquantifiable phenomenon as a primary dividing line between life and non-life, especially when it cannot even be observed in day to day life except on the scale of "microevolution".

With all of that said, an artificial organism might be capable of at least some limited evolution. However it need not be some wonderful seed from which myriads and myriads of increasingly complex artificial organisms could spring forth. To impose the condition of unlimited evolvability on an artificial organism would be inappropriate. Although I do not consider evolution a necessary prerequisite for something to be alive, the question of whether viruses could evolve or not is still interesting.

Evolution of Computer Viruses

When we begin talking about the evolution of computer viruses, it is necessary to clear up some confusion: Polymorphic, or self-mutating viruses are often called evolutionary, although that is not true in the sense of Darwinian evolution.

A polymorphic virus is a virus which encrypts its code differently each time it infects a new program. The primary purpose of this encryption is to defeat virus scanners which search for a string of code in order to locate viruses. The self-mutating virus simply avoids giving the scanner a fixed string to search for by encrypting its code differently each time.

This kind of a mutation scheme - in and of itself - does not fit the model for Darwinian evolution. Instead, it is somewhat like a chameleon camouflaging itself. The polymorphic virus changes what it looks like to the outside world every time it reproduces, but it doesn't change its essential function (unless additional features besides encryption have been added). None of the mutations, in and of themselves, improve the ability of the virus to survive. One mutation is not normally favored over another.

Obviously, a bug in a decryption scheme, or a fluke in an operating system or anti-virus program might give one encryption a better (or worse) ability to survive than another. However every time the virus reproduces, its children look completely different, but operate in essentially the same manner. No "survival of the fittest mechanism" can work in such an environment because the parents don't genetically pass any encryption information on to their children. In short, polymorphism should not be confused with Darwinian evolution.¹

Practically speaking, Darwinian evolution can only operate against a background of relative stability. An organism which passes on most of its characteristics to its progeny is a prerequisite. A few characteristics may change from time to time, resulting in a new genotype. That new organism will then normally pass on its new characteristics to its progeny. If the new organism is more successful than the old (i.e., on the average the new succeeds in reproducing more often than the old) then the relative population of new versus old can increase, and possibly replace the old entirely. Possibly a stable population of both types will result. In this way a population of organisms (or artificial organisms) can evolve from old to new.

It seems reasonable to suggest that genetic change - and the resulting evolution - can be of two types: accidental or pre- programmed. In nature, accidental evolution might be the result of a stray cosmic ray striking some organism and altering its genetic structure. Most such alterations will be immediately lethal, but some might only be harmful. Fewer still might be neutral, and rarely such an alteration might be beneficial. A "beneficial" alteration is essentially defined as one which would have some survival value for the organism. In such a situation, the mutated organism would reproduce successfully, and possibly replace the original in a large number of generations.

The second type of genetic change - pre-programmed - infers that some technique of modifying the genetic structure of an organism from generation to generation is built into its very coding. In the natural world, simple sexual reproduction is a good example. It affords a number of pre-programmed means for effecting genetic change. For example, human genetic information is broken up into 23 separate strands of DNA known as chromosomes, each one of which has an equal chance of coming from the father or mother at conception. In this way, planned change takes place from generation to generation. The child is not normally the same as either parent, but he is similar to both. In addition, a phenomenon called cross-over or chiasma occasionally occurs. When two chromosomes come into close proximity, they can occasionally break apart and combine with each other, so that the child inherits a chromosome which did not belong to either father or mother, but contains a segment from both. Cross-over provides an additional element of genetic flexibility which is evidently built right into the reproduction mechanism.

These pre-programmed means of genetic change give a population a way to adapt to environmental changes in an evolutionary fashion without having to rely on rare accidental mutations. The theory of evolution doesn't particularly say anything about how the genetic change takes place - only about what happens once variations exist.²

Potentially, any computer virus could be subject to accidental mutation. A power glitch while the virus is in memory, or a weak magnetic domain could conceivably change a bit, which would be passed on when replicating. Most such mutations would be disastrous for the virus, and result in a non-functional or crippled piece of code. However, if such a mutation occurred in a piece of dead code space, the virus would simply carry it along from generation to generation, with no ill side-effects. It is even conceivable that once in a while such a mutation could be beneficial.

Actually, the phenomenon of viruses carrying around changes from generation to generation is quite common (though the changes are not usually chance mutations). For example, the Stoned virus carries around with it the partition table of the last hard disk which it resided on.

Additionally, a computer virus can be designed to change itself in such a way that it will undergo some Darwinian evolution. Consider, for example, the INTRUDER virus discussed in The Little Black Book of Computer Viruses. It was a simple virus designed to infect EXE files. It contained a routine SHOULDRUN, which controlled the reproduction rate of the virus. The programmer could set this routine up to make the virus reproduce every time it executed, or only very rarely, There is no reason, however, that one could not design a SHOULDRUN routine which would modify its own reproduction rate. For example, the routine

; This routine returns Z if the virus should replicate
 
RUN_FLG		DB	7
 
SHOULDRUN:
	push	ds
	xor	ax,ax
	mov	ds,ax
	mov	bx,46Ch			;low word of current time
	mov	ax,[bx]			;since it's a fair random #
	pop	ds
	test	ah,1			;mutate or not? 50-50 chance
	jnz	NO_MUTATE		;no
	test	ah,2+4			;increase 75% chance or
	jnz	MUT_UP			;decrease 25% chance rate?
MUT_DOWN:
	shr	[RUN_FLG],1		;reproduction more likely
	jmp	SHORT NO_MUTATE		;by a factor of 2
MUT_UP:
	shl	[RUN_FLG],1		;reproduction less likely
	or	[RUN_FLG],1		;by a factor of 2
NO_MUTATE:
	and	al,[RUN_FLG]		;set Z flag properly
	ret				;and exit

will reproduce at a rate controlled by RUN_FLG. However, every other replication also modifies RUN_FLG randomly, either halving it or doubling it. In this way, over a number of generations, different versions of this virus will come into being, with all different replication rates, no matter what the initial rate was set to. This virus, which we call INTRUDER-II, is on the Program Disk.

INTRUDER-II virus will exhibit Darwinian evolution. Consider, for example, an ideal world in which no anti-virus software exists and everybody shares software with their neighbors. In such a world, the faster the virus reproduces, the more successful it will be, because reproducing slower has no inherent survival value. Thus, the world-wide population of this virus will be dominated by the fastest reproducing varieties, and the slowest reproducing varieties will be rare to non-existent. (See Figure 7.1)

Now, suppose an environmental change took place, and about half the world's PC's had a TSR anti-virus program installed on them which would catch this virus when it activates, but it could not scan for it and catch copies which hadn't activated. Suddenly, a slowly-reproducing version of our virus gains a certain advantage over a quickly reproducing one, because someone with anti-virus software will be less likely to notice it is there. Thus we would expect the population to shift over a period of generations toward the more slowly reproducing varieties. Some typical results are depicted in Figure 7.2. The program which does these calculations, INT_SIM, is included on the Program Disk too.

Figure 7.2: Population of INTRUDER-II with anti-virus.

In conclusion, computer viruses clearly can evolve, and use evolution to overcome challenges to their survival.

1 This does not, of course, mean that a polymorphic engine could not incorporate Darwinian evolution into its operation. In fact, we will discuss just such an engine later in this book.

2 Indeed, Darwin didn't even know about genetics when he proposed evolution. That was discovered by Gregor Mendel in 1866 and ignored by the scientific community, including Darwin, for years.

Conclusions

There are many mechanical aspects of life which contribute to our idea of what life is. I have not discussed all of the properties which have been proposed here, just those which appear most important and most certain. Many biological organisms posses all of these properties, yet exceptions can be found to every one of them. Thus, no one mechanical property can be used as a litmus test to say "this is alive and that is not" However, all of these properties seem closely tied to life and they give us a better idea of what life is, from a mechanical point of view.

There is a caveat though: when discussing the mechanical properties of life, we could not avoid philosophical issues. Trying to draw the line between self-reproduction and physics-driven replication proved more difficult than we imagined. Likewise, the whole idea of emergent behavior appeared somewhat illusory if we started looking at it too hard. We have to wonder, could the "emergence" inside a computer be fundamentally different than the "emergence" in the real world?

Humans are prone to resort to mere appearances to reinforce claims, to establish doctrines, and build models. Yet if we are to be good scientists, we must ruthlessly attack appearances to find out what substance they have. This will be our focus in the second part of this book.

For now, we can at least say that computer viruses fit our mechanical conditions fairly well. Only in the concept of emergent behavior do they appear to fall short That seems to be a limitation of single-user operating systems. Perhaps a multi-tasking OS/2 virus might buy us something in terms of emergent behavior. However, since we weren't intent on a litmus test in the first place, we don't disqualify a virus when it doesn't meet all our criteria (especially those of a questionable nature); neither do we call it alive if it does. We merely say it has stronger or weaker claims to life. And from a mechanical perspective, it seems safe to say that computer viruses have a fairly strong claim to "life".

Part II The Philosophy of Life

The Importance of Philosophy

In beginning our discussion of the philosophical aspects of life, I'd like to go back in time and look at pre-scientific ideas about what life is. At first exposure, such ideas may seem hopelessly antiquated and irrelevant to what we are trying to do. In fact, they are anything but irrelevant Some members of the Artificial Life community believe that if we can develop something functionally equivalent to a living organism, it will become actually alive. Certainly I am not averse to such an idea. Yet our concept of what "actually alive" even means is rooted in non-scientific ideas about what life is.

Science has done a great deal in terms of analyzing the mechanics of life, but its effect on our ideas of what is actually alive has been minimal. The very concept cannot be put in terms accessible to science. Biologists today often assume in the course of their research that a living organism is nothing more than a little ma-chine. After over a century of fighting about that idea, it seems to have become a given among biologists although largely a philosophical idea. As far as biological research is concerned, such questions aren't very important. Science is necessarily limited to studying the mechanics of life. Thus, I wouldguess that the average biologist just doesn't worry about it too much in his day to day work. Yet, in focusing on mechanics, science has not somehow redefined the common idea of what life is. That understanding is still primarily philosophical and religious.

The thought that a carbon-based organism (life) is just a little machine has gained a certain footholdin the minds of modern man. This foothold is tenuous though. That becomes altogether too clear when we turn the question around and ask, "Is my little machine alive?" It may have all the functional characteristics of life... but is it really alive? At this point there is a deep breath of hesitation. No answer is forthcoming. Somehow an honest man, who is not pushing some agenda, dares not give a certain word. Why does he hesitate? The deeper philosophical issues suddenly loom very large. And what are those deeper philosophical issues but the very ancient ideas about life which have persisted for millennia, and which most of us grew up with? So we explore them.

The big surprise we shall encounter is that these ancient philosophical ideas are much more than obsolete baggage to be rid of. They are of pivotal importance to what we are discussing, and they will bring us right back to the problems we had when discussing the mechanical aspects of life - problems like defining the boundary where physics-driven reproduction stops and self-reproduction begins, or the reality of emergent behavior.

Ancient Philosophy and Modern Science

It seems that all men everywhere consider life to be holy in the sense of something deserving deep respect, awe and reverence. Many primitive peoples simply worship living things, or use them to sacrifice to their gods. More sophisticated people work to protect life and preserve the environment Some scientists spend their whole lives studying life or trying to create it themselves, and many religious groups teach that God created it Whether worshiping it or protecting it whether viewing its creation as a goal to be attained, or as an incomparable divine act, men revere life. This attitude seems to be an aboriginal instinct which transcends space and time, language and cultural barriers. Though it may be expressed in many different and sometimes superficially conflicting ways, it is very real and pervasive.

I would not be surprised if every culture that is or ever was expressed this respect in one form or another, and thus had something to say about what life is and how to treat it. Although perhaps many cultures have something valuable to contribute to our understanding of what life is, I will concentrate primarily on western civilization. My reasons are twofold: Firstly, this is the area with which I am most familiar. Secondly, western ideas are important because science was born in western culture, so through it western ideas about life have attained a global importance. And since we are discussing science, western ideas are particularly relevant to our discussion.

The ideas which have informed occidental men about life from ancient times are rooted in Greek philosophy and the Bible. These strands of thought cannot be viewed as separate and antagonistic. Rather, they have been closely intertwined throughout history, and they have often informed and infiltrated one another.

Apart from the statement that God created all life, the Bible has very little to say about the nature of life in general. It says a lot about humanity, but very little about any other organisms. Often people infer that what the Bible teaches about man applies to some extent to other creatures as well, but these ideas are not developed in the text itself. Perhaps the most telling thing we can learn from the text comes from the words used to describe life. In Hebrew, one word for "life" is nefesh¹ which literally means breath. The word usually translated "spirit" is ruach - literally wind. The Bible says that life (nefesh) is in the blood² and that to spill the blood is to spill the life. If we read such terms literally, they appear to be purely functional statements about what life is. However, they clearly have some metaphysical content as well. This subtle marriage of the natural and the supernatural in describing life is not uncommon among the ancients. However, little more than this is spoken generally about living organisms in the Bible.

The Bible - especially the New Testament and some of the later, prophetic writings - paints a vivid picture of a supernatural world with life after death, angels and demons, rewards and punishments, and heaven and hell. However it appears this world belongs solely to God, man, and its other heavenly inhabitants. On the question of where all the other many and varied forms of life in this world fit into that scheme, we draw a blank.

Because the Bible itself is relatively silent regarding the philosophical nature of life in general, western society - even during its most orthodox Christian periods - has largely looked to the more speculative greek tradition for an understanding of what life is.

There are four major schools of thought among the Greeks that are worth taking at least a brief look at These can be broadly categorized as mysticism, atomism, the so-called harmony theories, and the Platonic/Aristotelian school of thought Each had something to say which has formed our thinking about life right down to the present day. Before delving in, however, we need to better understand the words which the Greeks used to talk about life.

Unlike the Hebrews, the Greeks had quite a variety of words at their disposal. There were matter-of-fact terms like bios, which describes a state of life, or manner of living, and zoe, which is a general term for the property of life in an animal, or life as opposed to death. However there are also the more theoretical terms like psuche and thumos, which denote terms closer to our own theological ideas of soul and spirit. The psuche was understood as the "sign of life" or soul, which also (possibly) survived after physical death, but it too, had the meaning of "breath." The thumos was seated in the psuche, and it had to do with the seat of feeling, emotion and thought. Thumos derives from thuo, which suggests a storm, violent motion, rushing wind, or strong desire. Again we find the terms wind and breath closely tied to life, with a deeper metaphysical connotation being quite clear throughout many of the writings of the ancient Greeks. Most of the discussions about life which the greek philosophers carried on revolved around the meaning and nature of the word psuche.

In the most ancient greek mythology, the psuche was something of an entrapped god, a mystical, supernatural being in a body. At death, this being left the body and went to Hades, at least for a period of time. Some believed in reincarnation, where the psuche would return to inhabit another body after aperiod of time, possibly being rewarded or punished in its new life for the sins of the past life. All of this theology focused on man, of course, and it has perhaps little to do with what we are talking about here.

Atomistic thinking, which dates back to Democritus (about 400 BC) and his disciples, Lucretius and Epicurus, was a radical break from greek mythology. They sought to understand the universe in terms of small, indivisible units, called atoms. There were four different kinds of atoms: earth, water, air and fire. In trying to explain everything in terms of these atoms, they of course sought to describe life - psuche - in terms of them too. Since fire was the smallest and most mobile type of atom, and since it was capable of causing motion (e.g. boiling water), the life-force was thought to be a type of fire, interspersed throughout the body. The body would continually lose these soul-atoms, and could maintain the necessary balance for life by inhalation, which brought them back ia Different philosophers came up with variations on this theme. For example, the Epicureans thought the life-force consisted of two parts, one located in a specific area of the body (be it the head, the heart, or somewhere else), and a part which was dispersed throughout the body.

Whatever the variation, atomists taught that the property of life was the result of the presence of a physical substance, or atomic structure. These ideas about life are commonly classified as materialist or substantialist theories.

The so-called harmony theory of life, usually associated with Philolaus and Empedocles, is also materialist in a sense. However its proponents did not view psuche as a material substance, but rather as a harmony, or proper ratio of material substances in the body. Like a musical instrument properly tuned, or a properly mixed chemical reaction (to use modem terms), a body could live if its substance was properly balanced and working together. In this school of thought, the psuche is not a material substance, but neither is it a transcendent supernatural entity.

Plato and Aristotle formulated a sophisticated challenge to the materialist/substantialist theory of life. Plato seems rather confusing on the subject of life. As a moral philosopher, his primary concern was for man, and most of his discussions revolve around man, In The Last Days of Socrates, he lays out elaborate plans in whichihepsuche survives death, goes to Hades, etc., etc., obviously borrowing from greek mythology. In the Timaeus, he espouses a form of Pythagorean number magic, in which the psuche is described in terms of numbers and their properties.

Plato's ideas may not be very clear, or well developed, yet he laid a groundwork which Aristotle was able to build on Plato was the first to divide the psuche into different parts. He did this differently at different times. For example, he divided it into reason and emotion, or into a higher, immortal part, and a lower, mortal part. He also developed the theory of forms which Aristotle used extensively in trying to understand the world. Most importantly, Plato was clearly searching for a non-materialist understanding of life. He did not divide the psuche into atoms, or physical parts, but saw it more in terms of psychological or mental states.

Aristotle was perhaps the most prolific writer on the subject of life. He wrote a number of books on the subject, examining all different kinds of life-forms, describing their physiognomy, their habits, classifying them and philosophizing about what life is. Aristotle approached the subject of life much like a modern biologist, seeking a general concept of what life is. However, as a student of Plato's, he sought a non-materialist understanding of life. He used Plato's ideas about forms as the basis for his work, rather than indulging in the same kinds of metaphysical and mythological speculations that his teacher did.

Aristotle's central work on the subject of what life is, is commonly known by its Latin title De Anima.³ E.g., "On the psuche", usually translated into English as "On the Soul." He describes a living, ensouled being as something which can produce movement and which may have intellect or perception. He breaks movement down into spatial movement, and movement connected with nourishment, growth, and perception. The psuche - the life principle - is then a form with these properties. When matter is endowed with this form it is alive. Aristotle suggested that living organisms could have different kinds of psuches. For example, plants have a psuche which admits of nutritive and reproductive motion, but not locomotive motion or perception.

At this point, you may be wondering what a form is. Good. You ought to be. Think of the word "form" in the sense of a mold which defines the character, shape and function of an object But also think of it as an archetypal idea. For example, a carpenter has an idea in mind when he makes a chair or a table. That idea is like a mold, which defines the shape of the object being constructed. Normally, when we think of an idea, though, we imagine it as being someone's idea, be it our own, or another person's, or even God's. The idea is a phenomenon of mind. In contrast, Plato and Aristotle seemed to view these forms as fundamental realities. They were not a phenomenon of this natural world, or of some supernatural world. They were not someone's idea, but rather Ideas which existed whether anyone thought of them or not.

Yet one cannot go so far as to say that these forms were the only ultimate realities in the thinking of Plato and Aristotle. For example, Aristotle believed in a God who was the prime mover in the universe - the origin and source of all things and all activity. What ultimate reality consisted of for these philosophers is often unclear, because they don't lay down a plan in any authoritative manner and stick to it. Since they did not depend on the authority of some scriptural cannon, they spoke of what seemed reasonable to them. And that changed from time to time.

Plato's mysticism makes his idea of life appear rather clearly supernatural. Aristotle, by staying away from mysticism, gives us a picture of life which is not purely natural, like the atomists, yet it is not supernatural in the sense we normally think of supernatural. For Aristotle, there was a form - an idea - behind each living organism. This form was not a part of the material world, and yet it defined the qualities of an object in the material world. It is not the fact that an object had a form associated to it that made it alive, but the nature of that form. Everything had forms associated to them in Plato and Aristotle's system. A chair did. A rock did. A lizard did. However, only the lizard had a form that imparted to it the ability to move, grow, procreate, and perceive what was happening to it That form made it alive.

Now, as I said before, these ancient ideas were not simply isolated schools of thought. Platonic philosophy had a great influence both within the church and in Jewish thought in ancient times. The Christian world-view ascended to predominance as the classical world collapsed (intellectually and politically) between the 4th and 6th centuries AD. Aristotle, whose works had been lost to the west for centuries, was rediscovered by way of Islamic culture in the 1 lth and 12th centuries, and hotly debated by theologians like Thomas Aquinas and Averroes. These thinkers effectively integrated Aristotelian and Christian thought With the revival of classical culture during the renaissance, one finds strong elements of both mysticism and atomism making their way into western thought So these different schools of thought have informed each other and antagonized each other throughout history.

In summary, then, the ancient greeks have handed down the following conflicting ideas to us:

The property of life is a natural phenomenon, either (a) a substance present in. the body, or (b) a proper organization of the substance of the body.
The property of life is not a natural phenomenon, it is either (a) a mystical, spiritual property, or (b) the result of a form or idea imposed upon the matter.

While the Bible is vague about life in general, its view of man is of the spiritual and mystical nature. Yet its view of life apart from man is perhaps not too different from Aristotle's line of thinking. Genesis says God created the different animals after their kind (Hebrew meen) several times.⁴ One is tempted to substitute the word form here. It seems to fit The text obviously suggests there was an idea of some type behind the making of each of the various creatures. Of course, one can not really view the biblical idea of kind in terms of some sort of self-existent Idea. The idea is clearly God's idea. None the less, the thought that there is an idea behind the organism sounds very Aristotelian.

So now we have an idea about what the ancients had to say about life. Obviously we could delve into the subject in great detail. Instead, I simply refer the reader to Hans Regnell's Ancient Views on the Nature of Life.⁵

Modern Science

The debate over the nature of life which began in ancient Greece thousands of years ago has continued right up to the present day. A superficial examination of modem science, and scientist's opinions might lead one to conclude that the atomistic school of though has won an astounding victory over Plato and Aristotle and over the mystics. Our understanding of the microscopic details of the universe have changed dramatically since ancient times. And most modern scientists, if questioned about it, would probably favor the idea that all phenomena in the universe - life included - could in principle be reduced to basic physics. This is the essence of the atomistic philosophy.

However this pervasive idea that atomism is ultimately correct is a matter of blind faith to most scientists, not proven fact. It is almost a sort of touchstone for the brotherhood of scientists, a measuring stick for how "scientific" some idea is. But in actual practice, atomism is often quickly jettisoned as a practical tool for understanding our world in favor of a much more Aristotelian approach. This is a fundamental paradox of modern science.

Let me illustrate this paradox with an example from physics that strikes to the core ofwhatl'm talking about. Up until the 1920's Newtonian mechanics was supposed to be the correct description of how all things worked, from subatomic particles to galaxies. The fundamental law of Newtonian mechanics is

$F=ma\hspace{100}10.1$

Force equals mass times acceleration. Apply a force F to an object with mass m and it will accelerate according to this law. In the 1920's quantum mechanics was formulated to account for experimental deviations from Newtonian mechanics in the subatomic realm The fundamental law of quantum mechanics is the Schrödinger equation,

$ih\frac{\partial}{\partial{t}}\Psi=H\Psi \hspace{100}10.2$

This says that the wave function $\Psi$ - which is a sort of probability distribution - evolved in time proportionally to the action of the Hamiltonian operator on it. These two laws are radically different One might wonder, can the law $F=ma$ be derived from the Schrödinger equation? Since everything is made up of subatomic particles, one would expect the Schrödinger equation to be somehow more fundamental. Thus, one would expect Newton's laws to result from a proper reduction of the Schrödinger equation. Such a reduction has been accomplished to a certain extent One can mathematically show that properly formed wave packets do obey Newton's law, within limits. Yet the problem goes much deeper. The whole way of understanding how the world works in quantum mechanics is radically different than in Newtonian mechanics. Quantum theory is a world of probabilities; Newton's is a world of deterministic certainty. How are the two reconciled? This is a deep question which has not been resolved in seventy years. To deal with probabilities, quantum theory introduces the idea of the observer as separate from a physical system under observation. The observer accomplishes what is called state vector reduction when he makes measurements on the system he's observing to determine its state. This observer is fundamental to the whole of quantum theory. If you try to do away with him, and incorporate him into the wave-function, you come up with nonsense.

Erwin Schrödinger triedto confront this problem by proposing a thought-experiment in which a cat is put in a box, along with a vial of cyanide which would be broken based on whether a single photon wa sreflected from a 50% silvered mirror. If the photon was reflected, the vial would be broken and the cat would die. If the photon was transmitted, the vial would not be broken, and the cat would live. According to quantum theory, the probability that the photon would be reflected was 50%. So, arguably, inside the box, you'd have a wave function in which the cat was 50% alive and 50% dead. With no observer, you had not accomplished state vector reduction, so this half-alive and half-dead state should persist until you opened the box and looked to see what happened. This obviously borders on insanity, but that is exactly what quantum mechanics would predict! You could even put a man in a space suit in the box, and he would be part of the system. What would he observe? Would he have a double existence, in which half saw the cat dead, and half saw it alive? To date, no one has resolved these questions.

So we have some deep and fundamental questions: How is it that a mechanistic theory about subatomic particles requires us to invoke the idea of an intelligent observer? And how is it that the determinism of Newtonian mechanics is reconciled with probabilistic quantum mechanics only by invoking such an intelligence? These are not easy questions. And they haven't been fully answered in seventy years. Yet scientists do not shy away from using both Newtonian and quantum mechanics. One cannot properly say that quantum mechanics is complete without the idea of an observer. And that is about as non-atomistic as you can get Likewise, if you can't build a solid connection between Newtonian and quantum mechanics, you can hardly deny that Newtonian mechanics leaves something to be desired in atomistic, reductionistic terms. Its very determinism can't be rooted in atomistic reality!

I chose this example because it may very well be that there is a rigorous solution to the problem. For example, Roger Penrose suggests that the answer may lie in a proper understanding of quantum gravity.⁶

VX Heavens