Collected Essays

by Rucker, Rudy

  It is conceivable that somewhere in the universe there may be things with phenomes that we would call living, but which are not grown from genomes. These geneless aliens might be like clouds, say, or like tornadoes. But all the kinds of things that we ordinarily think of as being alive are in fact based on genomes, so it is reasonable to base our investigations of A-Life on systems which have a genetic basis.

  If we’re interested in computer-based A-Life, it is particularly appropriate to work with A-Life forms whose phenomes grow out of their genomes. In terms of a computer, you can think of the genome as the program and the phenome as the output. A computer A-Life creature has a genome which is a string of bits (a bit being the minimal piece of binary information, a zero or a one), and its phenome includes the creature’s graphic appearance on the computer’s screen. Keep in mind that the phenome also includes behavior, so the way in which the creature’s appearance changes and reacts to other creatures is part of its phenome as well.

  Reproduction

  The second sex topic is reproduction.

  The big win in growing your phenome from a small genome is that this makes it easy for you to grow copies of yourself. Instead of having to copy your large and complicated phenome as a whole, you need only make a copy of your relatively small genome, and then let the copied genome grow its own phenome. Eventually the newly grown phenome should look just like you. Although this kind of reproduction is a solitary activity, it is still a kind of sex, and is practiced by such lowly creatures as the amoeba.

  As it happens, the genome-copying ability is something that is built right into DNA because of the celebrated fact that DNA has the form of a double helix which is made of two complementary strands of protein. Each strand encodes the entire information of the genome. In order to reproduce itself, a DNA double helix first unzips itself to produce two separate strands of half-DNA, each of which is a long, linked protein chain of molecules called bases. The bases are readily available in the fluid of any living cell, and now each half-DNA strand gathers unto itself enough bases to make a copy of its complementary half-DNA strand. The new half-DNA strands are assembled in position, already twined right around the old strands, so the net result is that the original DNA genome has turned itself into two. It has successfully reproduced; it has made a copy of itself.

  In most A-Life worlds, reproduction is something that is done in a simple mechanical way. The bitstring or sequence of bits that encodes a creature’s program is copied into a new memory location by the “world” program, and then the two creature programs are run and the two phenotypes appear.

  Mating

  The third sex topic is mating.

  Most living creatures reproduce in pairs, with the offspring’s genome containing a combination of the parents’ genomes. Rather than being a random shuffling of the bases in the parents’ DNA, genomes are normally mated by a process known as crossover.

  To simplify the idea, we leave out any DNA-like details of genome reproduction, and simply think of the two parent genomes as a chain of circles and a chain of squares, both chains of the same length. In the crossover process, a crossover point is chosen and the two genomes are broken at the crossover point. The broken genomes can now be joined together and mated in two possible ways. You can have squares followed by circles, or circles followed by squares. In real life, only one of the possible matings is chosen as the genome seed of the new organism.

  In computer A-Life, we often allow both of the newly mated genomes to survive. In fact, the most common form of computer A-Life reproduction is to replace the two original parent programs by the two new crossed-over programs. That is to say, two A-Life parents often “breed in place.”

  In a world where several species exist, it can even sometimes happen that one species genome can incorporate some information from the genome of a creature from another species! This phenomenon is called “exogamy”. Although rare, exogamy does seem to occur in the real world. It is said that snippets of our DNA are identical to bits of modern cat DNA. Gag me with a hairball!

  Mutation, Transposition and Zapping

  The fourth sex topic involves random changes to the genome.

  Mating is a major source of genetic diversity in living things, but genomes can also have their information changed by such randomizing methods as mutation, transposition, and zapping. While mating acts on pairs of genomes, randomization methods act on one genome at a time.

  For familiar wetware life forms like ourselves, mutations are caused by things like poisons and cosmic rays. Some mutations are lethal, but many of them make no visible difference at all. Now and then a particular mutation or accumulation of mutations will cause the phenome to suddenly show a drastically new kind of appearance and behavior. Perhaps genius, perhaps a harelip, perhaps beauty, perhaps idiocy.

  In the A-Life context, where we typically think of the genome as a sequence of zeroes and ones, a mutation amounts to picking a site and flipping the bit: from zero to one, or from one to zero.

  Besides mutation, there are several other forms of genome randomization, some of which are still being discovered in the real world and are as yet poorly understood.

  One interesting genome changer is known as transposition. In transposition, two swatches of some genomes are swapped.

  Another genome randomizer that we sometimes use in A-Life programs is zapping, whereby every now and then all of some single creature’s genome bits are randomized. In the real world, zapping is not a viable method of genetic variation, as it will almost certainly produce a creature that dies instantly. But in the more forgiving arena of A-Life, zapping can be useful.

  In the natural world, species typically have very large populations and big genomes. Here the effects of mating—sexual reproduction—are the primary main source of genetic diversity. But in the small populations and short genomes of A-Life experiments, it is dangerously easy for all the creatures to end up with the same genome. And if you crossover two identical genomes, the offspring are identical to the parents, and no diversity arises! As a practical matter, random genome variation is quite important for Artificial Life simulations.

  Death

  What would life like if there were no death? Very crowded or very stagnant. In imagining a counterfactual situation like no death, it’s always a challenge to keep a consistent mental scenario. But I’m a science fiction writer, so I’m glad to try. Let’s suppose that Death forgot about Earth starting in the Age of the Dinosaurs. What would today’s Earth be like?

  There would still be lots of dinosaurs around, which is nice. But if they had been reproducing for all of this time, the dinosaurs and their contemporaries would be piled many hundreds of meters deep all over Earth’s surface, in fact they would fill all known space. Twisted and deformed dinosaur mutations would be plentiful as well. One might expect that they the dinos have eaten all the plants up, but of course there would be no death for plants either, so there would be a huge jungle of plants under the mounds of dinosaurs, all of the dinos taking turns squirming down to get a bite. The oceans would be gill to gill with sea life, and then some. I think of the Earth before Noah’s flood.

  Would mammals and humans have evolved in such a world? Probably not. Although there would be many of the oddball creatures around that were our precursors, in the vast welter of life there would be no way for them to select themselves out, get together, and tighten up their genomes.

  An alternative vision of a death-free Earth is a world in which birth stops as well. What kind of world would that lead to? Totally boring. It would be nothing but the same old creatures stomping the same old environment forever. Like how the job market looks to a young person starting out!

  Meaningless proliferation or utter stagnancy are the only alternatives to death. Although death is individually terrible, it is wonderful for the evolution of new kinds of life.

  Evolution is possible whenever one has (1) reproduction, (2) genome variation, and (3) natural selection. We’ve already talke
d about reproduction and the way in which mating and mutation cause genome variation—so that children are not necessarily just like their parents. Natural selection is where death comes in: not every creature is in fact able to reproduce itself before it dies. The creatures which do reproduce have genomes which are selected by the natural process of competing to stay alive and to bear children which survive.

  What this means in terms of computer A-Life is that one ordinarily has some maximum number of memory slots for creatures’ genomes. One lets the phenomes of the creatures compete for a while and then uses some kind of fitness function to decide which creatures are the most successful. The most successful creatures are reproduced onto the existing memory slots, and the genomes of the least successful creatures are erased.

  Nature has a very simple way of determining a creature’s fitness: it manages to reproduce before death or it doesn’t. Assigning a fitness level to competing A-Life phenomes is a more artificial process. Various kinds of fitness functions can be chosen on the basis of what kinds of creatures one wants to see evolve. In most of the experiments I’ve worked on, the fitness is based on the creatures’ ability to find and eat food cells, as well as to avoid “predators” and to get near “prey”.

  So far in this essay we’ve talked about life in terms of three general concepts: gnarl, sex, and death. Computer A-Life research involves trying to find computer programs which are gnarly, which breed, and which compete to stay alive. Now let’s look at some non-computer approaches to Artificial Life.

  Biological A-Life

  In this section, we first talk about Frankenstein, and then we talk about modern biochemistry.

  Frankenstein

  The most popular fictional character who tries to create life is Viktor Frankenstein, the protagonist of Mary Shelley’s 1818 novel, Frankenstein or, The Modern Prometheus.

  Most of us know about Frankenstein from the movie versions of the story. In the movie version, Dr. Frankenstein creates a living man by sewing together parts of dead bodies and galvanizing the result with electricity from a thunderstorm. The original version is quite different.

  In Mary Shelley’s novel, Baron Viktor Frankenstein is a student with a deep interest in chemistry. He becomes curious about what causes life, and he pursues this question by closely examining how things die and decay—the idea being that if you can understand how life leaves matter, you can understand how to put it back in. Viktor spends days and nights in “vaults and charnel-houses,” until finally he believes he has learned how to bring dead flesh back to life. He sets to work building the Frankenstein monster:

  In a solitary chamber…I kept my workshop of filthy creation: my eyeballs were starting from their sockets in attending to the details of my employment. The dissecting room and the slaughter-house furnished many of my materials; and often did my human nature turn with loathing from my occupation…Who shall conceive the horrors of my secret toil, as I dabbled among the unhallowed damps of the grave, or tortured the living animal to animate the lifeless clay?

  Finally Dr. Frankenstein reaches his goal:

  It was on a dreary night of November, that I beheld the accomplishment of my toils. With an anxiety that almost amounted to agony, I collected the instruments of life around me, that I might infuse a spark of being into the lifeless thing that lay at my feet. It was already one in the morning; the rain pattered dismally against the panes, and my candle was nearly burnt out, when, by the glimmer of the half-extinguished light, I saw the dull yellow eye of the creature open; it breathed hard, and a convulsive motion agitated its limbs…The beauty of the dream vanished, and breathless horror and disgust filled my heart.

  The creepy, slithery aspect of Frankenstein stems from the fact that Mary Shelley situated Viktor Frankenstein’s A-Life researches at the tail-end of life, at the part where a living creature life dissolves back into a random mush of chemicals. In point of fact, this is really not a good way to understand life—the processes of decay are not readily reversible.

  Biochemistry

  Contemporary A-Life biochemists focus on the way in which life keeps itself going. Organic life is a process, a skein of biochemical reactions that is in some ways like a parallel three-dimensional computation. The computation being carried out by a living body stops when the body dies, and the component parts of the body immediately begin decomposing. Unless you’re Viktor Frankenstein, there is no way to kick-start the reaction back into viability. It’s as if turning off a computer would make its chips fall apart.

  The amazing part about real life that it keeps itself going on its own. If anyone could build a tiny, self-guiding, flying robot he or she would a hero of science. But a fly can build flies just by eating garbage. Biological life is a self-organizing process, an endless round that’s been chorusing along for hundreds of millions of years.

  Is there any hope of scientists being able to assemble and start up a living biological system?

  Chemists have studied complicated systems of reactions that tend to perpetuate themselves. These kinds of reaction are called autocatalytic or self-exciting. Once an autocatalytic reaction gets started up, it produces by-products which pull more and more molecules into the reaction. Often such a reaction will have a cyclical nature, in that it goes through the same sequence of steps over and over.

  The cycle of photosynthesis is a very complicated example of an autocatalytic reaction. One of the simpler examples of an autocatalytic chemical reaction is known as the Belusov-Zhabotinsky reaction in honor of the two Soviet scientists who discovered it. In the Belusov-Zhabotinsky reaction a certain acidic solution is placed into a flat glass dish with a sprinkling of palladium crystals. The active ingredient of litmus paper is added so that it is possible to see which regions of the solution are more or less acidic. In a few minutes, the dish fills with scroll-shaped waves of color which spiral around and around in a regular, but not quite predictable, manner.

  A Belusov-Zhabotinsky pattern in a cellular automaton.

  There seems to be something universal about the Belusov-Zhabotinsky reaction, in that there are many other systems which behave in a similar way: generating endlessly spiraling scrolls. It is in fact fairly easy to set up a cellular-automaton-based computer simulation that shows something like the Belusov-Zhabotinsky reaction—Zhabotinsky scrolls are something that CAs like to “do.”

  As well as trying to understand the chemical reactions that take place in living things, biochemists have investigated ways of creating the chemicals used by life. In the famous 1952 Miller-Urey experiment, two scientists sealed a glass retort filled with such simple chemicals as water, methane and hydrogen. The sealed vessel was equipped with electrodes that repeatedly fired off sparks—the vessel was intended to be a kind of simulation of primeval Earth with its lightning storms. After a week, it was found that a variety of amino acids had spontaneously formed inside the vessel. Amino acids are the building blocks of protein and of DNA—of our phenomes and of our genomes, so the Miller-Urey experiment represented an impressive first step towards understanding how life on Earth emerged.

  Biochemists have pushed this kind of thing much further in the last decades. It is now possible to design artificial strands of RNA which are capable of self-replicating themselves when placed into a solution of amino acids; and one can even set a kind of RNA evolution into motion. In one recent experiment, a solution was filled with a random assortment of self-replicating RNA along with amino acids for the RNA to build with. Some of the molecules tended to stick to the sides of the beaker. The solution was then poured out, with the molecules that stuck to the sides of the vessel being retained. A fresh food-supply of amino acids was added and the cycle was repeated numerous times. The evolutionary result? RNA that adheres very firmly to the sides of the beaker.

  The RNA evolution experiment is described in Gerald Joyce, “Directed Molecular Evolution,” Scientific American, December, 1992. A good quote about wetware appears in Mondo 2000: A User’s Guide to the New Edge, edi
ted by R. U. Sirius, Queen Mu and me for HarperCollins, 1992. The quote is from the bioengineer Max Yukawa:

  Suppose you think of an organism as being like a computer graphic that is generated from some program. Or think of an oak tree as being the output of a program that was contained inside the acorn. The genetic program is in the DNA molecule. Your software is the abstract information pattern behind your genetic code, but your actual wetware is the physical DNA in a cell.

  Genetic engineers are improving on methods to tinker with the DNA of living cells to make organisms which are in some part artificial. Most commercially sold insulin is in fact created by gene-tailored cells. The word wetware is sometimes used to stand for the information in the genome of a biological cell. Wetware is like software, but its in a watery living environment. The era of wetware programming has only just begun.

  Robots

  In this section we compare science fiction dreams of robots to robots as they actually exist today. We also talk a bit about how computer science techniques may help us get from today’s realities to tomorrow’s dreams.

  Science fiction Robots

  Science fiction is filled with robots that act as if they were alive. Existing robots already possess such life-like characteristics as sensitivity to the environment, movement, complexity, and integration of parts. But what about reproduction? Could you have robots which build other robots?

  A robot that reproduces by (a) using a blueprint to (b) build a copy of itself, and then (c) giving the new robot a copy of the blueprint. (Drawing by David Povilaitis.)

  The idea is perhaps surprising at first, but there’s nothing logically wrong with it. As long as a robot has an exact blueprint of how it is constructed, it can assemble the parts for child robots, and it can use a copying machine to give each child its own blueprint so that the process can continue. For a robot, the blueprint is its genome, and its body and behavior is its phenome. In practice, the robots would not use paper blueprints, but might instead use CAD/CAM (computer aided design and manufacturing) files.