If we could come up with the next breakthrough technology by sitting around thinking, “What will the next break-through technology be?” then we would all already be rich. So where do the great ideas come from? Some business classes or advice workshops will tell you to “solve a problem.” They say you should identify a need, and then fulfill it. That is a solid approach that has produced adequate results in some cases.
But I would like to talk about another approach: the contemplation of big, abstract philosophical ideas. Many young technology entrepreneurs have little patience with abstract scholarship because they don’t see the line that connects the Philosophy classroom to the Killer App. But the connection is there. Sometimes a single big, abstract question can generate dozens of technological breakthroughs that spread outward into areas ranging from medical research to video games.
In order to give you a sense of the power of a single philosophical idea, I’d like to tell you the story of a field of science that has had a huge impact on technology over the last two decades, but that you probably have never heard of: artificial life.
Santa Fe: evolution of an idea
The theoretical discipline of artificial life was born in the early 1990s. The character of the field was initially defined in a series of conferences in Santa Fe, New Mexico. Attending these conferences were biologists, mathematicians, engineers, programmers and philosophers. All were interested in talking about one of the most elusive and abstract topics that science has ever addressed: the definition of life itself. Their mission was focused on using different disciplines and perspectives to answer a few key questions about life. What are the defining properties of life? Can the defining properties of life be understood in an abstract way, independent of the details of biological life as it appears here on Earth? Can we re-create the defining properties of life in other things, like computer programs or robotic machines? One of the fathers of the field, Christopher Langton, said that while biology is the study of “life as we know it,” artificial life would be the study of “life as it could be.”
I had the opportunity, as a university student in 1992, to attend one of the early artificial life conferences in Santa Fe. Many of the presentations were focused on attempts to simulate the characteristics of living things. There were programmers who worked in digital graphics, demonstrating life-like flocking and herding patterns in simulated birds and animals, or demonstrating new algorithms inspired by biological growth mechanisms that gave rise to intricate and realistic pictures of plants growing in a field. Programs inspired by the study of swarms of insects were used to demonstrate that very complex behaviour could emerge when you have hundreds of very small, simple parts interacting using simple rules.
But there was also an energy at the conference that was driven by a purpose that went above and beyond mere simulation. I was lucky enough to hear presentations that talked about a program called “Tierra” by Thomas Ray. Tierra is a computer environment where small programs copy themselves in memory with small “mutations” that allow them to evolve over time.
This could be interpreted as a simplified simulation of an evolutionary process – but for Ray, Tierra was not a simulation. For Ray, and for many others at the conference, the ability to reproduce and evolve over time was the critical defining property of what it means to be alive. He called Tierra “an evolutionary approach to synthetic biology,” and he felt that the little programs evolving in the memory of his computer were a form of life.
There were more collections of synthetic “creatures” at the conference, appearing in many forms: robots that built other robots, or communities of simple “agents” trading information and resources in electronic worlds. The affection that these scientists felt for their creations was palpable. Although not all of them were entirely serious about the idea that they were creating actual living things, they clearly enjoyed thinking of their experimental worlds as sort-of alive, and thinking of themselves as progenitors.
There may have been a few people at the conference who had aspirations to the practical: people discussing genetic algorithms as a way of automating software development, for example, or people talking about using “swarm simulations” to control the flow of crowds at public events. But for the most part, my overwhelming impression of the conference was of ivory-tower misfits and idealists tackling the age-old question: what makes living things alive?
I was not alone in this impression. Anthropologist Stefan Helmreich spent a number of years living with scientists researching artificial life at the Santa Fe Institute in the 1990s. His book about his experiences there, Silicon Second Nature, goes into detail about the philosophical energy of the movement. Researchers in this community saw themselves strictly as scientists performing scientific research, but their language had strong philosophical overtones: they often described themselves as “gods” of the worlds that they had created.
Although they would approach the topic from varying disciplines and use different methods, the underlying drive of the research was about the ultimate abstract goal: the quest to define life itself.
So how did they do? According to Helmreich, artificial life as a philosophical endeavor did not fare well. “As far as biologists have been concerned, artificial life never really delivered on its promise to reboot theoretical biology,” Helmreich says. “The intersection of biology and computation has become much more workaday than epistemologically revolutionary. Bioinformatics has become where it’s at—and there’s very little bio-theory there.” It’s certainly true that artificial life never got the foothold in academia that (for example) artificial intelligence did.
“The people who were heavily involved in the artificial life conferences in Santa Fe in the 1990s have migrated their work to more traditional fields, such as systems biology or computer science. Stefan concludes, “I think it is fair to say that artificial life, in its Santa Fe incarnation, is over.”
Yet many of the practices and ideas that were explored in the Santa Fe conferences have been exported quite broadly into technology in the world today. From self-replicating robots, we now have configurable computer hardware that simulates self-organisation of biopolymers, or electrical networks that can optimise themselves by self-assembly. From the initial toy models of insect swarm behaviour, we now have complex distributed control programs that are able to perform real-time optimisation of traffic light patterns.
The flocking and herding simulations presented at those early artificial life conferences established many of the basic computational methods that are used in everything from movie CGI to video games to render realistic images of crowd behaviour. If you are doing scientific research, you can even go to the Cornell Creative Machines Lab and download an “automated scientist” that uses mathematical equations that evolve – much like Ray’s Tierra programs – to solve complex mathematical problems.
So although the scientists once located in Santa Fe have since dispersed, the work that they did on the “big philosophical questions” has produced many fruits in the technology world.
Paris: the importance of autonomy
There is a second thread to the story of artificial life. While the Sante Fe researchers focused primarily on the evolution of populations in their philosophical quest to understand (and create) life, a European school of thought was forming at the same time with a different view. Coming from a different epistemological background, researchers in Europe were more interested in what they called the autonomy of living systems: the fact that a living organism is one that is constantly in a state of self-regulation.
Living systems have a structure centered on monitoring, creating, and maintaining itself in the face of a changing environment. According to this school of thought, the necessary and sufficient defining characteristic of living things is that they are autopoietic: a term that comes from Chilean biologist Humberto Maturana and literally means “self-producing.”
The first European Conference on artificial life was held in Paris, France. The book that was eventually published from the papers presented there was called Toward a Practice of Autonomous Systems. In the introduction to that book, Paul Bourgine and Francisco Varela drew an explicit contrast between their own approach and the evolutionary approach championed in the United States. “We think artificial life can be better defined as a research program concerned with autonomous systems, their characterisation and specific modes of viability.
This view is not in contradiction with the [evolutionary view]; rather, we claim that the foregoing definition needs to be more precise, by focusing on those dynamical processes that assure the key features of autonomy more than any other dynamic properties present in living systems.” In other words: evolution is a property of living systems, and one that is worth studying, but what matters is autopoiesis.
Matthew Egbert teaches a course on artificial life at the University of Sussex, and is an active participant in the most recent incarnations of the European Conference on Artificial Life, which is still an annual event. When we discussed the philosophy of artificial life in Europe, he gave a passionate explanation of the difference between the evolutionary view and the autopoietic view of life:
One motivation in the study of autopoiesis is the desire to formulate a definition of life that does not focus upon the evolutionary history of how an organism came to be. Intuitively, it makes sense that a living system should be considered alive whether or not it is the product of evolution. If you had two identical elephants, one that came about due to natural evolution, and the other was manufactured by some elephant-production industrial plant, or cloned by some Star Trek-esque device, and if the two elephants were identical such that we were completely unable to distinguish between them, we’d have to say that they are both alive – right?
Similarly, just because something is the product of evolution, it does not follow that it is alive. Scientists and engineers use genetic algorithms to simulate a simple form of evolution on the computer to artificially evolve solutions to problems. The solutions are the product of a kind of evolution, but they are, of course, in no way alive. So, I would argue, being “alive” is a property of the way a system is organised, not how it came to be.
Although it is tempting to see these as two approaches as competing ideas about the basic definition of life, Egbert stops short of this combative interpretation. “It’s not like these are two bodies of incompatible theory. They are supplementary theoretical frameworks that we use to study life and adaptive behaviour. They complement one another well.”
Yet because of these differences in the “philosophy of life,” very different research agendas and technological tools have emerged from the European artificial life community. This research has focused on the creation of completely artificial “proto-cells” or simulating cell metabolism. They still acknowledge the importance of evolution as one of the behaviours of living things, but generally the research focuses on how evolution might emerge from, or be altered by, a living cell’s structure or metabolism.
This focus can be seen in recent pop-science announcements, such as last year’s announcement that scientists at Glasgow University were trying to form synthetic cells from inorganic molecules. The initial exciting discovery was that their cellular structure exhibited properties similar to those of organic cell membranes; the secondary quest was to see if it would be possible to get them to evolve.
The evolution of artificial metabolisms and artificial life based on the idea of self-production has a number of applications that are just beginning to be explored, ranging from materials science to medicine. I asked Matthew Egbert about the future technological promise of autopoietic theory. “There are labs all around the world that are working to design and create the first artificial cell,” he said. “Many of these groups continue to focus on evolution as a necessary feature of synthetic life, but increasingly, labs are considering the autopoietic organisation as important.
“I think that theory around autopoiesis and self-sensitive (or autopoiesis-based) behaviour can help these labs develop synthetic life that can be more robust and adaptive. In the longer term, these self-maintaining artificial cells might help us in medicine or in cleaning up the environment. Imagine, for instance, artificial cells that eat toxic materials in garbage dumps, consolidating them or converting them into a neutral waste.”
Artificial life moves forward
The 2011 European Conference on Artificial Life saw a merging of the American and European approaches to living systems. Dr. Carlos Gershenson, the head of the computer science department at Universidad Nacional Autónoma de México, feels that most current researchers interested in artificial life take a more unified, practical approach. “The two sides have slowly merged, in the sense that many people attended both the Santa Fe and European conferences in recent years. I think now you can find more diversity within Europe and USA than between them in a particular area.”
Nobody, apparently, is still arguing over the “ultimate definition” of life. Instead, the focus is on dynamics: the creation artificial entities that repair themselves, the way that swarms produce collective behaviour, and how evolution produces complexity. Much of the research looks at the interaction between these different elements of “living dynamics”.
For example, Matthew Egbert’s research looks at how autopoietic organisation can actually improve evolutionary adaptation, and Carlos Gershenson investigates self-organisation both within organisms and between organisms in social systems. Cutting edge applications range from the manufacturing of synthetic cells, to autopoietic models for intrusion detection inspired by models of the immune system, to the use of evolving programs that predict the stock market.
Does philosophy still play a role in driving research? Perhaps not for everybody, but it definitely does for some. Egbert, for one, firmly believes in the importance of understanding life through the lens of autopoiesis and autonomy:
It’s not uncommon these days to look at the news headlines and see that they’ve found a gene for so-and-so. I think I’ve even seen an announcement that they’d discovered a “gene for happiness”! This idea of genes predetermining and controlling our lives is an idea that takes away control and responsibility from us as individuals. Much of the study of autopoiesis is about understanding autonomy of life. Living systems, to an extent, make and break their own rules.
We are clearly not under the “remote control” of our genes, and we are emphatically irrational and unlike computers. Where does this come from? How can such autonomous systems exist? What are we?
Understanding autonomy is important not only for scientists or philosophers, but as Maggie Boden wrote in her 1996 article Autonomy and Artificiality, “what science tells us about human autonomy is practically important, because it affects the way in which ordinary people see themselves, which includes the way in which they believe it is possible to behave.”
So in the end, the tale of artificial life is a lesson in the importance of lifting your gaze above the entrepreneurial attitude of “fulfilling the next need” or searching for that perfect app to develop. We live in a pragmatic world, with young people fixed on the prize of inventing a gadget that everyone will love. But I would like to be an advocate for an alternative approach.
I entreat you to step back, from time to time at any rate, and ask the big abstract questions. It can be fun. It can be exhilarating. And, who knows, it may even end up leading to that next great breakthrough in technology for you, too.