Ehsan Masood talks to Stuart Kauffman, one of a pioneering generation of complexity theorists who struggled to define life in terms of advanced mathematics. Fifteen years on, his new book Investigations makes it clear that there may be more to life than Kauffman thought.
You thought you had a tight grip on life. What happened?
There鈥檚 nothing contradictory in what I鈥檓 saying now, compared to what I used to say. But there鈥檚 this feeling of saying: my God there is something here we hadn鈥檛 even thought about. It鈥檚 more a case of a track-change. I feel a bit like Wittgenstein. He made his reputation as a philosopher, then disappeared to Austria to teach in a high school. He returned 20 years later and basically trashed all his previous work in his book Philosophical Investigations. My book, Investigations is not as extreme an about-face as Philosophical Investigations.
Advertisement
So what does the new version of life look like?
My tentative definition of a living thing is that it is self-reproducing and does at least one thermodynamic work cycle.
What do you mean by 鈥渢hermodynamic work cycle鈥�?
I use the physicist鈥檚 definition of work and the work cycle. Think of a bacterium swimming up a glucose gradient in a solution. It鈥檚 capable of self-reproducing, but it鈥檚 also going through work cycles. And we would all say that it is going to get food. That is, the bacterium is acting on its own behalf. I call such a system an autonomous agent. We agents manipulate the Universe on our own behalf. This is true of all free-living organisms that I know. They reproduce and do thermodynamic work cycles. In my previous books I talked about life as the emergence of self-reproducing molecular systems. Up until five years ago, I would have said, that鈥檚 it-that鈥檚 life. Self-reproduction is sufficient for life. I would now say, you鈥檝e got to do a work cycle as well.
You also now say that science as we know it can鈥檛 describe life. Why?
When we describe the activities of living things, we use words like 鈥渄oing鈥�. For example, when birds build a nest, they are 鈥渄oing鈥� something by themselves. And when tigers hunt gazelles and gazelles run away, that鈥檚 what both are 鈥渄oing鈥�. Physics can鈥檛 describe birds, gazelles and tigers because physics is a language of something that has happened or is happening-like when a ball rolls down a hill. This is an event. It鈥檚 something that has happened. It鈥檚 not 鈥渄oing鈥�. I鈥檓 saying that physics and chemistry and the corresponding mathematics can鈥檛 deal with this right now. They don鈥檛 have a language for it at all.
OK, but your definition of life appears to be framed in a way that maths and physics can describe. Where do they fail?
There鈥檚 something in a work cycle that physics can鈥檛 yet explain, and that鈥檚 how spontaneous and non-spontaneous processes come to be linked. Imagine a steam engine where the hot gas is expanding and pushing down the piston in a cylinder. In thermodynamics, this is called a spontaneous process. But what if a railway employee turns up by chance and pulls up the piston to recompress and reheat the gas? This is a non-spontaneous process, because it鈥檚 the employee and not the gas that鈥檚 doing work on the piston. The machine links the spontaneous and non-spontaneous processes, and while physics can describe the spontaneous, it can鈥檛 cope with the linked non-spontaneous process. Now think of the whole biosphere. The biosphere too is a combination of linked spontaneous and non-spontaneous events. Can you predict how the biosphere will evolve? You can鈥檛.
Give me an example of what you mean?
Take Darwinian natural selection. You can鈥檛 predict it. Not even with the best models. In the book, I use the story of Gertrude the flying squirrel as an example. Gertrude is a really ugly squirrel who lived up a tree 63.5 million years ago. She鈥檚 unpopular because she has a flap of skin from her wrist to her ankle on both sides. One day an owl comes slicing down from a nearby tree to catch Gertrude for lunch. She鈥檚 scared witless and jumps out of her tree, extending her arm-and finds that she鈥檚 flying. Before you know it, she鈥檚 become a heroine, gets married, and all her kids have flaps of skin, which is why we have flying squirrels today. Or take another example. Years ago, some guys were designing a tractor. They needed a huge engine block, and they tried to build a chassis to mount it on. But the chassis kept buckling under the weight of the block. Finally, one of them said: 鈥淭he chassis needs to be massive and rigid, and that鈥檚 just what the engine block is like. So why don鈥檛 we use the engine block as the chassis and hang everything off that?鈥� This was spur-of-the-moment invention.
So what do these examples tell us?
Both pose a big problem: how can we develop a mathematics that accounts for radically novel evolutionary adaptive events, like the flying squirrel, the tractor chassis, or even the Internet, when we cannot pre-state what will come into existence? The biosphere keeps inventing new states, behaving in new ways. And we can鈥檛 say ahead of time what they will be. I am utterly befuddled by this. I鈥檓 proud that I鈥檝e got there, but frustrated because I don鈥檛 know how to state it crisply, and I certainly don鈥檛 know how to explain it using mathematics. I鈥檓 also thrilled by it, because I鈥檓 probably the first one to say it.
Does it matter that we can鈥檛 describe life using the language of maths and physics?
Well, all of us understand Newtonian physics, at least in outline. Physics, whether Newtonian or contemporary, wants to account for what happens in the Universe. But the examples above seem to point to a limitation on our capacity to predict. This is neither quantum uncertainty nor the uncertainty associated with chaotic dynamics. Rather, it appears to be the case that we simply do not have the concepts ahead of time to describe what will emerge in the biosphere or econosphere. So there seems to be some limit in the way the physicists taught us to do science.
What does it mean for our attempts to create life, to reproduce life?
I think it means I may have told us how to reproduce life. That means I鈥檝e given a simple description for what it will take to make life. It鈥檒l happen sometime in the next half-century or so. We are already making molecular systems that reproduce, and we are making molecular motors. Somebody is going to figure out how to make a real molecular example of what I鈥檓 talking about.
You鈥檙e something of an entrepreneur as well. Did you think of patenting your definition of life?
Yes I did. I thought, my God, I could obviously patent this. Then I thought, I may have stumbled across the definition of life. If I have, who am I to patent it? It鈥檚 God鈥檚 work, it鈥檚 not my work.
Do you believe in God?
I don鈥檛 believe in a personal God. But I felt this is too important to go and try and patent and make money out of. It鈥檚 my obligation just to publish it.
Yes, but why didn鈥檛 you file a defensive patent to stop someone else from profiting from your discovery?
More power to them. If there is ever going to be a technological revolution that鈥檚 based on this idea, you need patents so that you can get the monopoly, to get money, to make your idea go. That鈥檚 why we use patents.
But a defensive patent would have preserved your principle that you shouldn鈥檛 make money by exploiting a definition for life . . .
That鈥檚 true, I never thought of that. I could have taken out a defensive patent and then licensed it for free.
What are your current business activities?
That鈥檚 a long story. But the brief version is that I founded BiosGroup four-and-a-half years ago. I鈥檓 the founder chairman of the board and chief scientific officer. I鈥檓 also an external professor at the Santa Fe Institute, around which the sciences of complexity crystallised. Management consulting companies began to visit Santa Fe, sensing something new and potentially important and I did some work for McKinsey and Coopers & Lybrand. Later, Ernst & Young offered to put up $6million for a company that would apply complexity theory to business and management. I got very excited about it, in part because there is an entrepreneurial side to me. I鈥檝e started three companies. Two biotech companies and BiosGroup. We have 130 people in seven cities on two continents. We鈥檙e valued at over $80 million and have done around 45 projects for Fortune 500 companies.
What kind of work have you done?
The Procter & Gamble company wanted us to figure out how to reduce their supply chain by 75 per cent. We built five agent-based computer models in which we simulated every aspect of the supply of P&G products from warehouses to retail outlets. We included every little detail, including how many pallets go into trucks that distribute P&G products, and whether full truckloads or less than full truckloads were shipped.
This kind of business modelling isn鈥檛 uncommon. How is yours different?
I鈥檒l give you an example. Ask yourself, what are the ingredients of a good supply chain? It鈥檚 that products move smoothly from the suppliers to the plant, to a warehouse, to the shelves in the stores. There are no hiccups with transport. So the correct quantity of, say, shampoos gets loaded onto the right number of trucks. Also, that retail stores always have a ready supply of shampoos, not too many and not too few. We called this 鈥渓aminar flow鈥�. When we modelled P&G鈥檚 supply chain in this way, if everything was working well it would look like a smoothly running river. The supplies come from upstream and flow out smoothly to the final customer.
Is that how it worked out?
In reality there were problems in the system. For example, the cheapest way to transport bottles of shampoo is to send full truckloads to retail stores. But we found that this creates obstacles in our model, like boulders in a river. For example, full trucks take longer to load and unload. And they make it harder for retailers to be able to respond to sudden changes in demand for a product. So we ran some models in which more trucks were used but which weren鈥檛 always full. We found that the turbulence had gone and laminar flow was restored. We鈥檝e since run actual trials of the model and it looks like it鈥檚 working.
You鈥檝e won a MacArthur fellowship-it鈥檚 often called a 鈥済enius award鈥�. Do you think you鈥檙e a genius?
I鈥檓 not going to answer that.
Other people certainly think you are.
Well I鈥檓 glad of that. I don鈥檛 think I鈥檓 smarter, but I do think that I鈥檓 more creative than a lot of people, and that doesn鈥檛 get measured on an IQ test. I think that Investigations is an example of asking questions that nobody would think to ask. And that鈥檚 what God gave me in the way of a gift. I hope I鈥檓 using it responsibly.
Investigations is published by Oxford University Press, 拢8.99, ISBN 019512104X