I got a Michael Chrichton book for Christmas. As I was poking around his web site, I happened upon a recent speech by the author titled Fear, Complexity, & Environmental Management in the 21st Century. Chrichton had rattled many feathers with State of Fear (I found the book a fun read but the message in it was mostly lacking). This speech seems to be the evolution of his thinking. The main points are:

1. The environment is a complex system.
2. It’s difficult to predict long-term trends in complex systems. Large scale violent events have been common in the world forever.Spreading doomsday fears doesn’t help the situation. Often the doomsayers don’t have good data or are in the game for political or financially-motivated reasons.
3. The environmentalist movement should stop thinking simplistically about the world and start treating the problem as one of complex systems management.
4. We need to stop the cycle of fear.

It’s hard to argue with these high-level messages but I do think Chrichton misses two important points about self-regulating complex systems such as the environment, financial markets, etc. First, every such system has a certain rate at which it can respond to change. Financial markets can respond very quickly. Ecosystems respond much more slowly, many orders of magnitude more slowly if we are talking about genetic changes. Second, the feedback & response mechanisms the self-regulating systems have evolved are naturally dependent on the types of events they have experienced in the past.

When you put these two together, it’s easy to see how certain type of shocks may significantly impact or perhaps break even the best self-regulating system either because the system is unprepared to respond to the type of shock it has received and/or because its response cycle simply takes too long.

I know nothing about environmental management but I have had some experience with complex system management. Years ago I did research on soft AI (nowadays called soft computing). Soft computing is often applied to complex systems, which lack agreed-upon formal models defining their operation. Many of the soft computing sub-disciplines such as neural networks, neuro-fuzzy systems or evolutionary computing employ a learning model for creating a solution to a problem.

Let’s consider a really simple example of a neural network that takes the height of an individual as input and produces that individual’s weight as output. If the neural network is trained on data from Chinese peasants, it likely won’t do a good job predicting the weight of Nordic individuals (because it won’t have had inputs in that height range) or Samoans (because of the different body types for the same height).

Well, by the same argument, it is not clear how natural or man-made systems would respond to stimuli they have not previously experienced or ones whose cumulative effect is so significant that it overpowers the speed with which the system can adapt.

In the case of the environment, an examples would be is a large meteoric impact or a nuclear holocaust. More relevant examples are ozone layer depletion, CO2 increase, etc. The key issue is less whether the magnitude of the change is catastrophic and more whether the response required to prevent a significant undesirable deviation from the status quo (delivered through a combination of Nature, governments, the private sector and individuals) can be quick & effective.

In the case of financial markets, examples include new financial instruments or trading strategies. Insider trading is one example. It used to be perfectly legal to trade on private information. Not anymore, because we realized that allowing it would compromise the very structure of the self-regulating market system, which is based on the assumption of (nearly) perfect information.

In the case of the technology industry, the coming of the Internet is a good example. The magnitude of the change and the speed with which it came about took the rest of the hi-tech industry by surprise. It could not respond fast-enough. As a result, the very landscape of the industry changed significantly.

Many in the industry thought that Web services will bring about the same type of radical industry change. It didn’t happen. The existing ecosystem was able to absorb the shock by co-opting the new technology in their existing product lines and adjusting sales and marketing messages accordingly.

So, in a funny turnabout, Chrichton’s speech made me think about tech booms and how one can shock industries past the breaking point in order to introduce truly revolutionary change.