Jul 06 2004
Suppose that we are comparing toasters and cats. Both are about the same size, and are familiar objects.
If we try to understand the toaster, we find that the greater the precision we measure it, and the greater degree to which we understand its failings, the better we are able to fix it. We can take a toaster apart and put it back together again. Our understanding of how a toaster fails and how it thrives are nearly equivalent. A toaster is exactly the sum of its parts.
A cat, however, is much different than a toaster. It really doesn’t help to measure a cat to greater precision, and understanding how a cat dies is an entirely different way of knowing than how a cat thrives. We cannot dissect a cat and reassemble it and have the same cat. The cat is greater than the sum of its parts.
When systems become cat-like, the gap between our understanding of their failures and our understanding of the way they thrive widens. Emergent properties – the difference between the whole and the sum of the parts – are invisible to the toaster-like perspective.
If we DEFINE cats according to the “strengths” of toasters, we find that they make very poor toasters because they don’t generate enough heat, they don’t sit still, and they are unpredictable. This analysis would confirm our suspicions that cats are poor toasters.
Back in the days of typing memos on typewriters, we didn’t have to worry about Internet-based viruses. However, connecting all of our word processors together introduced a cat-like behavior. The more we connect things, the more the emergent properties come to dominate the behavior of the system, and the harder they are to understand from the toaster-like perspective.
The moral of this story is that we are facing increasingly cat-like issues, but are generally stuck with toaster-like perspectives with which to understand them.
Let’s look at another form of order – the snowflake. These condense with the right conditions of temperature and humidity. They are amazingly similar, and amazingly diverse at the same time. The process is highly scalable; snowstorms or light flurries can happen.
Snowflakes are even hard to understand from either the toaster-like or the cat-like perspectives. They represent a phase transition which may be quite unexpected from simply looking at a cloud.
In a sense, the web was a snowstorm-like phenomenon, condensing from the proper initial conditions, some of which were provided by the state of the Internet itself, and some of it crystalized by Sir Tim Berners-Lee’s vision by simply introducing HTTP, URL, and HTML standards.
The gist of all this is, that there is a huge difference between fixing the world’s problems with toaster-like precision, and making the world a better place by condensing good things with snowflake-like scalability, order, and diversity.
So how do we create an environment in which good things condense like snowflakes, rather than being built like toasters?
There is a lot of science with which to inform this thinking. Physics has come a long way from the days of the “clockwork universe.” Check out Statistical Mechanics, and some collected links on Percolation Networks.