Wednesday, September 27, 2006

Living in a distributed world

In this post I had mentioned how ants work together to find a short way past obstacles, and said that they are not the only living things that work in that way. Engineers, though, have a different way of looking at it than pure scientists. These types of systems are called Distributed Control Systems, meaning that there is no central controlling authority, but instead, each individual follows a set of simple rules and this alone is enough to achieve the system's objectives.

Such a concept may seem alien at first. We humans are used to a hierarchical system of getting things done... in governing citizens, in managing corporations, in controlling manufacturing systems, in practically everything, we have a system of smaller parts reporting to progressively larger parts. But there is evidence to show that simple rules followed by a large number of peers can lead to as complicated a global behaviour as might be required.

I'm going to take an example of a non-living thing this time. The following experiment is called Conway's Game of Life. Consider a two dimensional matrix, in which each cell can have a state of "dead" or "alive". Each cell has eight neighbours; if zero or one of them are alive, the cell dies of loneliness. If two or three neighbours are alive, the cell lives. If four or more neighbours are alive, the cell dies of overpopulation. A dead cell "comes to life" if it has exactly three live neighbours.

Depending upon the initial state, an astounding variety of beautiful and complex patterns have been observed to arise out of these simple rules, including different types of oscillators, explosions, firing guns, even moving spaceships!

A snapshot of the Game of Life Applet: From http://www.ibiblio.org/lifepatterns/

While I used this only to demonstrate that simple rules can result in complex, self-organizing behaviour in a distributed world, the fact remains that a lot of the "management" in nature occurs this way. It doesn't take an engineer to understand the problems with centralized control in a large hierarchy; if you've tried getting something past the bureaucracy of your university, workplace or government, you know it all too well yourself! But engineers understand how difficult it is to design a distributed control system. It is only in extreme circumstances that they do indeed deploy systems in such a manner - for example, in large sensor networks, where thousands of sensor nodes may be released in forests, underwater, anywhere!

We live in a distributed world, in which the sum of parts is inevitably greater than the whole. Sometimes we scientists get a glimpse of it in our computer simulations, but you only need to look out of your window to understand. When you see a flock of birds flying in formation. When you see a beehive. And once you truly understand it, you will see the world through brand new eyes.

7 comments:

Sakshi said...

Well written.
" but you only need to look out of your window to understand."
I think that is the reason why I am in biology - because I see the complex gene talk that occurs in the simplest task in nature :)

Born a Libran said...

Conway's game of life - the first cellular automata. Set up a whole field of different kind of simulations which are used in such diverse fields as Fluid Mechanics and Systems Biology. Interesting stuff.

Born a Libran said...

By the way, why is there no category for this post?

Prashanth said...

Couldn't think of one... I'll do something about it now..

OCTFCU Online Banking said...

I'm pretty sure everything starts with simplicity, at some point or another.

buy a essay said...

I agree with you that there is evidence to show that simple rules followed by a large number of peers leads to as complicated a global behavior as it might be required.

Harris Bank Online Banking said...

I once saw ants make a raft out of dead ants. They're totally going to outlive us.