Scientist uses smart building blocks
In 2009, Dr Stephen Wolfram, the enfant terrible of contemporary science, released a breathtaking computer program that creates a completely new way to handle and query complex human knowledge.
Even though his program is still in its infancy, it already poses a real threat to the search-engine technology used by Google and others. He has created nothing less than a new approach to machine-based problem solving.
A former child prodigy who published articles on quantum physics at the age of 17, Wolfram is best known as the inventor of the widely used Mathematica program.
His new knowledge-computation engine, called Wolfram/Alpha, is based on his Mathematica system. It is able to solve all sorts of scientific queries and problems users may have.
However, try the same sort of queries with Google or Yahoo and all you will get is long lists of references mentioning the same words you just typed in. Underlying the program is a fresh conception of science, one Wolfram first proposed in 2002 in his book A New Kind of Science.
Wolfram realised that even the most complex computer programs with millions of lines of code may not be able to solve difficult scientific problems.
But there are a variety of simple computer programs that can be defined in just a few lines of code and can display astonishingly complex and unpredictable behaviour. So, instead of analysing what very complex programs may or may not be able to do, Wolfram analysed in depth a certain class of basic program types, called cellular automata, which, despite their simplicity, can emulate even the most complex natural processes in nearly any scientific field.
A cellular automaton or cellular machine can easily be defined. The simplest are one-dimensional machines and can be arranged in a horizontal line. Each simply consists of a number of attached cells (usually graphically represented as squares) and some simple rules that define how the state of each cell changes depending on the state of the cells in its immediate environment (see Picture 1).
Such machines are called 'cellular' because the concept suggests an arrangement of cells just as they exist in cell tissue, where each cell is adjacent to and influenced by neighbouring cells. Once such a cell assembly is initialised with a state for each cell, the 'computing' process can start.
Each cell evaluates the state of its neighbours and applies the relevant rules to change its own state. When each cell has updated its own state, the process starts again. The effect of the successive application of these rules can be seen in Picture 2. It can be easily observed that some rules show uninteresting results (see Rule X), whereas some other rules show very complex and unpredictable behaviour (see Rules Y and Z).
Wolfram did not invent such machines. Nor was he the first to study them. But he may go further than most to generate insights into the complex processes these machines can emulate and why.
His idea is that the more complex processes of nature cannot adequately be described by rigid mathematical formulas. Instead of more complex theories, his approach is to use simple cellular automata to study scientific problems and processes.
Using this cellular concept, he has been able to produce his search program.
Eberhard Schoneburg, a computer scientist, is the CEO of Hong Kong-based Artificial Life
