Networks: Innovation, growth and sustainable development


The emergence of the Internet as a measureable manifestation of our social and economic relationships changed the domination of networks in our lives. From about 2000, the internet has allowed us to study and understand the type of networks in which we live, and to model their behaviour. The Internet has fundamentally changed the distribution of wealth. The rich became richer simply because of the larger scale of the trading network and stretched national wealth distributions. Network effects are therefore likely to be responsible for much of the perceived increases in inequalities in the last 20-30 years, and policies to tackle poverty must therefore address the extent to which the poor can engage with society’s networks of wealth creation. The greatest challenge to continued growth and prosperity, and therefore to peace and justice, is climate change. The potential cost of inaction on climate change could be as high. Our self-organising social networks have structured our societies and economies, and are now reflected in our technology networks. We can now replicate their evolution in computer simulations and can therefore better assess how to deal with the greatest challenges facing us in the next few decades.

Networks dominate our lives. They dominate our economic and social behaviour, yet until the last few decades we knew little about how they behave and grow.

This changed with the emergence of the Internet, not just as a communications tool, but as a measureable manifestation of our social and economic relationships. From about 2000, the internet has allowed us to study and understand the types of networks in which we live, and to model their behaviour.

These insights have since been transferred across to economics, ecology and biology, providing new insights into the nature of evolution in all areas. There have been numerous valuable successes and these have raised expectations that the science and modelling of net- works can make a major contribution to economic policies, sustainable development and medicine. The application that has raised most expectation is in economics. However, there are limits to what can be done. From my experience in the last few decades, let me summarise what we now know, and what we can and can’t do.

1. Self-Organising Networks

There is now no doubt that social and economic systems have most of the characteristics of one of the four types of self-organising networks: scale-free, “small-world” networks. It is worth reminding ourselves what these characteristics are: A power-law (nx) relationship between nodes (people, busines- ses etc) and the number of links between them, and a power-law distribution of other key parameters (wealth, influence, and also fluctuations around dynamic equilibria etc.). Such networks emerge naturally in nature and society, and give rise to the dominance of “hubs” – key individuals and businesses; to the inequalities we know so well as Pareto’s 80/20 rule; to “non-linear” responses to change, and to the fluctuations we see in equity markets.

The dynamics of such networks can also now be well modelled. They emerge naturally as a result of simple rules of “preferential attachment” to more connected nodes and are robust to disruption (which is why the Internet was designed this way), and therefore also to deli- berate change by policy intervention. Their evolution is also “path-dependent”, so the effect of a policy intervention depends on the starting point and previous history. They are in cont- inuous evolution, don’t have static equilibria, and are susceptible to “boom and bust cycles”.

2. The Challenge to Economic Analysis and Projection

The economic models used by all main-stream economic institutions are still “dynamics tochastic general equilibrium” models. However, the recognition that our economic and social systems are complex dynamic networks requires new modelling approaches, notably “agent-based” modelling, that explicitly recognise the interconnectedness of different agents (people, businesses and parameters). These models can replicate and “explain” many of the characteristics of our economic systems. However, economic and social change can only be studied by letting computer simulations of the interplay of complex interactions “play out”, and computer simulations can never reflect the full complexity of human behaviour and interactions. We will therefore need to continue to use a combination of General Equilibrium models for short-term steering of our economic systems, and “agent-based” network models for more long-term analysis of growth, innovation and sustainable development.

3. Boom and Bust in Networks

A first useful insight emerged very quickly with the boom and bust of “Internet-related” companies in the period 1998-2003. PC-based dial-up access to the Internet and GSM mobile telephony are both classic “network” technologies; the value depends on the square of the number of users.* Growth is therefore initially hyperbolic: It follows Metcalfe’s law (faster than exponential), but then slows as the utility of additional links and participants diminishes. All growth patterns are self-limiting; network growth also slows and eventually saturates, as in the case of the penetration of GSM telephony and PC-internet use, and now social networks such as Facebook. Equity value of Internet and GSM-related companies which had been grossly inflated by extrapolations of ever-faster “exponential growth” col- lapsed as the growth-rates slowed. A simple model of self-limiting hyperbolic growth well reflected the evolution of technology use, with growth rates following its first differential and equity-values, the second differential. A more complex but analogous pattern underlies the current economic crisis, with its origins in the business, banking and government networks which followed an unsustainable enthusiasm for debt-financed investment.

Peter Johnston: Senior Advisor, European Policy Centre
* The value of a network is related to the number of participants and number of others they can connect to: the square of the number of participants.

Pages: 1 2