Thursday, February 24, 2011

Complexity studies offer rich insights

Feb 23, 2011
By Su Guaning

HUMAN societies, especially cities, are enormously complex organisms. For illustration, consider a simple dynamic system comprising a retailer-wholesaler-manufacturer distribution chain - the 'Beer Game' devised at MIT's Sloan School of Management.

Picture a student in his dormitory, at 3am, staring at his computer in puzzlement. He is the 'Retailer', who usually sells four cases a week of a local brand 'Lover's Beer'. He has a stockpile of 12 cases in inventory, and routinely orders four cases a week from the wholesaler to replenish his stock.

During Week 2, his sales double to eight cases. He duly orders eight more cases to replenish his stock. But during Week 3, only four cases arrive - there is a four-week delivery delay - and he continues to sell eight cases. So he orders 12 more in anticipation of future demand.

By Week 5, his stock has run down to zero. By Week 9, there is a backlog of 11 cases, with no relief in sight as the delivery man fails to show up. He orders even more cases.

Meanwhile, across campus, there is another student, the 'Wholesaler'. He sees the demand from his retailers climb through the roof and yet the supply from the brewery fails to materialise. By Week 9, his backlog goes up to 43 cases.

A third student plays the role of the 'Brewery'. She sees an even bigger spike in demand as the orders pour in from the wholesalers. She ramps up production in a crash programme. Everyone is screaming: 'Give me more Lover's Beer!'

All of a sudden, from Week 14, the beer arrives like a flash flood. The same sales volume cannot keep up with the supply. Demand collapses, overwhelming not just our retailer and wholesaler but also our brewery. By Week 24, there is an excess of 500 cases of Lover's Beer.

What triggered this disaster? A simple demand spike due to a pop song featuring Lover's Beer!

While each player merely responds to his immediate surroundings, they collectively exhibit behaviour dictated by the structure of the system as a whole. The 'Beer Game' has been played many times by different people - including in Singapore - and the outcome is invariably determined not by the players but by the 'system dynamics'.

The property market here and the prices of certificates of entitlement (COEs) for cars also exhibit large gyrations due to inherent delays and human complexity. In the case of the property market, the process from the release of land to completion of a residential project can take years. Faced with temporary shortages, buyers panic and bid up prices. This results in boom and bust cycles. Delays and complex human behaviour can make any system prone to wild gyrations.

Whether it be Singapore or the Tianjin Eco-City, building sustainable cities requires careful system designs to moderate and prevent such wild swings. But while we can routinely design quantifiable systems such as aircraft, human beings are infinitely more complex. It is impossible to derive a detailed model of human society that would allow a government to determine the impact on citizens of every policy it adopts.

We saw this in the case of COEs. The theory is simple: control the number of new cars put on the road each month by a bidding system - and voila, traffic will flow more freely. Real life, however, is not as accommodating.

Even if we managed to control perfectly the number of cars on the road, traffic congestion depends on the decisions of hundreds of thousands of individual drivers: what route they decide to take; speed of travel; whether they slow down to look at a traffic accident; and so on. Add to these the factors that prevent us from controlling well the number of cars on the road and the problem really becomes intractable.

The insights that scientists have gained from studying complex systems in the aggregate may nevertheless be useful.

Analysing the operation of a car engine by going down to the single molecules of fuel and oxygen is practically impossible due to the large number of molecules and the unknown initial conditions. But the macro laws of physics governing compression, heat transfer, expansion and production of mechanical energy enable us to produce macroscopic models of performance to design workable engines. Can we do the same for sustainable cities?

The difficulty would lie in the absence of physical laws governing macroscopic human behaviour. Physicists have nevertheless tried to apply their insights to human societies with some success.

Dr Geoffrey West, a theoretical physicist who is a frequent visitor to NTU's Institute of Advanced Studies, is one example. He and his colleagues at the Sante Fe Institute have applied physics methodology to complex organisms ranging from rodents, elephants and blue whales to all sizes of cities. He has discovered amazing 'laws' governing cities. For example, when a city doubles in size, its economic output not just doubles but increases by a further 15 per cent.

This is one example of what is called 'complexity studies'. While not immediately useful for policy-making, it can provide useful insights. There is a unique collection of high-powered intellectuals at Santa Fe, including Nobel physics laureate Murray Gell-Mann, dealing with similar problems typically considered intractable.

The human condition is a fascinating source of inspiration for poets, artists and musicians. It is also a wonderfully complex, interwoven tapestry of patterns and even hidden codes. We should venture into complexity studies so as to contribute to the richness of human studies and possibly improve public policy.

The writer is president of Nanyang Technological University. Think-Tank is a weekly column rotated among eight leading figures in Singapore's tertiary and research institutions.

No comments: