On 16 June, over a billion people from the subcontinent had their attention focused on a small patch of greenery in Manchester, UK, the site of a cricket World Cup clash between India and Pakistan. The tumultuous political history of the two, combined with the highly charged sporting rivalry between the two teams, always makes for an intense spectacle that brings out indescribable emotions among the citizenry of both nations. And yet, this perceptible animosity belies a great sense of off-the-field camaraderie between the players of the two teams, as individuals. This veritably extends in general to Indian and Pakistani individuals when they come together as South Asians, for example, as graduate students in an American university. What would explain this intriguing phenomenon, where group behaviour makes for a striking contrast to Individual behaviour?
An answer to this question was put forward 40 years ago by Nobel Prize-winning economist Thomas Schelling in his book Micromotives And Macrobehavior. Schelling argues that individual motives do not directly explain group behaviour, and illustrates this with a simple mathematical model. He studied ethnic segregation in residential communities in US cities. Making the liberal assumption that an individual would prefer to live in a neighbourhood where, say, a third of its members belonged to the same community, he built a mathematical model to allocate neighbourhoods to individuals. The outcome was surprising and insightful: Although individuals needed only a third of their neighbourhood to be composed of members of their community, the model built neighbourhoods where close to 80% of its members were of a single community. Schelling’s model became a seminal work in the field of game theory and agent-based computational economics.
Model thinking is neither new nor uncommon. The current renewed imperative for Model Thinking arises from the deluge of data that organizations today find themselves with. Data in itself cannot inform decision-making. Data needs to feed a structured model, which could then logically represent the essence of a situation consistent with the recorded data. This model can then be tweaked to study the impact of inputs and assumptions on outcomes, as well as to predict future outcomes based on perceptible trends in the data. While using a model can lead to great insights and informed decision-making, any model by itself is bound to be wrong. A single model will most likely not capture every assumption or variable that might impact the eventual outcome. Using a number of diverse models is often helpful in ensuring that different factors are considered, and results in better decision-making. Philip Tetlock, a political scientist at the University of Pennsylvania, in his book Expert Political Judgment demonstrates that a nimble, broad-based approach built on diverse traditions yields more accurate predictions than a single-pronged, inflexible approach.
Moreover, the utility of models is human in its essence. While machines and algorithms can sort through big data to optimize one node of a network, a human perspective is necessary to understand and solve problems for a complex and interconnected network, taking into account human preferences and objective functions. A classic example of this is the use of trading algorithms in stock and commodity markets, which may provide superior returns, but only for a while. Sustained Warren Buffett-esque outperformance requires a human understanding of multiple models.
The concept of modelling can be extended to general thought. Most of us subconsciously build mental models that help us understand and interpret the world. Over time, we tend to favour one model over others and begin to apply that model indiscriminately to all situations. For example, managers take pricing decisions after considering the cost of delivery, utility to the consumer, and competitive pressures. A manager with a dominant profit-and-loss mental model may focus more on cost considerations at the expense of the other two. Each mental model partially opens our window to the world; the more models we use, the clearer our view. We piece together the bits of reality gleaned from each into one holistic view of the world.
This has a direct impact on how we think of policy and governance in an interconnected world. If we were to apply an agent-based model to governance, it’s apparent that behavioural change at the global level can happen most effectively through interventions at the individual level by shaping human preferences and values. When the scale of interactions at a micro level reaches a tipping point, fundamental macro-level changes can happen. This places great responsibility on education as an effective means of shaping the values of an individual for the interconnected world of the future.
The essence of model thinking in the 21st century—which is to apply multiple models on copious data to develop diverse perspectives that inform decisions—can and must be applied to the most important global problems facing humanity today. Model-thinking can not only help us understand possible scenarios of the future but also inform decisions we can take today to ensure the most favourable outcomes under various future scenarios. The preponderance of data and computing power today makes model thinking substantially more effective, impactful and, therefore, imperative.
Kapil Viswanathan and P.R. Venketrama Raja are respectively, vice-chairman of Krea University, and chairman of Ramco Group
The post Model thinking as an imperative in the age of data appeared first on CommentWise.