Predicting The Future

Since ancient times people have attempted to predict the future, but where they sought answers from mystical sources, we can apply the scientific method. Predicting the future has also appeared within works of fiction. For example, in Foundation, written by Isaac Asimov, psychohistory was used as a method for predicting large scale future events, by combining an analysis of history, sociology and mathematics.

Predicting The Future

Futurology

The term futurologist is applied to people who attempt to predict future events by looking at historic trends and analysing data, including climate, geopolitics and demographics. Although some might consider futurology to be pseudoscience, its practitioners believe they are applying the scientific method, by looking for patterns and applying probability.

We are familiar with predictable events, such as the changing seasons. Meteorologists use computer models of historic data to produce weather forecasts that many of us consult before going out for the day. When planning long term strategies, people working in business and government also attempt to predict the future, by gathering relevant data and taking the actions that they believe will have a desired outcome, based upon their expectations.

Predictive Models

When building predictive models, the greater the range and depth of relevant data considered, the more likely predictions will be accurate. Determining all of the relevant variables will be a challenge and developing robust models requires input from a wide range of perspectives. However, specialists are often unaware of some relevant factors and individual perspectives can be shaped by limited personal experience, assumptions and preferences that reduce the accuracy of predictions.

Within a controlled environment, such as a card game or board game, the number of variables will be limited and knowable, enabling probability to be calculated with a good degree of confidence. However, predicting the future on a global scale in the real world, confronts anyone who attempts to do so with an almost infinite combination of variables to consider. In addition to this, the accuracy and quality of data can vary and random or seemingly unrelated events can influence outcomes.

Probability

Physicists describe the behaviour of subatomic particles in quantum mechanics as being unpredictable, so that the position of a particle is uncertain. However, in the aggregate the behaviour of large numbers of particles can be predicted to a high level of accuracy, using probability. The concept of psychohistory in Foundation was modelled on the kinetic theory of gases, in which movement of a single molecule could not be predicted, but the behaviour of a large volume of molecules was predictable.

Applying probability to human society, we might not be able to predict small scale events accurately, but on a larger scale knowing the relevant variables and having good quality accurate data could deliver useful results. For example, we cannot with confidence predict the score in a particular football match or a companies share price on a specific day. We would expect to have greater success though when predicting performance over a longer time period, if we had access to relevant historic and current statistical data and a proven model with which to analyse the data.

Scenarios

Returning to the world of physics, chaos theory describes small changes having large effects, such as a butterfly flapping its wings and causing a storm. Therefore, rather than attempt to predict exact outcomes, when building predictive models, we should instead develop scenarios. They can be used to describe a range of possible outcomes that are based upon observable patterns, possible future human behaviour at population levels and allow for the complex and unpredictable nature of chaotic systems.

Currently machine learning is being used to predict outcomes as varied as climate change to the behaviour of customers on eCommerce platforms. Increasingly powerful artificial intelligence, could allow us to build more detailed models that will deliver more accurate predictions of future events. Human nature, as measured by The Five Factor Model, is largely unchanged across time, but other factors would be less predictable, even at scale. Although there will always be a degree of uncertainly, the process of considering the impact of our current actions upon future events could help us to make better decisions.

Learning From History

Rather than build predictive models, input data and allow time to pass, so as to determine the accuracy of their predictions, models can be evaluated by looking at historic events. From ancient Greece or Rome, to more recent periods of history, we could feed relevant data for a specific year into various models and measure how accurately they each predict events during subsequent years. Models that predict past events with a reasonable degree of accuracy, could be fed contemporary data and the results evaluated in an attempt to predict the future.

Predicting the future of our world on a global scale would require highly complex models. They would likely use artificial intelligence and models they build could be beyond human comprehension. Valuable insights might be gained from their scenarios for our future, but how much trust should we place in machine algorithms rather than our own judgement when making decisions? What moral dilemmas could such a potentially powerful resource confront us with and what regulations might be put in place?

Posted in Technology.