Can the future really be predicted?

For several decades, researchers have heatedly debated the statistical properties of real-world time series. On one end of the spectrum, researchers put forward that the majority of the real-world time series are nondeterministic and that some of them exhibit `random walk'; these researchers advocate that real-world time series are unpredictable. On the other end of the spectrum, researchers have devoted their entire careers trying to predict time series based on the past history of the time series. A variety of time series prediction techniques have been widely used in domains such as weather forecasting, electric power demand forecasting, earthquake forecasting, and financial market forecasting. As real-world time series are affected by a multitude of interrelating macroscopic and microscopic variables, the underlying models that generate these time series are nonlinear and extremely complex. Therefore, it is computationally infeasible to develop full-scale models with the present computing technology. As a result, researchers have resorted to smaller-scale models. Despite advances in forecasting technology over the past few decades, there have not been algorithms that can consistently produce accurate predictions with statistical significance. Therefore, this position paper investigates whether real-world time series are deterministic or nondeterministic. This paper argues that nondeterminism does exist and that real-world time series exhibit unpredictability. As a result, consistently accurate time series prediction can be considered to be impossible. However, short-term time series prediction may be possible temporarily if one could somehow discover a simple model that can momentarily represent an intricate system.