Occam's Razor

Most of us have heard the term Occam's Razor (also spelled Ockham's Razor or Ocham's Razor) and may understand that it is a principle that favors simple explanations over complex ones. The word “razor” refers to “shaving away” assumptions or complexities until arriving at the core of the matter. This principle is named after the English Franciscan friar William of Ockham (1287–1347), but scholars before him, including Aristotle (384–322 BC) and Ptolemy (c. AD 90–c. AD 168), often formulated it more crisply than found in the philosophical and theological writings of Ockham himself. Ptolemy stated: “We consider it a good principle to explain the phenomena by the simplest hypothesis possible.” It is important to realize that Occam's Razor is not an absolute scientific principle, stating that the simplest possible explanation or hypothesis is always the correct one. Instead, it suggests that when confronted with multiple explanations that are equally plausible, the simplest, the one requiring the fewest assumptions, is most likely true. How does Occam's Razor apply to groundwater science and engineering, if at all? We are no longer weighing competing theories for the movement of groundwater or the fate of contaminants in it. We currently have universally accepted governing (differential) equations for both groundwater flow and contaminant transport under a variety of conditions. No need to resort to Occam's Razor here. However, we are weighing competing solutions to these equations (models) in real-world settings. Here, we have many options. We can construct complex models that include as many complexities as we can imagine or use very parsimonious models that represent only the essential processes, or anything in between. In practice, limited data availability often forces one's hand. Under those conditions complex models lack much of the underlying data and can only be constructed using a multitude of assumptions. Here is where Occam's Razor directs us to use a more parsimonious model, “shaving away” undesirable (and less important) complexities, thus avoiding too many assumptions. But even when field data are not all that sparse, Occam's Razor is a good guiding principle. Shaving away complexities exposes the essential processes that determine the outcome of the model. Such a simple model is better understood and easier explained to stake holders (or in a court of law). A simple model is also more easily calibrated against field data. And finally, it is a better candidate for hypothesis testing. However, exactly which complexities are to be shaven away? How do you avoid throwing out the baby with the bathwater? As Schwartz et al. (2017) point out; you first have to fully understand the underlying science before you are able to arrive at “elegant simplicity” that enhances our understanding of the problem. Otherwise you risk “naïve simplicity” thereby seriously misrepresenting the system you are trying to model. So there are two prerequisites for responsibly applying Occam's Razor to groundwater modeling: (1) The underlying science must be fully settled and (2) the modeler must fully understand that science. And herein lies the crux – modelers must be well-trained groundwater scientists with a good scientific intuition as to what to include and what to ignore in a model. The training of groundwater modelers, therefore, must start with the underlying science and involve solving numerous basic, but fundamental, groundwater flow problems either by hand or on the computer. This will lead to sound modeling intuition with which one can and will responsibly apply Occam's Razor.