Including causal mechanisms in model formulations is the key to predictive success. Yet it would appear that much of our latest (specifically ecophysiological) understanding accruing from phenomological evidence and experimental work is often not being included within model structures. Dysfunctional equations (which fail to capture mechanistic understanding, and which lead to incorrect model behaviour) are often used instead, the use of fixed Monod formulations to simulate multiple-nutrient interactions in phytoplankton, or Holling type II descriptions of resource-limited predatory activity, being typical examples. In some instances, dysfunctional equations may be adopted through sheer ignorance, a worrying prospect given our incomplete knowledge of many processes occurring in marine ecosystems. One wonders, for example, to what extent the parameterisations used in the current generation of complex ecosystem models being developed for climate studies, and the predictions thereof, can be relied upon. Here, we investigate the underlying problems leading to the use of dysfunctional models within marine food web systems, our perusal of the subject suggesting that ignorance is by no means the only factor, other reasons being indifference, inertia, and subservience to simplicity.