Scientific parameterization and its validation: comparing the universal models of fisheries economics with the invalid modeling of stock assessment

Abstract

Here I compare the valid parameterization of fisheries economics with the data-fitted parameters of a stock assessment’s modeling, invalid models whose predictions cannot be falsified (which is another way of saying the models are not universal). The distinction between (a) a model’s valid universal prediction and (b) a model’s invalid prediction that cannot be falsified can be summarized in terms of a model’s parameterization, as: high falsifiability = paucity of parameters = simplicity Models of stock assessment; invalid models whose predictions cannot be falsified I used Ricker’s version of a Gordon Schaefer model (Corkett, 2002, my Fig. (c)) to illustrate a non falsifiable MSY prediction of stock assessment, a prediction that, like that in Fig. 1B, takes the logical form of a ‘there-is’ proposition (Popper, 1959, p. 68), as: The universal laws of physics, for example, are simple: they have few parameters and a high degree of falsifiability. It is this high degree of falsifiability or negation that guides all engineers by showing them what cannot be achieved and should not therefore be tried as part of ‘trial and error’ engineering. By contrast, the models of fish stock assessment are not simple: their data-based parameters change with the changing data. These models are not universal; they apply only to a particular fishery situation and are incapable of guiding all fishery managers (all social engineers) by indicating what cannot be achieved in a social engineering. Under Karl Popper’s limited (and formal) definition of empirical science data-fitted models proffer policy advice to fisheries management that is formally invalid.