Paradox, accounting values, and intelligible regulation

Expected utility theory, which includes estimating the probabilities of uncertain future outcomes, is the classical model for rational economic decision making, and, by implication, rational valuation and financial reporting regulation. In Wittgensteinian terms it is a ‘hinge’ of the language game in which these practices are embedded. When rendered explicit, however, this ‘hinge’ appears to be formally incoherent. The exploration of this problem has consequences for all of our arguments over the epistemological underpinnings of accounting reports – whether realist, representational, constructivist, or otherwise.
Arguably, there are two complementary primitive models that underlie real-world probability estimation. Taken together, they generate a version of Goodman’s inductive paradox (other versions of which also arise for non-inductive empirical generalisation). This, in its turn, is related to Kripke’s paradox, which arises when we try to give behavioural accounts of rule following, and so of participation in a language game.
This paper explicates this type of paradox in the context of commercial decision making, and considers its consequences. The existence of paradoxes should render the system that generates them completely incoherent, but (paradoxically ...) they seem to be generated by any attempt to give complete accounts of some of the normative fundamentals which underlie linguistic practice - such as truth-telling, validity and rule-following.
Whether or not these paradoxes represent a serious threat to the coherence of the empirical or behavioural sciences, it might be objected that commercial decision making methods and financial regulation rarely aspire to the kind of rigour that these disciplines attempt to achieve. Part of the argument of this paper will be that the intelligibility of commercial language suggests an approach to these paradoxes which is not obvious from more traditional philosophical perspectives.

The intentionality of belief renders certain belief claims by participants in a shared language game incorrigible (within the game), in the sense that they can be doubted only by doubting the seriousness or quality of participation. If certain statements about rule following and word meaning have this same quality, then there is a way of avoiding the consequences of Goodman’s and Kripke’s paradoxes, and of sterilising the probability estimation paradox for any playable commercial language game.