Law & Society Blog

Many normative issues are, in fact, unsolved empirical questions. For example, if the deterrent effect of criminal punishment could be measured reliably, much (not all) of the philosophical thicket surrounding the justification of punishment would disappear. This is because, as a general matter, there tends to be greater consensus with respect to high-level principles, for example, that core criminal behavior warrants a punitive response, than with respect to the particulars, for example, that drug abuse, specifically, (i) ought to be criminalized; and (ii) ought to carry long jail terms. Greater empirical understanding transforms such questions of ends into questions of means. If we knew empirically that (i) drug abuse did little more than endanger the health of the perpetrator-victim; and (ii) long jail sentences had little to no deterrent effect, we would, over time, probably move from a punitive model of dealing with drug offenses to a public health solution. The same is true for many other similarly charged issues.

Empirical analysis (using the term with all the appropriate caveats, that is, understood as the subscription to a realm of socially constructed reality, which is less controversial than competing or complementary realities) transforms black boxes, or non-trivial machines, into transparent boxes or trivial machines. (H. v. Foerster). Trivial machines, no matter how complicated, can be understood in terms of input and output, ex post causation, and means-ends rationality, in other words, they are synthetically determined, independent of the past, analytically determinable, and predictable. Non-trivial machines, in contrast, cannot be understood in that manner; their behavior depends on changing inner states and continuous self-reference. Non-trivial machines are historically dependent and unpredictable. Therefore, to a significant extent, normativity is a strategy for coping with a lack of reliable empirical information. Without the ability to rely on normative expectations, the (re-)actions of an observer of a non-trivial machine would be random and highly unstable.

For most of human history, the future was, in fact, a black box. Dealing with uncertainty required normative techniques, most significantly religion. Then, in the 16th and 17th Century, future became increasingly transparent through the discovery of natural laws (which were “eternal” in the sense that they applied to past, present, and future situations) and through mathematical tools to describe chance uncertainty in terms of risk. The demystification of the modern world in (what we call) the Age of Reason consisted in large parts of a replacement of normative models of the world with technical models, that is, a transformation of formerly non-trivial machines into trivial machines. The two most significant classes of non-trivial machines that have remained to this day are persons and social systems.

Today, problems with trivial machines are problems of science; that is, they have been farmed out to the scientific system, a functionally differentiated sub-system of society that administers the programs for applying the binary code values true/false to factual propositions. Social consensus about trivial machines is, therefore, more or less stable. (Compare that to the 17th Century where natural philosophy was still competing for that post. I should point to Neal Stephenson’s gripping/hilarious account in Quicksilver.) In contrast, problems with non-trivial machines, whose behavior often frustrates our expectations, require normative strategies. At the core of a normative strategy is the partial replacement of a relevant but unknown “is” with an “ought.” Normative strategies reduce the significance of frustrations of our cognitive expectations (which are inevitable, because we cannot reliably predict the behavior of a non-trivial machine) and increase our willingness to rely on normative expectations, even in the face of disappointment. Put differently, as Niklas Luhmann explained, in the event of frustration, we tend to abandon cognitive expectations, but we hold on to normative expectations. For example, if I leave the building at night and find my car stolen, I quickly abandon any cognitive expectation of being able to drive home. However, I refuse to abandon the normative expectation of others to honor my property rights. Against that backdrop, punishment is not meant to deter prospective thieves by increasing the expected costs of criminal actions or to rectify a past wrong, in fact, punishment is not overly concerned with the perpetrator at all. Rather, punishment is the symbolic affirmation that continuing reliance on a normative expectation (for example, that of property) is not foolish but is what society expects me to keep expecting.