Lorraine Daston wrote:
EPILOGUE
More Honored in the Breach
… When Hamlet remarked that the Danish custom of making a racket to applaud the king’s drunken toasts was “more honoured in the breach than in the observance” (Hamlet, I.4), he was expressing contempt for the custom. But resonant phrases, especially Shakespearean ones, have a way of taking on a life of their own, and this one has come to mean that a rule can be even more emphatically affirmed by its exceptions than by its observance. Both the long history of rules traced in this book and the current experience of living without a rulebook suggest that the 180-degree turn of Hamlet’s phrase is not an accident. The exception does prove the rule, in the double sense of both testing and confirming it. Rules don’t simply have exceptions; they define and are defined by their exceptions, as right defines left and as fence defines breach.
Context choreographs the pas de deux between rules and their exceptions. Rules formulated to guide practice in situations in which the unexpected is the expected, whether in running a monastery or besieging a city, build in examples and exceptions. These thick rules are prepared to deal with any and all eventualities. Rules formulated for more stable, standardized circumstances, whether applying an algorithm to a routine calculation or setting speed limits for city streets, barely mention exceptions. Such would-be thin rules flourish in the same settings that averages do: where what happened in the past is a reliable guide to what will happen in the present and future. An immense amount of infrastructure, both human and material, goes into making the world safe for thin rules. Workflows for calculations; sidewalks and broad, straight streets for city traffic; schooling and sanctions for everyone. Even under the most propitious circumstances, in which rules are so effectively drilled into schoolchildren that a slight change can provoke a national wave of protest, as in the case of orthography, rules must be constantly shored up by the editor’s red pencil and the demon spellchecker. The court of equity in English law or the dictionary of the Académie française, both venerable institutions established to adjudicate between rules and their exceptions, offer eloquent testimony to the fact that exceptions we shall always have with us.
There is, however, a vast difference between exceptions, boon companions of all rules, and states of exception, the suspension of rules altogether. States of exception substitute the prerogative of the ruler for the rule, whether in the form of the divine miracle that violates natural law or the head of state who waives the rule of law in an emergency or in a pardon. In such situations, discretion soars to a maximum; predictability correspondingly plunges to a minimum. Unlimited discretion has stirred uneasiness since at least Aristotle, but thick rules depended on this faculty. One historical arc traced by this book is how the evolution from thicker to thinner rules has in part been driven by growing distrust of discretion, variously impugned as arbitrary, capricious, inconsistent, unpredictable, unfair, opaque, self-serving, and even tyrannical. More precisely, low tolerance for discretion indexes rampant distrust in society: governments that don’t trust their citizens to decide where it’s safe to park or whether to report lottery winnings on their taxes; citizens who don’t trust their governments to treat rich and poor alike or not to pocket bribes and fees. Under these circumstances, all exceptions become suspect, and states of exception most of all.
Yet dramatic breaches of natural and social order are not the only way to loosen the hold of rules. More effective in the long run is to change rules so frequently and so drastically that none can take hold in the first place: rule vertigo. Miracles and emergencies are by definition flashes in the pan. If the Red Sea had remained forever parted after the safe passage of the Israelites it would have become just another natural attraction; an emergency that stretches on for years becomes the way we live now. Yesterday’s showy exceptions eventually become today’s rules with the passage of time. In contrast, rule vertigo undermines the very idea of a rule if it persists too long. If yesterday’s rule becomes tomorrow’s exception, then no rule can harden into habit or solidify as norm. The galloping pace of fashion doomed the medieval and early modern sumptuary regulations that tried to discourage extravagance; conversely, it took well over a century of stubborn persistence on the part of the Parisian authorities, who issued and re-issued the same sanitation regulations over and over again, to entrench rules in everyday conduct. Rules succeed best when they make themselves superfluous, when stopping at a red light or queueing up to board a bus or plane becomes second nature, a long-term process. As politicians charged with containing a pandemic in dynamic circumstances have discovered, the faster rules change, the weaker the hold of any rule, no matter how urgently promulgated. Rules in general begin to decay, a graver threat to order than any short-lived state of exception.
How can rules cope with variability, instability, and change without losing their grip? Each of the three ancient meanings of rules as model, algorithm, or law points to a different strategy. Laws come in many shapes and sizes, from local regulations teeming with specifics to august natural laws proclaimed for all of humanity. But whether general or specific, the more closely laws themselves imitate the permanence and predictability they aim to create, the stronger their normative authority, even if enforcement is sporadic and sanctions are mild. Fundamental or constitutional law profits from this insight by ensuring that amendments are few and far between. Legislation too often altered sows uncertainty about what the rules are, much less how to follow them. When times change or laws conflict or exceptions crop up, formidable argumentative resources of equity, casuistry, analogy, precedent, and prerogative are mustered to stretch existing law to fit the unforeseen case.
Algorithms escape context by ignoring it. Mathematical problems contain only those details needed to solve them, no more and no less. Observatories, census bureaus, and banks standardized Big Calculation in the same way that factories standardized mass manufacturing in the nineteenth century: by making the work mechanical, with or without machines. But context, with all of its disruptive details and special cases, inevitably creeps back in, as legions of behind-the-screens human workers who must repair the errors made and damage wrought by algorithms online know. Machine learning algorithms that work impressively well in development stages can be thwarted in practice by the most minute changes in input data. Making the world safe for algorithms turns out to mean freezing context: a world without anomalies or surprises.
And models? In the end, the one ancient meaning of rules that seemed to go extinct around 1800 may prove to be the most enduring. Rules-as-models are the most supple, nimble rules of all, as supple and nimble as human learning. Whether the model was the abbot of a monastery or the artwork by a master or even the paradigmatic problem in a mathematics textbook, it could be endlessly adapted as circumstances demanded. In an age of exact copies, whether churned out on the assembly line or spread online as viral images, imitation has come to mean mindless imitation. But wherever traditions or simply genres exist, whether in the sciences or the arts, whether the elegy or the still life, imitation-without-copying perpetuates the lineage without fossilizing it. Just as in the case of a living language, rules can be formulated for how to construct a grammatical sentence, write a play, compose a symphony, conduct a laboratory experiment. Yet following models remains a more efficient and flexible way to learn than following explicit rules—even for those activities that are the most rule-bound, like playing chess. Moreover, models as implicit rules pave the way for explicit rules, just as a grammatical paradigm of the conjugation of a specific verb paves the way for an explicit general rule of conjugation. A well-chosen model—a paradigm, to remain with grammar—is already halfway to a generalization. Models bridge the ancient philosophical opposition between universal and particulars, rules and cases. And they circumvent the modern philosophical problem of how to interpret rules unambiguously altogether: ambiguity in a model is a feature, not a bug.
Why then did rules-as-models not only disappear but become downright paradoxical by the mid-twentieth century? The implicit rules of models hardly ever existed apart from explicit rules; for the most part, they worked together to regularize and refine practice. One way of rephrasing the question is, under what circumstances do explicit rules no longer need the support of implicit ones? One answer—Wittgenstein’s answer—is, never: even the most apparently straightforward, unambiguous rules—algorithmic rules such as how to continue a numerical series—cannot evade interpretation. Wittgenstein’s solution was in essence to reinvent implicit rules: rules as customs or institutions—in other words, rules-as-models. But this philosophical response, however valid, begs the historical question: why did it ever come to seem that explicit rules could do without implicit ones? This book’s answer has been that success—slow, fitful, fragile, partial but real—in creating islands of uniformity, stability, and predictability fostered the dream of rules without exceptions, without equivocations, without elasticity. Both the mechanical algorithms and the natural laws were versions of this dream of rules that followed themselves, everywhere, always. The models that mediated between rules and the unruly world could be kicked away like the scaffolding from a completed building.
These dream worlds have never been fully realized, but in some places, at some times they were approximated. From standardized cooking measurements to the rule of law, from street safety to reliable statistical projections, parts of the world have become rulier—easier to govern by rules because less unruly. So impressive were these partial successes that approximation was mistaken for perfection, the mistake that Wittgenstein’s problem lays bare. But it is a problem that would not have occurred to Aristotle, or even Kant, both of whom wrestled with the more ancient philosophical problem about rules: how to square universals and particulars. It took a transformation in ideas about what rules could be—explicit, exacting, unqualified, and unequivocal—to generate the first new philosophical problem about rules in over two thousand years.
The two philosophical problems about rules, ancient and modern, are still very much our problems, as this book has tried to show. But the same historical circumstances that gave rise to the second, modern problem have greatly hampered solutions to both. Explicit rules not only pushed out rules-as-models; they also made the cognitive skills needed to follow rules-as-models—or almost any rules—suspect. Discretion, judgment, and reasoning by analogy, all faculties required to select which rule suits each case and to tailor the rule for a better fit, are in danger of slipping into the murky regions inhabited by intuition, instinct, and inspiration, all opaque to critical scrutiny. Worse, the faculties that save explicit rules from themselves can appear both unfair and irrational. Bureaucratic rules define fairness as treating everyone alike, regardless of circumstances. Their notorious rigidity stems from the fact that any deviation from uniform application counts as prima facie evidence of corruption, not wise discretion. Like modern ideals of fairness, rationality itself has become a matter of explicit rules, mechanically applied. The fact that the forms of reasoning needed to rescue explicit rules from exceptions and equivocations cannot themselves be spelled out as explicit rules renders them ipso facto irrational. Irrational, but also necessary: hardly a rule can be applied without discretion, judgment, and analogy.
Enemies of rules chafe against the restrictions that rules impose. Good sense seems thwarted at every turn; new and better ways of doing things are strangled by red tape; mechanical rules enforced by actual machines make no allowances for the natural diversity of persons and circumstances. Who has not cursed at an obdurate computer program or online algorithm, well knowing that we curse in vain? Yet every rule, no matter how rigid, no matter how cut-and-dried, is also an occasion for covert rule reasoning. Every time we seek to follow (or evade) a rule, we are honing the very faculties explicit rules banished: judgment, discretion, analogy. Which rule applies best to this case? Must it be tweaked to fit better? What exactly does the rule stipulate? Should the spirit or the letter of the rule take precedence? In normal times, our judgments about such matters are so swift and sure as to be invisible to us. But in abnormal times, when we are thrown into the breach without a rulebook, we once again become aware that there are no rules to help us reason about rules.
The Source:
Lorraine Daston, Rules: A Short History of What We Live By, Princeton University Press 2022