Spirit Vs Letter


Spirit takes many forms—rarely can we establish its exact throughlines; we may recognize when we see it, but only in person, in the fullness of its situation.

Letter—the specification of spirit—can attempt to capture some of the shapes and guises in which spirit manifests. Though this translation process cannot succeed in full, spirit on its own cannot be legislated, cannot be uniformly instituted as expectation, cannot tile itself across a superorganism.[1] Insofar as it can, it is hard and unaccountable, relies on trust and discretion, defies monitor. So we are stuck with letter.

Insofar as letter invariably fails to capture spirit, each rounding off and uncovered patch of ground, each poor synopsis or compression, shifts the behavior of the organization or individual away from their desired spirit, toward some other, emergent, perverse nonintention. Often in this process, the substitution of letter for spirit is forgotten—there is a slow forgetting that something complex and extraverbal has been surrogated into language, into measurement, into specification, and the surrogate comes to stand in as spirit itself. This process is known as surrogation.

In the story of King Midas, Dionysus promises the king any wish he desires, and Midas wishes that all he touch turn instantly to gold. We know what he “really” was after—voluntarism, for one—but did he even know, precisely, what he wanted when he wished it? Had he specified its parameters and constraints, or assumed a human audience would infer the spirit of his desire? Unfortunately, the total space of intelligence is much broader than the space of human minds: Midas was met only by a trickster god, and his sloppy presentation of spirit into letter became the undoing by which he is known.


Rules, as we will see, do not rid us of the spirit of behavior so much as its letter of expression—what we can call a desire and its outlet. Like the problem in artificial intelligence of the “nearest unblocked strategy,” the values and intentions which gave rise to a now-banned behavior do not disappear on its banning. (Nor, technically speaking, does the banned behavior itself, which has merely had its cost-benefit equation altered to decrease instances.) Instead, these desires have merely been re-channeled into the “nearest unblocked” action.

We can see this in the West’s ongoing drug war, and the explosion of “research chemicals” in the 2000s and 2010s. These chemicals are called analogues because they are “off-by-one”—near-copies that evade the law’s letter even as they deliver similar effects, defying its spirit.

Unfortunately, writing laws based on spirit is intractable; it must be converted to letter, as the spirit depends on interpretation, and is therefore difficult to evaluate. In this sense, letter laws are formalizing reductions which, in their lossy compression, gain only the advantage of minimizing vagueness. In this way they are very similar to job performance metrics or the grade-point average: their major benefit is evading vagueness, but the vagueness has not disappeared from reality, it has merely disappeared from the evaluation. Moreover, the necessity of converting spirit into letter in institutional structures, in order to coordinate superorganisms, gives rise to problems like Goodhart’s Law and surrogation. Spirit is abandoned, and letter becomes the incentive structure (or “reward function”) that governs actor behavior.

One does, occasionally, see spirit-based enforcement—parents and children are an obvious case—but these instances typically require authoritarianism, a dictator who has “authority” in interpreting both the law’s spirit and the accordance of a subject’s actions with it. Such decisions need not be publicly legible, consistent, or fair—in part because they are often enacted on small populations, and in part because the authority does not answer to the public.

  1. Cooperative enterprises organized by prestige economies to a common purpose (Simler 2016). ↩︎