Day 46 · The Evolution of Law and Justice

Law Is Not Deduced From Justice. It Grows Out of Trouble.

12 August 2026 (Wednesday) · BigCat's Time Machine
A stele no judge ever cited, a compression that threw away ninety-five percent, one line in a writ saying «use your discretion for now», and a company lawyer's brief — all four texts were later canonised as the origin of justice.
EVENT · 01

A Stele No Judge Ever CitedThe Code of Hammurabi · c. 1754 BCE, Babylon

c. 1754 BCEBabylon · SusaJean Bottéro 1992

Hammurabi (reigned c. 1792–1750 BCE) spent three decades unifying southern Mesopotamia and raised the stele late in life: basalt, some 2.25 metres tall, 282 provisions. A French expedition under Jacques de Morgan unearthed it at the end of 1901 — not in Babylon, but at Susa in south-western Iran, where the Elamite king Shutruk-Nahhunte had carried it off as loot around 1158 BCE.

Every provision is cast as «if … then …». Penalties are tiered by status: put out the eye of a propertied man (awilum) and lose your own; put out the eye of a dependent (mushkenum) and pay one mina of silver; put out the eye of a slave and pay his owner half the slave's price. The real puzzle lies elsewhere — among the thousands of excavated Old Babylonian trial records, not one cites any article on the stele. Its prologue and epilogue, meanwhile, are the king speaking about himself: to make justice appear in the land, so that the strong may not oppress the weak.

Were it genuinely enforceable statute, judgments should have left citation traces. Jean Bottéro (Mesopotamia, 1992) therefore reads it as a royal specimen of justice and a monumental proclamation — the rhetoric of kingship rather than positive law. Raymond Westbrook and Martha Roth counter that it crystallised the consensus of contemporary practice: judges ruling by custom carried that same scale in their heads and had no need to cite it. Judicial practice has to be read from other material — the Fengzhenshi documents among the Shuihudi Qin bamboo slips, excavated in Hubei in 1975, record procedures for inspection and interrogation. That is what law looks like when it lands, and it does not look like a stele.

The values on the wall versus the real promotion criteria; the clause in the compliance handbook versus the check that actually blocks a deploy. The second one is the system's real behaviour.

The authority of a text is not the same as the text being used.
When was the last time the written standard your team keeps was genuinely produced as the basis for a decision?
EVENT · 02

Two Thousand Volumes Into 150,000 Lines — and No UndoJustinian's Digest · 530–533, Constantinople

530–533Constantinopleius commune

The substance of classical Roman law was not legislation but the jurists' answers to particular cases (responsa). By the sixth century the writings of Ulpian, Paulus, Papinian and others had piled into a mountain of mutual contradiction, and which authority a court cited was close to luck. Justinian (reigned 527–565) resolved to close the question; his instrument was the quaestor Tribonian.

On 13 February 528 he ordered a code of imperial constitutions; in December 530 he authorised Tribonian to convene a commission of sixteen to compile the Digest, allowing ten years. It was finished in three: roughly two thousand volumes and three million lines cut down to some 150,000 — less than one part in twenty preserved. It took effect on 16 December 533, together with a decree that the original works could no longer be cited in court and that commentary on the compilation was forbidden. What had been cut gradually disappeared. Five centuries later, around the 1070s, an ancient manuscript resurfaced in Italy; Irnerius began lecturing on it at Bologna, and the European ius commune was under way.

Without the compilation, Roman law would likely survive only in fragments, as Athenian law does. With it, the original context is gone for good — every line of Ulpian we read has passed through Tribonian's scissors, and we hold no master copy to check it against. In the first half of the twentieth century, scholars after Mommsen hunted enthusiastically for «interpolations», at one point believing the Digest had been rewritten throughout; after the 1970s the fashion receded and the mainstream came to see the changes as mostly excision. The argument cannot be settled precisely because the comparison text was destroyed by that same decree.

Compressing scattered documents into a single source of truth buys searchability and enforceability; the price is edge cases you can never roll back. Knowledge-base consolidation and model distillation are the same trade — the only difference is whether anyone kept the original.

Compression buys usability and charges you irreversible context.
That single source of truth you maintain — do the deleted parts still exist somewhere, or did you also issue a decree forbidding citation of the originals?
EVENT · 03

The Jury Was a Stopgap in an Institutional VacuumFrom Ordeal to Jury · 1215–1219, Rome and England

1215–1219Rome · EnglandRobert Bartlett 1986

In early thirteenth-century England, criminal guilt was settled by ordeal: carry a red-hot iron nine paces, then have the hand bound and inspected after three days for festering; or be bound and lowered into consecrated water, where sinking meant innocence. The judge was nominally God, but the officiating and the interpretation belonged to the priest present. In November 1215 Pope Innocent III convened the Fourth Lateran Council.

Canon 18 forbade clergy to pronounce the blessings that made an ordeal work — a single prohibition removed the ordeal's entire source of legitimacy. English criminal justice promptly seized up. In January 1219 the regency government of the boy king Henry III sent writs to the itinerant justices conceding that nothing had yet been settled, and instructing them to deal with prisoners awaiting trial at their own discretion for the time being. The judges reached for what lay nearest, converting the jury of neighbours already used in land disputes and in presentment into a finder of fact. It was no right at first: the accused had to «consent» to jury trial, and those who refused were subjected under the Statute of Westminster of 1275 to peine forte et dure — crushing weights applied until they answered.

The same vacuum was filled with something else on the Continent. Revived Roman-canon procedure demanded statutory proof: two eyewitnesses, or a confession. Where witnesses were lacking, confession was the only route, and torture became an institutional necessity of evidence-gathering. Robert Bartlett (Trial by Fire and Water, 1986) argues that the ordeal did not die because reason defeated superstition, but because the church withdrew and each region improvised its own replacement. Without Canon 18 the English jury might have been delayed a century or two — the fork between common law and civil law begins as an act of filling a blank.

When an old anchor of trust fails, what takes its place is usually whatever was already lying around, later ratified as principle. Backstopping AI output with human review today is less a design than the only thing currently available.

How an institutional vacuum gets filled depends on what is at hand, not on what would be best.
How long has that «temporary» mechanism in your system been running? Is it still being paid down as debt, or is it now maintained as a design principle?
EVENT · 04

International Law's First Argument Was a Brief for Stolen CargoGrotius and the Santa Catarina · 1603–1609

1603–1609Singapore Strait · Amsterdamvan Ittersum 2006

On 25 February 1603 Jacob van Heemskerck, a captain of the Dutch East India Company (VOC), seized the Portuguese carrack Santa Catarina in the Singapore Strait. Her cargo of Chinese silk and porcelain fetched around three million guilders at auction in Amsterdam — close to half the VOC's founding capital. The trouble followed at once: the Dutch and the Portuguese were not formally at war, so on what authority does a trading company capture another nation's ship? Some Mennonite shareholders demanded their money back on grounds of conscience.

The VOC engaged Hugo Grotius (1583–1645), a lawyer barely into his twenties, to build the case. Between 1604 and 1606 he wrote De Iure Praedae: under natural law a private party may execute justice itself where no public authority is available to appeal to; and the sea, by its nature, can be neither occupied nor exhausted, so the Portuguese monopoly over the East Indies route was itself unlawful. The manuscript went unpublished at the time; only its twelfth chapter appeared anonymously in 1609, as Mare Liberum. With De Iure Belli ac Pacis in 1625, Grotius became the father of international law.

Freedom of the seas might well have been proposed without this capture — but its first systematic defence as a universal principle did emerge from an interested brief. And the position tracks power: in 1635 England's John Selden argued in Mare Clausum that waters can be sovereign territory, at a time when English shipping was being squeezed by the Dutch; a century later, roles reversed, it was the English who championed free navigation. Martine van Ittersum (Profit and Principle, 2006) uses the manuscripts to show Grotius acting for years as the VOC's retained pen; Richard Tuck (The Rights of War and Peace, 1999) holds his universalism to be sincere, and merely convenient for expansion.

Open standards, data portability, «fair use» for training data — the loudest advocates are usually the current challengers, and the position often reverses once they hold the incumbent seat. Read the position after you read the book.

Universal principles are often born as briefs for particular interests — which does not invalidate them, but does determine where their edges lie.
The industry principle you most agree with: who first proposed it, and from what position? Which side were they on at the time?

Four Texts: The Trouble Came First, the Principle Afterwards

Each was written to settle one concrete problem; posterity read all four as general statements about justice.
Text
The trouble it actually solved
What was really written
Later canonised as
c. 1754 BCE
Hammurabi's stele
a king needing to prove he was just
282 status-tiered specimen rulings
«the first written law code»
530–533
the Digest
jurists contradicting one another
a 95%-cut compilation, rivals banned
«the canon of Roman law»
1215–1219
the jury
nobody left with standing to judge
one line in a writ: use your discretion
«the bulwark of liberty»
1603–1609
Mare Liberum
one shipload of loot needing a story
chapter twelve of an unpublished brief
«the origin of international law»

Going Deeper

Why is the title «the first law code» worth distrusting?
Because it conflates two things: a text existing, and a text being in force. Hammurabi's stele exists but appears in no judgment; the Digest took force because the same decree suppressed its competitors. To judge whether a rule is actually running, do not read what it says — look for citation traces, in judgments, tickets, incident reviews. Writing it down only makes it a candidate; being cited is what ships it.
Jury and torture came out of the same prohibition. What does that say about institutional evolution?
That institutions are not designed so much as forced by constraints. One canon created the same vacuum in two places; England had a jury of neighbours lying around, the Continent had the revived Roman-canon rules of proof, so one grew a jury system and the other grew judicial torture. That is what path dependence means: a small difference in starting conditions amplified over centuries into two incommensurable systems — the fork is rarely a choice, it is an inventory.
Taken together, where does the legitimacy of law actually come from?
Four different sources here. The stele rests on proclamation (the king says it is just); the Digest on monopoly (no rival text may be cited); the jury on custom (do it long enough and it becomes a right); freedom of the seas on argument (derived from natural law). The most fragile is argument — it reverses with the arguer's circumstances. The most durable is custom, because it does not require anyone to go on believing a particular reason.
If we were writing rules for AI systems today, what is the most concrete warning these four episodes give?
Do not write only the principles — design the citation trace alongside them. Hammurabi's stele cannot be tested for effect precisely because nobody ever cited it, and most AI governance documents are in the same condition: well written, with no mechanism recording which actual decision they were invoked in. An enforceable rule needs three things: a trigger, an adjudicator, and a record. Missing any one of them, what you have written is a stele.