Statecraft: Law, Records, and Administration

Difficulty: free to describe, extremely hard to install. Prerequisites: writing, mathematics and record-keeping, money and finance.

Why it matters

Conquering territory is comparatively easy. Holding it is an administrative problem, and the states that lasted were the ones that solved it.

Alexander conquered from Greece to the Indus in a decade, and his empire fragmented within a year of his death because it was held together by one person's presence. Rome held its territory for centuries with roads, a census, written law, and a professional civil service — and the eastern half ran on that machinery for another thousand years. The Mongols conquered the largest contiguous land empire in history and then, notably, adopted the administrative systems of the people they had conquered, because conquest is a different skill from government.

The pattern is consistent: the technologies of administration outlast the technologies of war, and they're free.

The core problem

A ruler can only personally know a few hundred people and personally see one place at a time. Everything beyond that has to be run through agents you cannot watch, using information you did not gather yourself.

So every institution below is really an answer to one of three questions:

1. What is out there? — census, survey, maps, accounts.
2. How do I make agents act as instructed when I'm not looking? — law, audit, rotation, salaries, courts.
3. How do people know what the rules are? — written law, published standards, courts of record.

Get those three right and the state runs itself. Get them wrong and you have a warlord with a temporary reach.

Part 1: Know what you have

The census and the register

You cannot tax, conscript, or plan without knowing who and what exists. A census is the foundational act of statecraft, and the Domesday Book (1086) — a county-by-county record of who held what land, how many ploughs, how many mills, and what it was worth — is the model. It was, in its own words, so thorough that "there was no single hide nor yard of land, nor indeed one ox nor one cow nor one pig which was left out."

Record, at minimum: households, land, and livestock. Add mills, boats, looms, and forges if you can — those tell you where the productive capacity is.

Repeat it periodically. A register that isn't updated becomes fiction within a generation, and taxing people on a fifty-year-old survey is how you get revolts.

Understand what it costs the people being counted. A census is also the instrument of conscription and taxation, and populations have always known that, which is why they resist and under-report. Expect that, and don't be surprised by it.

Cadastral survey and land registration

Written, mapped records of who holds which piece of land, with boundaries. This does three things at once:

• Makes land taxable systematically rather than by negotiation.
• Makes disputes resolvable by evidence rather than by force or by whoever remembers loudest.
• Makes land transferable and mortgageable, which turns it into capital — see money and finance.

Survey it with the tools in water supply and the geometry in mathematics. Mark boundaries physically with stones, banks, or hedges, and record them with measured bearings and distances, not just "to the old oak."

Be honest about the second edge of this. Written land registration also converts customary and common rights — grazing, gleaning, firewood, which the poorest depend on and which nobody holds a document for — into nothing, because they don't appear on the register. Enclosure across early modern England did exactly that and dispossessed a great many people entirely legally. If you install a land register, record the common rights too, or you will have quietly stolen from the people least able to object.

Maps

A state that cannot picture its own territory cannot govern it. Combine the surveying instruments from water supply and sanitation with the triangulation in mathematics: fix a carefully measured baseline, then build a network of triangles outward from it by measuring angles only. Errors don't accumulate the way they do with chained distances, and this is how every accurate national map was made before aerial survey.

Maps let you plan roads, assess territory, move armies, settle boundary disputes, and — not least — see that a region exists at all.

Part 2: Make agents behave

This is the hard one, and most of the answers are unglamorous.

• Pay officials properly, in money, from the centre. An unpaid or underpaid official will extract from the population, because they have to eat. Corruption is usually a compensation problem before it is a moral one.
• Never farm the taxes. See money and finance. A contractor who has bought the right to collect will extract everything and you will get a fraction and all of the blame.
• Rotate postings and separate powers. Don't let the same person assess the tax, collect it, hold it, and audit it. Rome sent governors for fixed terms; China rotated magistrates and forbade them from serving in their home province, precisely so local kinship ties couldn't capture them.
• Audit, and make the audit independent. Double-entry accounts (see mathematics) are self-checking by construction: if the books don't balance, something is wrong, and the system tells you.
• Written receipts, in duplicate. A taxpayer holding a receipt is a taxpayer who cannot be charged twice. This one practice does more against petty extortion than any inspector.
• Provide an appeal route to the centre. People with a genuine complaint and no channel become people with a genuine grievance and a weapon.

Examinations and merit

The Chinese imperial examination system selected officials on demonstrated competence rather than birth, from roughly the Sui dynasty onward. It was imperfect — it favoured those who could afford years of study, and it rewarded literary skill over administration — but against the alternative of pure hereditary appointment it was a large improvement, and it gave the state a huge pool of trained, loyal, mobile administrators.

The transferable idea: select on tested ability, and make the test public. A hereditary aristocracy is a small talent pool with no incentive to perform.

Part 3: Law

Write it down and publish it

Unwritten custom is whatever the powerful remember it to be. Written, published law is checkable law, and the act of publishing is itself the reform — Hammurabi's stele, the Roman Twelve Tables, and every later code all put the rules where ordinary people could see them, which is what makes appeal possible.

Publish it where people can read it, and — once you have printing — print it cheaply and widely. Law nobody can consult is law nobody can invoke.

The principles worth arguing for

These are political rather than technical, and each is worth more than any machine on this list:

• The same law applies to everyone, including officials and, ideally, the ruler. This is the single most radical and most valuable idea here.
• Predictability over severity. A moderate punishment reliably applied deters far better than a savage one applied erratically. Draconian codes with capital punishment for minor theft produced juries who simply refused to convict.
• Proportionality, and a scale of penalties.
• Evidence over ordeal, oath, or torture. Trial by ordeal, combat, and compurgation determine who is strong or well-connected, not who did it. And torture reliably produces confessions and unreliably produces truth — a tortured person says whatever stops the torture. This is not a modern squeamish view; it was argued in the 18th century on exactly these evidentiary grounds and it is simply correct.
• Courts of record. Written judgments create precedent, and precedent makes outcomes predictable, which is what lets people plan, contract, and invest.
• Contract enforcement. Merchants trade where agreements are enforceable. A reliable commercial court attracts trade the way a good harbour does.

Rule-of-law as an economic technology

This is the non-obvious part. Secure property and enforceable contracts are economic infrastructure. If a ruler can seize your workshop at whim, nobody builds a workshop worth seizing — they hold portable wealth and keep their heads down. If title is secure and contracts hold, people invest, lend, specialize, and build things that take years to pay back.

Every material in this repository is worth more in a society where the person who builds the kiln expects to keep it.

Part 4: Reach — the relay system

The empire's size is set by how fast information travels.

The relay post is the answer, and it is purely organizational: stations at intervals of a day's ride or less, each with fresh horses and riders, so a message travels continuously at the speed of a galloping horse rather than the endurance of one animal.

• Persian angarium (5th c. BCE) — Herodotus's "neither snow nor rain nor heat nor gloom of night" was written about these couriers.
• Roman cursus publicus.
• Mongol yam — stations every 40 km, which let Karakorum communicate across Asia in weeks rather than seasons, and which is a serious part of how a nomadic confederation governed a continent.

Roughly 250–350 km per day is achievable with a good relay network, against perhaps 50 for a single rider. That difference is the difference between a province you govern and a province you nominally own.

Build it with roads, and see communication at distance for optical signalling, which is faster still.

Part 5: Standards

Boring, cheap, and disproportionately valuable.

• Weights and measures, enforced by physical reference standards held publicly and checked periodically. Fraud in measurement is the commonest commercial dispute in every pre-modern economy, and eliminating it is nearly free. Make the standards reproducible from nature so they can be recreated rather than merely copied — see navigation and timekeeping.
• A single calendar, with the Gregorian leap rule.
• A standard currency of consistent fineness.
• Standard documents — deeds, contracts, receipts in fixed forms, which makes them checkable and archivable.
• A capital's time and measures propagated outward by the relay network.

Part 6: What actually holds a conquered territory

Since the framing is conquest, the honest historical answer:

1. Don't destroy what works. The Mongols, Romans, and British all largely kept existing local administration and taxed through it. Replacing a functioning bureaucracy with your own costs more than it yields.
2. Co-opt the local elite. Give them a stake — office, status, a share of revenue. Elites with something to lose police the population for you. This is cynical and it is how it was done.
3. Tax moderately. The most common cause of rebellion is not conquest itself but a tax increase or a new tax after conquest.
4. Respect religion and custom where you can. These provoke revolt out of proportion to their administrative importance.
5. Roads and garrisons, which are the same investment: roads let a small force reach anywhere quickly, which substitutes for a large force everywhere. See land transport for the roads and fortification and logistics for what actually limits a force's reach, which is supply rather than fighting strength.
6. Justice that is better than what came before. A conquered population that can get a fair hearing against your own officials will tolerate a great deal.
7. Feed them in famine. Granaries and relief are cheaper than suppression, and they buy legitimacy that force cannot.

And the thing that actually did most of the conquering, historically, was disease — see medicine and public health. The Americas were not defeated by steel. That is the darkest fact in this repository and it should not be left out of a page about conquest.

Failure modes

Era notes

Written law codes go back to Ur-Nammu (2100 BCE) and Hammurabi (1750 BCE). Census-taking is at least as old. The Persian relay system is 5th century BCE; Roman provincial administration and the cursus publicus run through the imperial period; the Chinese examination system dates from the Sui and Tang; the Domesday survey is 1086.

None of this needs technology you don't already have on this list. It needs writing, arithmetic, and the willingness to bind your own officials — which is the genuinely hard part, because every one of these reforms takes power away from someone who currently has it.

Your knowledge advantages: the audit-and-rotation package for controlling agents you cannot watch; receipts as an anti-corruption tool; independent double-entry audit; triangulation for accurate mapping; the argument that predictable moderate law outperforms savage law; that torture produces confessions rather than truth; recording common rights so a land register doesn't dispossess the poor; and the central insight that institutions outlast people, and that is the only durable form of conquest.