Long before the spreadsheet, merchants, priests and kings built tools to decide under uncertainty: clay tokens sealed in clay envelopes, liver models, sea loans that priced the autumn storms, partnership contracts and double-entry ledgers. Part 1 follows the Age of Record, when the hardest problem in finance was making a claim survive time, distance and dispute.

Conviction Before the Number

Every investment committee ends the same way: someone has to say yes. The memo, the model and the data room exist to make that yes defensible, to the partners in the room and to the limited partners who will ask about it years later. We call the result conviction: a belief strong enough to commit capital, and shared enough that others will commit alongside you.

Conviction is not a feeling. Across five thousand years of trade, it has been manufactured by tools. In this series, a decision-making tool is any technology, ritual, contract or institution that changes what a decision-maker can remember, verify, price, challenge or agree on. Some are obvious, like the ledger. Others are easy to miss: a clay envelope that let a buyer check a seller's claim, a seasonal clause in a Greek sea loan, a divination rite that forced a council to argue before it acted.

The series follows these tools across three ages. Part 1, the Age of Record, runs from roughly 7500 BCE to 1500 CE, when the central problem was memory. Part 2, the Age of Number, covers the seventeenth to twentieth centuries, when probability, prices and models turned records into forecasts. Part 3, the Age of Judgment, examines the last fifty years: decision science mapped where human reasoning goes wrong, AI began to take on the machine work of analysis, and the question became what remains ours to decide.

A note on judgment

That last word deserves care, because it is easily misread. In everyday use, judgment carries heavy connotations: a final verdict handed down, or the "human judgment" that behavioural research shows to be biased and noisy. We mean something older. The word entered English in the mid-thirteenth century from the Old French jugement: a legal trial, a sentence, and also the power to decide. It comes from the verb jugier, "to judge", and ultimately from the Latin iudicare.

In this series, judgment means that power to decide, exercised by people who carry responsibility for the outcome. Records, numbers and models inform it; none of them holds it. Every tool in this story exists to equip that power, and conviction is what it produces when it is well equipped.

Throughout, we track five recurring functions. We call them the pillars of conviction:

Every tool in this story strengthened at least one of them. Most of the failures began when one of them was allowed to weaken. And every tool, in its own way, attacked the three problems deal teams still face today: how long it takes to reach conviction, how good the research behind it is, and how well the instruments of decision serve the people who use them. Speed, depth and tooling are the thread that runs from a clay token to an AI-drafted memo. Follow it, and the story of finance becomes the story of how we learned to decide better, faster.

Era I. Clay and Contract: Mesopotamia, c. 7500 to 1700 BCE

The token and the envelope

Our story begins in the farming villages of the Near East, with a decision tool small enough to fit in a closed hand. From around 7500 BCE, farming communities across the Near East used small clay tokens (cones, spheres, discs), each standing for a measure of goods such as grain or livestock. The archaeologist Denise Schmandt-Besserat, in Before Writing (University of Texas Press, 1992), argued that these tokens were the ancestors of writing itself.

Tokens could record a claim, but they could not prove it. The decisive step came around 3500 to 3300 BCE. Parties to an agreement sealed tokens inside a hollow clay ball, a bulla, and rolled their seals across the outside. The envelope solved a problem every deal team still knows: how do two parties who do not fully trust each other fix what was promised? The contents could not be altered without breaking the seal, and anyone in doubt could break it and count. Soon the tokens were pressed into the surface of the envelope, then onto flat tablets. By about 3200 BCE the impressions had become signs, and at Uruk the signs became proto-cuneiform. Of the roughly 4,900 proto-cuneiform texts catalogued from that period, about 85 percent are administrative. Writing was invented largely to keep accounts. (The direct line from token to sign is debated; see Piotr Zimansky's 1993 review.)

The bulla is the first tool of Evidence: a claim, a source, and a way to check one against the other. It also saved time: instead of relitigating a dispute from memory, the parties could open the envelope and settle it. It is the first shortcut to conviction in our story.

Kanesh: a venture fund in a sack

Once claims could be recorded and checked, they could also travel. Around four thousand years ago, merchants from the city of Aššur on the Tigris ran a trade network reaching Kanesh, modern Kültepe in central Anatolia. They sent tin and textiles by donkey caravan and brought back silver and gold. Excavations have recovered around 23,000 of their clay tablets: letters, contracts, debt notes and court verdicts, most written between about 1950 and 1836 BCE. It is one of the richest private business archives of the ancient world, reconstructed by Mogens Trolle Larsen in Ancient Kanesh (Cambridge University Press, 2015).

Among those tablets lies the instrument that most resembles a modern fund: the naruqqum, literally "the sack". About a dozen investors, sometimes including the trader himself, pooled capital denominated in gold and entrusted it to a single merchant for a fixed term of ten years or more. One partnership for a trader named Amur-Ištar ran for twelve years from 1908 BCE, with 30 minas of gold. Shares could be inherited or sold, and new investors could buy in. The managing trader took a share of the profits for his work. William Goetzmann, in Money Changes Everything (Princeton University Press, 2016), reads the naruqqum as a precursor of the joint-stock company. A modern reader sees something even closer: a closed-end fund with a long life, a general partner, limited partners and a carried interest.

Which raises the question every LP still asks: what makes a decade-long commitment possible? Nobody commits gold for ten years on trust alone. What made the naruqqum possible was the archive around it: letters reporting prices in Aššur and Anatolia, sealed contracts naming witnesses, and records that heirs could produce before a judge. The archive was the fund's Memory, and it was also its governance. Investors were not only men: Cécile Michel's Women of Aššur and Kaneš (2020) documents women who financed, lent to and managed this trade.

Law as a risk model

If archives gave private investors memory, the state soon gave them rules. Around 1754 BCE, Hammurabi of Babylon published the law code now preserved on a basalt stele in the Louvre. Its credit provisions read like an early term sheet. Interest was capped at 20 percent on silver and 33⅓ percent on grain, reflecting the higher risk of harvest loans. Law 48 addressed catastrophe: if a storm or drought destroyed a debtor's crop, he owed neither grain nor interest that year, and the debt tablet was rewritten. It is a force majeure clause, written nearly four thousand years before the term existed. By fixing the rules in advance, the code turned a whole class of uncertainty into Terms that lender and borrower could price in advance, rather than negotiate after every catastrophe.

The first price series

Rules fixed the terms of a deal. What was still missing was a view of the market over time. For six centuries, from 652 to 61 BCE, astronomers in Babylon kept diaries of the night sky. Alongside planetary positions and river levels, from 385 BCE they regularly recorded the market prices of six commodities: barley, dates, mustard, cress, sesame and wool. The historian R.J. van der Spek and colleagues have turned these entries into a price series spanning more than three centuries. The scribes believed heaven and market moved together. They were wrong about the cause and right about the method: systematic observation, recorded at regular intervals, is how patterns become visible. Mesopotamia had, in effect, built the first research desk. What no record could do was look into the future, and for that, people turned to the gods.

Era II. Asking the Gods: Structured Uncertainty, c. 2000 to 400 BCE

A database of omens

Faced with a future that no archive could reveal, Mesopotamian kings consulted specialists. In 1935 the archaeologist André Parrot found 32 clay models of sheep livers in the royal palace at Mari, on the Euphrates. Mesopotamian diviners practised extispicy: they examined the liver of a sacrificed sheep and read its marks against a standardized map of zones. Some of the Mari models carry "historical omens", records of the signs observed before specific past events. Over centuries, scribes compiled thousands of these into omen series written as conditionals: if the liver shows this, then that will follow.

It is easy to dismiss extispicy as superstition. As a decision process, though, it had features modern committees struggle to reproduce. The template was standardized, so two diviners examined the same zones in the same order. Readings were tied to written precedent. And the procedure could be repeated, so a king could ask the same question twice and compare answers.

Recording the forecast, and the outcome

Half a world away, the same need produced an even more disciplined method. In China, the Shang kings at Anyang divined with turtle plastrons and ox shoulder blades, applying heat and reading the cracks. The inscriptions carved on these oracle bones, from the reign of Wu Ding (about 1250 BCE) to the fall of the Shang around 1046 BCE, follow a fixed structure that the historian David Keightley, in Sources of Shang History (1978), described as preface, charge, prognostication and verification. The diviners recorded the question, the king's forecast and, in many cases, what actually happened.

That last element is remarkable. It is a forecast log, the discipline that Philip Tetlock's forecasting tournaments would rediscover three thousand years later, and that we will meet again in Part 3. A recorded prediction can be scored. An unrecorded one tends to be remembered as right. (Scholars still debate whether the charges were questions or statements of intent; either way, the record linked expectation to outcome.)

The ambiguity engine

Recording a forecast is one discipline; arguing over it is another. In 480 BCE, with the Persian army advancing, Athens consulted the oracle at Delphi. According to Herodotus, the answer said that a "wooden wall" alone would not fall. Some Athenians took it to mean the old thorn hedge around the Acropolis. Themistocles argued that the wooden wall was the fleet, and persuaded the assembly to stake the city on its ships. The victory at Salamis followed.

Historians suspect the oracle's wording was polished after the event. The episode still captures something essential. The oracle did not decide; it framed a question and returned it to a deliberating body, ambiguous enough to require argument. What decided the matter was Challenge: rival interpretations tested in public before Athens committed its fleet and its future.

Randomness as a strategy

The most surprising defence of divination came much later, from the twentieth century. In 1957 the anthropologist Omar Khayyam Moore published a short paper, "Divination: A New Perspective", in American Anthropologist. He studied Naskapi hunters in Labrador, who chose where to hunt by heating a caribou shoulder blade and reading its cracks. Moore proposed, cautiously, that the cracks worked as a randomizer: they kept the hunters from settling into predictable patterns that the game could learn to avoid. Michael Dove later described something similar among the Kantu' of Borneo, whose bird augury spread farming plots across sites and so diversified the risk of crop failure.

Read this way, divination was an early debiasing technology. It broke the pull of recent experience, introduced diversification, and gave groups a ritual that ended deliberation once it had run its course. The theology was wrong. Some of the decision science was surprisingly right. Yet a ritual could only frame a decision. To commit capital across open water, merchants needed something they could price.

Era III. Pricing the Voyage: Athens and Rome, c. 400 BCE to 200 CE

The sea loan

That something was the contract. Classical Athens fed itself with grain shipped from the Black Sea, and financed that trade with a distinctive contract: the maritime loan. A lender advanced money against a ship's cargo. If the ship arrived safely, the borrower repaid the principal plus a steep premium. If it was lost, the debt was cancelled. Credit and insurance were fused into a single contingent claim.

One such contract survives, quoted in the speech Against Lacritus attributed to Demosthenes. Two lenders advanced 3,000 drachmas for a round voyage to the Black Sea. The rate was 225 drachmas per thousand, 22.5 percent for the voyage, rising to 300 per thousand if the ship left the Pontus after the rising of Arcturus in mid-September, when autumn storms made the passage more dangerous. It is a risk premium priced by season, written down and enforceable in court.

Edward E. Cohen's Athenian Economy and Society: A Banking Perspective (Princeton University Press, 1992) shows how such loans sat within a network of bankers, guarantors and witnesses. What matters for our story is the conversion: an unknowable event (a storm, a pirate, a wreck) became Terms both parties could evaluate before committing.

Shares in the Republic

Athens priced the risk of a single voyage. Rome took the pooling of capital further. The Republic outsourced tax collection and public works to companies of publicani, the societates publicanorum. These firms raised capital from investors who held partes, shares, without being liable as full partners. Cicero mentions shares bought at "a very high price", which suggests a market in them. Ulrike Malmendier's study "Law and Finance 'at the Origin'" (Journal of Economic Literature, 2009) argues these companies anticipated features of the modern corporation, including a separate legal persona and transferable stakes. How actively the shares traded is contested.

The principle holds either way. Once ownership is divided among many, the decision to invest needs a shared, legible basis that outsiders can assess without sailing the ship or collecting the tax themselves. That is Consensus turned into an institution, and with it a new demand: research that strangers could rely on. Over the next thousand years, the tools to meet that demand would come from the east.

Era IV. Zero and the Networks: India, the Islamic World, China and the Andes, c. 100 to 1250 CE

Forty ways to embezzle

The Arthashastra, a Sanskrit treatise on statecraft attributed to Kautilya, shows how seriously the ancient world took the quality of the numbers behind a decision. It devotes much of its second book to the treasury, revenue and the superintendents of state enterprises. Its chapter on the bureau of accounts lists forty ways officials can embezzle funds: a taxonomy of fraud written as an audit checklist. Its date is contested; tradition places it in the fourth century BCE, while Patrick Olivelle's scholarship dates most of the text to the early centuries CE. Either way, it treats verification as a permanent function of government, not an occasional inquiry.

Fortunes, debts and nothing

Audits need arithmetic, and the most important arithmetic of all was being invented on the same subcontinent. In 628 CE the mathematician Brahmagupta completed the Brahmasphutasiddhanta at Bhillamala in north-west India. It set out rules for calculating with zero and with negative numbers, which he called "debts", alongside positive numbers, "fortunes". A debt minus zero is a debt; a fortune minus zero is a fortune. Negative numbers entered mathematics through the language of the balance sheet.

The numerals travelled west through the Islamic world. Al-Khwarizmi's treatise on Hindu reckoning, written in Baghdad around 825, carried them to Arabic-speaking scholars and later, in Latin translation, to Europe. Place-value notation made calculation possible on paper by anyone trained in it, and made it auditable in a way the counting board never was: the working stayed on the page. Analysis became faster, and for the first time anyone could check it.

Partnership and reputation

Numbers were not the only thing the Islamic world passed on. It also refined the contract that financed Mediterranean trade for centuries. In the mudaraba, or qirad, an investor supplied capital and a merchant supplied labour; profits were shared by an agreed ratio, while losses of capital fell on the investor. Abraham Udovitch's Partnership and Profit in Medieval Islam (1970) argued that this form shaped the Italian commenda, though Roman, Byzantine and Jewish precedents also have their advocates.

We know how these partnerships worked thanks to the Cairo Geniza, a storeroom in a Fustat synagogue where written documents were deposited rather than destroyed. From it, S.D. Goitein's A Mediterranean Society reconstructed the letters of eleventh-century traders reporting prices, shipments and the conduct of agents across the sea. In a 1993 paper in the American Economic Review, Avner Greif argued that these "Maghribi traders" enforced honesty through a coalition: an agent who cheated one merchant would be shunned by all. Jeremy Edwards and Sheilagh Ogilvie have challenged that reading, showing that the traders also used courts and formal contracts. Both sides agree on the essential: shared information across a network was the basis for trusting a distant agent. Memory and Consensus worked together.

Claims without metal, records without ink

While Mediterranean traders built trust through letters, merchants at the other end of Eurasia were learning to trust paper itself. In Sichuan they issued paper deposit receipts, jiaozi, to avoid carrying heavy iron coins. In 1024 the Song state took over their issue, creating the first government paper money. A note is a pure claim: its value rests on confidence in the issuer's records and restraint.

On another continent entirely, a civilisation showed that rigour did not even require writing. In the Andes, the Inca state ran an empire of millions without alphabetic writing, using khipus, knotted cords that recorded census and tribute. In 2018 Manuel Medrano and Gary Urton matched six colonial-era khipus to a Spanish census of 1670, showing that the knots encoded the tribute owed by named individuals. A record need not be written to be rigorous.

Fibonacci brings the method home

The threads of this era (numerals, partnerships, trade networks) converged on an Italian port. In 1202 Leonardo of Pisa, known as Fibonacci, published Liber Abaci. He had learned Hindu-Arabic numerals in North Africa, where his father represented Pisan merchants, and his book taught Italian traders to calculate partnerships, currency conversions, interest and profit shares. William Goetzmann's paper "Fibonacci and the Financial Revolution" (NBER, 2004) argues that it also contains the first known present-value analysis, a method for comparing cash flows received at different times. If Goetzmann is right, the core of every DCF model reached Europe in a merchant's arithmetic manual. Italy now had the numbers. What its merchants did with them would shape finance for the next three centuries.

Era V. The Merchant Revolution: Italy, c. 1150 to 1500

The commenda

They began by financing voyages. In Genoa and Venice, the commenda became the standard way to finance a voyage. In its classic form, a resident investor provided all the capital and a travelling merchant provided the labour; on return, the investor took three-quarters of the profit and the traveller one quarter. The split is incentive design: the merchant shares the upside, and the investor's loss is capped at the capital committed. The carried interest of modern venture and buyout funds descends, conceptually, from these contracts.

Pricing reputation

As voyages multiplied, Italian merchants separated insurance from lending. Giovanni Ceccarelli's research on late-medieval marine insurance shows underwriters pricing risk with no probability theory at all, and with striking precision. A Venetian galley paid premiums of around 1.5 percent, where a sailing ship on the same route paid 3 to 4 percent. In 1399 a captain with a poor reputation paid 14 percent when the usual rate was 5 to 8. In 1385 a ship rumoured lost was insured at 85 percent. Premiums moved with vessel type, track record and the latest news. Underwriters were pricing information, and merchants learned the inputs from handbooks such as Francesco Pegolotti's Pratica della mercatura (c. 1310 to 1340), a guide to routes, weights, exchange rates, duties and how to test the quality of goods.

In the name of God and of profit

If underwriters priced information, the most successful merchants made a business of collecting it. The archive of Francesco Datini, a merchant of Prato who died in 1410, was rediscovered in his house in 1870. It holds about 150,000 letters, more than 500 account books and ledgers, some 300 partnership deeds and around 400 insurance policies. Iris Origo's The Merchant of Prato (1957) drew from it the portrait of a man who ran companies from Avignon to Barcelona by correspondence, demanding constant reports on prices, cargoes and partners. His ledgers open with the invocation "in the name of God and of profit". Datini's edge was not capital. It was information: more of it, more often, and filed so it could be found again. In today's language, he had built a faster route to conviction than his rivals.

The Medici control system

Datini ran a network of companies; the Medici turned that network into a system of control. The Medici Bank, founded in 1397, became the largest financial house in Europe. Its structure, reconstructed by Raymond de Roover in The Rise and Decline of the Medici Bank (Harvard University Press, 1963), was a set of separate partnerships. Each branch (Rome, Venice, Geneva, Bruges, London and others) was a distinct firm, in which the Medici usually held at least half the capital and the branch manager a minority stake. Managers sent annual balance sheets to Florence, where the general manager reviewed them, questioned doubtful debts and approved new lines of business.

It was a decision system: local judgment, central review, aligned economics. De Roover also traced its decline. In the late fifteenth century, managers in Bruges and London lent heavily to rulers, among them Charles the Bold of Burgundy and Edward IV of England, and central review failed to stop them. When challenge weakens and incentives drift, a strong control system decays into a formality.

The method of Venice

The tool that would outlast them all was being codified at the same time. In 1494 the Franciscan friar Luca Pacioli published in Venice his Summa de arithmetica, geometria, proportioni et proportionalita. A section titled Particularis de computis et scripturis described double-entry bookkeeping, which Pacioli called the method of Venice. He did not invent it: the Farolfi company ledger of 1299 to 1300 already shows complete double entry. Print spread it across Europe.

Double entry records every transaction twice, as debit and credit, so the books must balance. An error or a fraud in one place leaves a trace in another; the system checks itself. Werner Sombart claimed capitalism was unthinkable without it. Basil Yamey replied in 1949 that it was neither necessary nor sufficient. Jacob Soll's The Reckoning (2014) offers the most useful reading for investors: accounting mattered most where it was tied to accountability, where someone actually read the books and acted on what they found.

What the Age of Record Taught Us

For five thousand years, the tools that produced conviction did one thing above all: they made the basis of a decision outlast the moment of deciding. Each pillar had its instruments, and each instrument, in its time, shortened the road to conviction, deepened the research behind it, or gave decision-makers a better tool to work with.

The parallels with a modern deal process are not decorative. A data room is a descendant of the bulla: a sealed set of claims that a counterparty can open and test. An investment memo carries the burden of the Kanesh letters: it must let someone who was not in the room understand, years later, why capital was committed. And the Medici decline is the oldest version of a story every committee knows: controls that nobody enforces become paperwork.

These tools shared one limit. They recorded what had happened. A Kanesh merchant knew last season's tin price; he could not estimate the probability that a caravan would be lost. Medieval underwriters priced risk by experience, reputation and rumour, not calculation. Records could tell a decision-maker what was true. They could not tell him what was likely.

In the summer of 1654, two French mathematicians, Blaise Pascal and Pierre de Fermat, began exchanging letters about a gambling puzzle: how to divide the stakes of a game interrupted before it ends. Their answer would change how the world thinks about the future. That is where Part 2, The Age of Number, begins.

References and Further Reading

Mesopotamia

1. Schmandt-Besserat, D. Before Writing (University of Texas Press, 1992); summary: Tokens and Writing: the Cognitive Development
2. Zimansky, P. Review of Before Writing, Journal of Field Archaeology 20(4), 1993
3. Unicode Consortium / CDLI. Proposal to encode proto-cuneiform, 2019 (corpus statistics)
4. Larsen, M.T. Ancient Kanesh (Cambridge University Press, 2015); review, New York Journal of Books
5. Larsen, M.T. "Partnerships in the Old Assyrian Trade", Iraq 39, 1977
6. Dercksen, J.G. "The Golden Interval of Old Assyrian Trade", NINO/Peeters, 2022
7. Michel, C. Women of Aššur and Kaneš (SBL Press, 2020)
8. Goetzmann, W. Money Changes Everything (Princeton University Press, 2016)
9. The Code of Hammurabi, trans. L.W. King, Avalon Project, Yale Law School
10. van der Spek, R.J. et al. Commodity Prices in Ancient Babylon, 385 to 61 BC

Divination and decision

11. Universität Hamburg CSMC. Liver models from Mari, Manuscript of the Month 46
12. Keightley, D. Sources of Shang History (University of California Press, 1978); Inscribed turtle plastron, Smithsonian
13. Herodotus, Histories 7.141 to 7.143
14. Moore, O.K. "Divination: A New Perspective", American Anthropologist 59, 1957; discussed in Shallit, J. "Randomized Algorithms in 'Primitive' Cultures"
15. Dove, M.R. "Uncertainty, Humility, and Adaptation in the Tropical Forest: The Agricultural Augury of the Kantu'", Ethnology 32(2), 1993

Athens and Rome

16. Demosthenes 35, Against Lacritus
17. Cohen, E.E. Athenian Economy and Society: A Banking Perspective (Princeton University Press, 1992)
18. Malmendier, U. "Law and Finance 'at the Origin'", Journal of Economic Literature 47(4), 2009

India, the Islamic world, China and the Andes

19. Olivelle, P. King, Governance, and Law in Ancient India: Kauṭilya's Arthaśāstra (Oxford University Press, 2013)
20. MacTutor. Brahmagupta, University of St Andrews
21. Udovitch, A.L. Partnership and Profit in Medieval Islam (Princeton University Press, 1970)
22. Goitein, S.D. A Mediterranean Society, vol. I (University of California Press, 1967)
23. Greif, A. "Contract Enforceability and Economic Institutions in Early Trade: The Maghribi Traders' Coalition", American Economic Review 83(3), 1993
24. Edwards, J. and Ogilvie, S. "Contract Enforcement, Institutions and Social Capital: the Maghribi Traders Reappraised"
25. China celebrates paper currency anniversary, The Standard, January 2024
26. Medrano, M. and Urton, G. "Toward the Decipherment of a Set of Mid-Colonial Khipus from the Santa Valley, Coastal Peru", Ethnohistory 65(1), 2018; Harvard Gazette coverage
27. Goetzmann, W. "Fibonacci and the Financial Revolution", NBER Working Paper 10352, 2004

Italy

28. Vicente García, J. "The Commenda Contract", Oxford Business Law Blog, 2022
29. Ceccarelli, G. "The Price for Risk-Taking: Marine Insurance and Probability Calculus in the Late Middle Ages", Journal Électronique d'Histoire des Probabilités et de la Statistique, 2007
30. Pegolotti, F.B. La pratica della mercatura, ed. A. Evans (Medieval Academy of America, 1936)
31. Britannica. Francesco di Marco Datini; Origo, I. The Merchant of Prato (1957)
32. de Roover, R. The Rise and Decline of the Medici Bank (Harvard University Press, 1963); Edler de Roover, F. "The Medici Bank", 1943
33. Pacioli's Summa, 1494, HistoryofInformation.com
34. Lee, G.A. "The Coming of Age of Double Entry: The Giovanni Farolfi Ledger of 1299 to 1300", Accounting Historians Journal 4(2), 1977
35. Napier, C. "Basil Yamey: Historian of Accounting", LSE
36. Soll, J. The Reckoning (Basic Books, 2014)

‍

‍