Four Ways a Promise Became Money

The Pactolus river yielded electrum, a natural gold-silver alloy whose gold content swung between roughly 30 and 55 percent, so every trade began with weighing and assaying — uncertainty about fineness was itself a tax. Croesus (r. c. 560–546 BCE), the last king of Lydia, attacked exactly that: the refinery excavated at Sardis (Ramage and Craddock, 2000) shows the Lydians could separate gold from silver and strike coins of stable fineness in pure gold and pure silver — history's first bimetallic coinage.

The decisive move was not the metallurgy but the stamp punched into the metal. A stamp does not stop forgery; it does something else — it shifts the cost of "verify it again every time" onto the king, once and for all, and the king's return is seigniorage. In 546 BCE Cyrus stormed Sardis and Croesus lost his kingdom, but the monetary form outlived him — the Persian daric ran on the same logic, and the Greek city-states followed.

Platform identity verification and signed open-source package distribution do the same work: compressing "check your counterparty every time" into "check one endorser once" — and the verification cost saved is exactly the endorser's room to charge.

Sichuan ran on iron cash — the court did not want copper coin draining out of the province — and iron cash was cheap and heavy enough that a single bolt of silk cost tens of catties of metal to carry. In the late tenth century sixteen Chengdu merchant houses jointly issued jiaozi, deposit notes redeemable in iron cash: private warehouse receipts, in effect. Then came the predictable sequence — principal diverted, redemptions refused, houses failing.

In 1023 the court established the Yizhou Jiaozi Bureau and took issuance into state hands, laying down three rules that read as a prototype central bank: issue by term — each term three years, old notes exchanged for new at expiry, which gave the currency a recall mechanism; a cap — roughly 1.25 million strings per term; and a reserve fund — 360,000 strings of iron cash held back, about 28 percent of the issue, the first fractional reserve ever written into law. It held for close to a century. What broke it was not the design but the fiscal position: during the Chongning era the notes were renamed qianyin and pushed into Shaanxi to fund campaigns, and the term limits and caps fell one after another; Southern Song huizi and the Yuan paper notes later repeated the pattern.

Stablecoin reserve disclosures and redemption promises pose the same problem: issuance costs nothing, and the difficulty is binding the issuer — especially when the issuer is also the largest spender.

Giovanni di Bicci de' Medici founded the Medici Bank in 1397, and its organizational form was more advanced than its bookkeeping: each branch was a separate partnership, with the head office holding the controlling stake and branch managers buying in for a share of profits — liability and incentive both partitioned. In 1494 Luca Pacioli published his Summa de arithmetica in Venice, one chapter of which gave the first complete printed description of Venetian double-entry bookkeeping. He did not invent it — thirteenth-century Genoese ledgers already show the form — what he did was standardize it and make it reproducible.

The heart of double-entry is not arithmetic but redundant checking: every transaction is entered twice, on both sides, and a trial balance that fails to balance signals either an error or a fraud. Its real consequence is that ownership and management can separate — a financier need not be present to audit an agent in Bruges. Yet in that same year, 1494, the Medici Bank ended along with the family's expulsion from Florence. Raymond de Roover's The Rise and Decline of the Medici Bank (1963), built on the surviving account books, reaches a bleak conclusion: the bad debts were concentrated in political loans to Charles the Bold of Burgundy and Edward IV of England — the ledgers recorded the exposure perfectly clearly, and nobody had the standing to refuse those clients.

Observability dashboards, risk scores, audit logs — most post-incident reviews find the metric had been flashing for a while. What is missing is rarely the data; it is the authority for a front-line operator to tell a high-pressure client no.

In July 1944, with the war still unfinished, delegates from 44 nations met to rebuild the monetary order. Two plans were on the table. Keynes proposed an International Clearing Union with an accounting unit, the bancor, built around symmetric adjustment: deficit countries must tighten, and surplus countries would pay a charge on accumulated balances, forcing them to expand or revalue. The plan of Harry Dexter White of the US Treasury instead pegged the dollar to gold at $35 an ounce, pegged other currencies to the dollar, and left the burden of adjustment almost entirely with deficit countries.

The outcome was settled by power, not by the merits of the design: the United States then held roughly two-thirds of the world's monetary gold and was the sole net creditor — surplus countries do not vote for rules that penalize surpluses. White's plan won. In 1960 Robert Triffin, in Gold and the Dollar Crisis, named the self-destruct clause built into the system: the world needs dollars as reserves, so the United States must run persistent deficits to supply liquidity, and those very accumulating deficits destroy the credibility of dollar-gold convertibility. On 15 August 1971 Nixon closed the gold window, ending the convertibility pledge.

De-dollarization talk and alternative cross-border payment rails are stuck on the same network effect: for any single country the cost of exiting exceeds the gain unless a critical mass moves together — the same arithmetic as protocol migration in distributed systems.