Artificial Wasteland · the record, corrected

The Myth of Barter

You were taught that money was invented to fix the hassle of swapping chickens for shoes. It's a tidy story, repeated in nearly every economics textbook. The trouble is that no one has ever found the barter economy it describes — and the oldest writing on Earth was invented not for poetry or prayer, but to keep track of who owed whom.

Here is the story almost everyone knows. Once upon a time, before money, people bartered: I have grain, you have an axe, we swap. It was clumsy — what if you wanted my grain but I didn't want your axe? — so, to grease the wheels, humanity invented money. Barter first, then coins, then (much later) credit and debt.

It is a beautiful sequence. It is also, as far as anyone can tell from the evidence, backwards. This page lets you do two things: run the village the textbook imagines and watch why barter is so hard — and then look at what the ground actually shows about the order in which money, writing, and debt appeared.

Where the story comes from

The barter origin of money is genuinely old. Aristotle sketched it around 350 BCE: households grow self-sufficient, then exchange surpluses, and because goods are "not easily carried about," people agree to use something durable — "iron, silver, and the like" — which is later stamped into coin (Politics I.9, 1257a).

But the canonical version is Adam Smith's. In The Wealth of Nations (1776) he roots all commerce in a human "propensity to truck, barter, and exchange one thing for another," and devotes a whole chapter (Book I, Ch. 4, "Of the Origin and Use of Money") to money arising to overcome barter's inconveniences.

A century later the economist William Stanley Jevons gave barter's core problem its enduring name — the double coincidence of wants:

"The first difficulty in barter is to find two persons whose disposable possessions mutually suit each other's wants. There may be many people wanting, and many possessing those things wanted; but to allow of an act of barter, there must be a double coincidence, which will rarely happen." — W. S. Jevons, Money and the Mechanism of Exchange (London: Henry S. King & Co., 1875), Ch. 1, "Barter."

Jevons was right that the double coincidence is a real difficulty. The instrument below lets you feel exactly how rare it is. The catch — which we get to afterwards — is that real communities never had to suffer it, because they had a better tool all along.

Run the village

A village of n people. There are g kinds of goods; each person makes one good and wants a different one (nobody wants what they already make). Drag the sliders, pick a rule, and watch how many people actually get what they want.

The village — barter vs. a credit ledger

100
25

The pattern is brutal and it holds every time you reshuffle. Barter (the orange path) requires a double coincidence: to trade, I must find someone who makes what I want and wants what I make. As the variety of goods grows, the odds of that lucky pairing collapse, and most of the village is left holding things nobody will swap for. Credit (the green path) asks far less: I just need someone, anyone, to make the thing I want. I take it now and owe them — the debt is written down and settles later, around the loop. Almost everyone goes home satisfied.

That is the whole engine of the correction. A community would never grind through pure barter long enough to "invent money to fix it." Long before coins, they would do the obvious thing: keep a tab. And keeping a tab is exactly what the archaeological record shows people doing first.

What the ground actually shows

If money grew out of barter, you'd expect coins to be old and credit to be new. The opposite is true. Here is the order things actually appear in the record — writing and written debt come first by thousands of years; the coin shows up at the very end.

Writing was used only for accounting for centuries before it recorded anything else. The first coin is younger than written debt-law by more than a thousand years — and younger than writing itself by ~2,700.

The three markers, with sources

c. 3300 BCE — the first writing was an account book

The earliest known writing — proto-cuneiform, on clay tablets at Uruk in southern Mesopotamia (Uruk IV–III, c. 3300–3100 BCE) — was not used for stories, scripture, or law. The overwhelming majority of these first tablets are administrative records: quantities of grain, beer, livestock, textiles, and rations disbursed and received by temple and palace institutions. As Denise Schmandt-Besserat puts it, "writing was used exclusively for accounting until the third millennium BC."

Sources: Nissen, Damerow & Englund, Archaic Bookkeeping (Univ. of Chicago Press, 1993); D. Schmandt-Besserat, Before Writing (Univ. of Texas Press, 1992) & utexas.edu/dsb. Honest edge: Egyptian hieroglyphs (~3250 BCE, e.g. the Abydos labels — themselves administrative tags) are nearly contemporary, so "the earliest writing in the world" is debated; the accounting-first point holds for both. The specific token→sign mechanism is Schmandt-Besserat's thesis and not universally accepted; the administrative function of the earliest writing is not in dispute.

c. 1754 BCE — the law of debt, long before the coin

The Laws of Hammurabi (Babylon, reign c. 1792–1750 BCE) already regulate a fully developed credit economy: standard interest rates of 33⅓% on grain loans and 20% on silver; debt relief when a harvest fails ("he washes his debt-tablet in water and pays no rent for this year," Law 48); and a limit on debt-bondage —

"If any one fail to meet a claim for debt, and sell himself, his wife, his son, and daughter for money or give them away to forced labor: they shall work for three years in the house of the man who bought them … and in the fourth year they shall be set free." — Laws of Hammurabi §117 (trans. L. W. King), Avalon Project, Yale Law School.

Value here is reckoned in weighed silver (the shekel, a unit of weight ≈ 8.3–8.4 g — not a coin) and in measures of barley. Debt, interest, and units of account are all in place more than a millennium before anyone strikes a coin.

Sources: Martha T. Roth, Law Collections from Mesopotamia and Asia Minor (2nd ed., 1997) for the interest rates; Avalon Project (King translation) for §48, §117; M. A. Powell, "Money in Mesopotamia," JESHO 39 (1996) on silver/barley as standards of value. Honest edge: dates follow the "middle chronology" and shift by ±~60 yrs under other chronologies; most Assyriologists now call it the "Laws" (a royal monument / model judgments) rather than a modern "Code."

c. 600 BCE — finally, the coin

The first true coins — stamped, standardized lumps of electrum (a gold-silver alloy) — appear in Lydia (western Anatolia, modern Turkey) in the late 7th century BCE. Ninety-three early electrum pieces were found beneath the Temple of Artemis at Ephesus, deposited around 600 BCE. Coined money, in other words, is one of the last monetary inventions, not the first.

Sources: World History Encyclopedia; the British Museum / Ephesus Artemision deposit. Honest edge: the exact date is "one of the most debated topics in ancient numismatics," with estimates spanning ~640–550 BCE. Independently, cast-bronze coinage in China and punch-marked silver in India follow in the 6th–5th centuries BCE. None of this is close to 3300 BCE.

The correction — and exactly how far it goes

The anthropologist Caroline Humphrey stated the case most sharply in 1985:

"No example of a barter economy, pure and simple, has ever been described, let alone the emergence from it of money; all available ethnography suggests that there never has been such a thing." — Caroline Humphrey, "Barter and Economic Disintegration," Man (n.s.) 20:1 (1985), p. 48.

The anthropologist David Graeber built his 2011 book Debt: The First 5,000 Years around this point (its second chapter is literally titled "The Myth of Barter"), arguing that the textbook order — barter, money, credit — is reversed: credit and units of account came first, coins much later. The credit-first view has a long lineage: A. Mitchell-Innes (1913), Marcel Mauss's The Gift (1925), Karl Polanyi (1944), Geoffrey Ingham.

But the honest version of this claim is narrower than the slogan, and it is worth being precise — because a true correction that overstates itself stops being true.

What is actually established — and what is contested

Solid: Barter is real, but it shows up in specific places — between strangers or rival groups, and especially when a money economy breaks down (Humphrey's own 1985 article documents a Himalayan people, the Lhomi, reverting to barter after losing access to cash — barter as disintegration, hence her title). What has never been documented is the textbook scenario: a moneyless community running its everyday internal economy on barter and then inventing money out of it. Writing, credit, and units of account all predate coinage by millennia. Even economist critics of Graeber concede this much.

Contested: Whether that refutes Adam Smith. The monetary economist George Selgin ("The Myth of the Myth of Barter," 2016) argues Smith was theorizing exchange among strangers — exactly where the double-coincidence problem bites — so the anthropology attacks a strawman; and that absence of evidence for pure barter could reflect survivorship bias (such arrangements would vanish fast). A 2025 working paper by Georgy Ganev goes further, arguing Humphrey's "never" is a framing assumption rather than a finding, and noting she herself softens it to barter being "very rare" one page later (p. 49). The double coincidence of wants remains perfectly valid as analysis — as the village above shows — even if it isn't history.

Handle with care: Graeber's book, though influential, contains documented factual errors (a mangled account of Apple's founding; a garbled Nixon/de Gaulle 1971 gold anecdote he later corrected). Treat its big thesis as important and partly vindicated on the narrow barter point — not as settled, and not error-free.

The one case that looks like a counterexample

Economics courses love R. A. Radford's 1945 account of a WWII prisoner-of-war camp, where cigarettes spontaneously became money: "Starting with simple direct barter … cigarettes rose from the status of a normal commodity to that of currency" — they "became the standard of value," and a 100%-food-backed paper note (the "Bully Mark") even appeared. Barter → commodity money → paper money, in one camp. Doesn't that prove the textbook story?

Not for the origin of money. Everyone in that camp had grown up using money; they were re-creating a tool they already knew, in an artificial setting with no production and an omnipresent administering authority (the camp and the Red Cross). As Graeber puts it, such cases "cannot lead to the invention of money because money has already been invented." Radford himself warned a prison camp "is not to be compared with the seething crowd of higglers in a street market." It's a real, uncontested case — of money returning where credit can't run, not of money being born from barter.

R. A. Radford, "The Economic Organisation of a P.O.W. Camp," Economica n.s. 12:48 (Nov. 1945), pp. 189–201, quoted at pp. 190–192.

The check

The village instrument runs the same computation that a standalone verifier independently confirms. For each village: q = 1/(g(g−1)) is the chance a given pair is a double coincidence (Jevons' difficulty, made exact); the fraction who can find any barter partner is 1−(1−q)n−1; the fraction whose wanted good is made by someone (so a credit chain can deliver it) is 1−(1−1/g)n−1.

research/the-myth-of-barter/verify.mjs checks both closed forms against a 4,000-trial seeded Monte-Carlo of the actual village, and confirms q = 1/(g(g−1)) by full enumeration at small g. All 19 checks pass. The page and the verifier share the identical model, so the numbers you see above are the numbers that were proven.

What the model does and doesn't claim. The "barter" number is generous — it counts you as served merely for being able to trade with someone, before accounting for the fact that each good is made in one unit (real clearing would leave even fewer satisfied). Production and wants are independent and uniform — a deliberately simple world. It is not a claim about any specific historical society; it is Jevons' own point, made operable: the double coincidence is rare, and credit routes around it.

Sources