Money Is a 6000-Year-Old Distributed Systems Problem

Money is a means of exchange: you give me something valuable, I give you money in return. But once you start looking at how money has worked across history, it stops looking like a finance story and starts looking like a distributed systems story. Every phase of money solved a specific engineering problem that the previous phase couldn’t — and the problems are ones any software engineer will recognize.

Let’s walk through it, and for each phase let’s be precise about what was actually being trusted, what was being verified, and where the system broke.

Barter — the peer-to-peer discovery problem#

The earliest trade was barter. I have goats, I want silver, so I find someone with silver who wants goats, and we make a deal. Historians trace this back to around 6000 BC in Mesopotamia (though anthropologists increasingly argue that “pure barter economies” are more of a textbook myth than a real historical phase — debt and credit relationships often came first).

Regardless of how pervasive it was, barter as a mechanism has two obvious engineering problems:

No shared protocol for value. How many goats equal how much silver? There’s no standard. Every transaction renegotiates the exchange rate from scratch. It’s like every HTTP client and server inventing a new wire format on each request.
The double coincidence of wants. You need a counterparty who both has what you want and wants what you have. In distributed systems terms, this is a peer-to-peer matching problem with no index and no broker. Discovery is O(n) at best, and often fails entirely.

Barter doesn’t scale. It works for a village of 50 people, not a civilization.

Commodity money — introducing a shared protocol#

The fix: pick something everyone agrees has value, and use that as the medium. Cattle were one of the earliest (the word “pecuniary” comes from Latin pecus, meaning cattle). Cowrie shells were used across Asia and Africa from around 2000–1000 BC — durable, hard to fake, easy to count. Mesopotamia and Egypt traded gold and silver as bars and wire about 4,500 years ago.

This is the first real protocol. Nobody specifically wants your goats — they just need to accept the shells, because they know the next person will too. The double coincidence of wants collapses from “find someone who wants goats” to “find someone who accepts shells,” and almost everyone does.

But there’s still a verification problem: how do you know the silver bar you’re being handed is actually silver, and weighs what it’s supposed to? Every transaction requires a scale and, ideally, a way to assay the metal. Verification cost is high and happens at every step.

Coins — attestation, not backing#

Around 600 BC, the kingdom of Lydia (modern-day Turkey) struck the first official coins. They were electrum — a natural gold/silver alloy — and stamped with a lion’s head. China independently developed coinage around the same period.

This is where the common “coins were backed by the government” framing goes wrong, and the distinction matters: early coins weren’t backed by anything external. They were made of the valuable stuff itself. The silver was the value. If you melted a Roman silver denarius, you got exactly the amount of silver the denarius was worth.

So what was the stamp doing? It was a signature — an attestation. The issuer was saying:

“This coin contains X grams of Y alloy. You don’t have to weigh it. You don’t have to test the metal. Trust the stamp.”

The stamp didn’t create value. It certified value that was already physically present. In engineering terms, coins didn’t introduce backing; they moved verification from per-transaction to per-issuance. Every medieval merchant no longer needed a scale and an assayer. They could look at the stamp and transact in seconds.

This is the same pattern as TLS certificates, signed Git commits, or any attestation system: a trusted party signs something once, and everyone downstream gets to skip verification.

Debasement — the first proto-fiat#

The interesting failure mode of coinage is that the issuer can cheat. And they did, constantly.

The Roman denarius is the canonical case study. Augustus (reigning at the end of the 1st century BC) set it at around 95–98% silver, about 3.9 grams. It held that purity for over a century. Then:

Nero (~64 AD): dropped weight and silver content.
By 200 AD: under 60% silver.
By 270 AD: around 5% silver, with a thin silver wash over a bronze core.
By ~300 AD: the denarius was effectively gone.

The face value stayed nominally the same. The metal content collapsed. This is called debasement, and it’s essentially the issuer exploiting the fact that users trust the stamp instead of the metal. The Roman state was running a multi-century operation to extract value from its own money supply — pay soldiers and fund wars with coins that carried the authority of the old denarius but only a fraction of the silver.

The consequence is Gresham’s Law: bad money drives out good money. When new low-silver coins circulate alongside old high-silver coins at the same face value, everyone spends the bad ones and hoards the good ones. The good coins vanish from circulation. The denarius disappeared from Roman commerce not because it was banned, but because any sane person melted theirs down or buried them the moment a debased version existed.

For an engineer, Gresham’s Law is a game-theoretic result: in a system where two tokens are treated as equivalent by protocol but have different underlying costs, rational actors will always spend the cheaper one and hold the more expensive one. The “good” token self-removes from circulation. It’s the monetary version of “cheapest valid solution wins.”

This was the first real-world experiment with partial fiat: the stamp claims value that the metal no longer fully backs. It worked until it didn’t. Roman inflation in the 3rd century got so bad that the empire eventually stopped pretending and transitioned to barter and in-kind tax collection in many regions — an actual regression to an earlier phase of the stack.

Paper money — indirection and abstraction#

Carrying coins for large transactions is physically painful. A medieval merchant moving a year’s worth of coins across a country needed a cart and a guard.

Paper money — starting in China during the Tang Dynasty (618–907 AD) and made official by the Song Dynasty in the 11th century with a currency called jiaozi — fixed this with classic indirection. The paper note was a pointer. The actual value was metal sitting in a vault, and the note was a redeemable claim on it. Europe caught up on this about 600 years later.

This is exactly the trick that makes modern software tractable: the thing you hold doesn’t have to be the thing, it just has to reliably refer to it. File handles aren’t files. URLs aren’t web pages. The note isn’t the gold.

Tally sticks — a medieval distributed ledger#

From around 1100 AD, medieval England used tally sticks as an accounting and currency system. They’re interesting because they’re an almost perfect early example of a distributed ledger.

A tally stick was a piece of hazelwood notched to record a debt — notch size indicated amount (thumb-width for £100, little finger for £20, and so on). After notching, the stick was split lengthwise so both halves carried an identical record. The creditor kept the longer half (the “stock” — the origin of “stockholder”), the debtor kept the shorter half (the “foil”). The crucial property: the irregular grain of the split meant only the original two halves fit back together. You couldn’t forge a matching tally after the fact — the wood itself served as a tamper-evident, pair-bound record.

Think about what that gives you:

Immutable record: you can’t re-notch without it being obvious.
Pair-bound verification: proof of validity requires both halves, which are held by opposing parties.
No central authority needed for verification of an individual transaction. If the halves match, the record is real.

England ran large portions of its royal accounting on tally sticks for about 700 years. It’s a pre-electronic distributed ledger, using the physics of wood grain as its cryptography.

Banking — the first financial platforms#

By now we have coins, paper claims, and debt records. What we don’t yet have is an institution that does the unglamorous work of holding deposits, issuing credit, moving money across distances, and keeping books that outlast any single transaction. That’s banking, and it’s where money starts to behave less like an object and more like a platform.

The Medici — Italy invents the modern bank#

Banking in a recognizable form emerges in 14th–15th century Italy — Florence, Venice, Genoa. Wealthy grain merchants started acting as lenders and currency exchangers on the side, and eventually the side business became the business. The Bardi and Peruzzi families ran the first large multi-branch banks in the early 1300s. Both collapsed in the 1340s when Edward III of England defaulted on enormous war loans — the first major systemic banking failure caused by sovereign default, a pattern that would repeat a lot.

The Medici Bank (founded 1397 by Giovanni di Bicci de’ Medici) was the successor and the most influential. Its real contribution wasn’t size — it was infrastructure. Four innovations in particular are load-bearing for everything that came after:

Double-entry bookkeeping. Every transaction recorded twice: once as a debit, once as a credit, on opposite sides of the ledger. The books only balance if the two sides agree. This is a checksum. It doesn’t prevent fraud entirely, but it makes unintentional errors self-announcing and makes intentional fraud much harder to hide. In engineering terms, double-entry is a ledger with a built-in integrity constraint — every write is paired, and imbalance is a compile-time error. Luca Pacioli formalized it in 1494, but the Medici had already been running it for most of a century.
Branch network. The Medici ran offices across Florence, Venice, Milan, Rome, Avignon, Geneva (later Lyon), Bruges, and London. A merchant in Bruges could deposit with the local Medici branch and a trading partner in Florence could draw from the Florence branch against that deposit. This is cross-node accounting with eventual settlement between branches. They were running a distributed system with 15th-century latency — weeks to months — and they mostly kept it consistent.
Bills of exchange. A merchant in Florence could write a paper instrument in florins, payable in pounds at a future date in London. This is simultaneously (a) a proto-wire transfer, (b) a foreign-exchange trade, and (c) short-term credit. Crucially it was also a workaround for the usury ban — more on that in the next section. Bills of exchange are the ancestor of every modern instrument from letters of credit to swap contracts.
Holding-company structure. The Medici Bank was legally a set of separate partnerships — one per branch — owned by a central partnership controlled by the family. If one branch blew up, the others were legally insulated. This is sharding and blast-radius containment, 500 years before the word “microservice” existed.

The Medici Bank eventually collapsed too, in 1494, mostly from bad loans to European princes and poor oversight of distant branches. The risk model that made it powerful — lending to sovereigns — is also what killed it. But the patterns survived.

Goldsmiths — accidental invention of fractional reserve banking#

Two centuries later, in 17th-century London, banking bootstrapped itself again, this time through goldsmiths. Wealthy people stored their gold in goldsmiths’ vaults (they had the strongest vaults in town) and got paper receipts saying “bearer is entitled to X ounces of gold.” These receipts started getting traded as money directly, because it was easier than redeeming them.

Then the goldsmiths noticed something: not everyone redeems at once. On any given day, only a small fraction of receipt-holders actually came to claim their gold. Which meant… they could issue more receipts than they had gold in the vault, and as long as redemptions stayed below the cash reserve, nobody would notice.

This is fractional reserve banking, and in engineering terms it’s deliberate overcommitment of a resource. The database has 100 rows claiming ownership of an asset, but the underlying asset only has 60 physical units. It works as long as reads stay distributed over time. It breaks catastrophically if everyone reads at once — which is exactly what a bank run is. A coordinated read that exceeds the reserve.

The whole modern banking system is still built on this model. Your “deposit” at a bank isn’t sitting in a vault; it’s been loaned out. The bank holds a small fraction (varies by regulation, often around 10%, sometimes effectively 0% after the 2020 US reserve requirement change) as liquid cash. The rest is IOUs from other parties. Your account balance is a claim that is only honorable if not too many claims are exercised simultaneously.

The Bank of England — banking becomes state infrastructure#

In 1694, a group of London merchants lent the English crown £1.2 million to fund a war with France, and in exchange got a charter to form the Bank of England — the first modern central bank. It was nominally a private institution issuing banknotes, but it had a special relationship with the state: its notes were accepted for taxes, and the government’s debt was parked there. In effect, the bank’s promises became indistinguishable from the state’s.

This is the pattern for every central bank that followed. A central bank is where the government’s credit and the banking system’s credit are deliberately entangled — so that when a bank run threatens the system, the central bank can print new money to cover redemptions (lender of last resort), and when the government needs money, the central bank can buy its debt. The trade-off: you get systemic stability, you lose a hard constraint on money supply.

Engineering analogy: the central bank is the system’s garbage collector and memory allocator combined. It can always find more memory (print money) to prevent an out-of-memory crash (bank run, debt default). That flexibility is powerful, but it also means there’s no hard ceiling on allocation — which is why inflation is a feature of every fiat system, not a bug.

The usury problem — why interest was banned for most of history#

Every system of banking runs on lending at interest. But for most of recorded history, lending at interest was considered a serious moral crime, and all three major Abrahamic religions explicitly banned it.

The bans#

Judaism prohibits ribbit (interest) on loans between Jews (Exodus 22:25, Leviticus 25:36–37, Deuteronomy 23:19). But Deuteronomy 23:20 carves out loans to non-Jews — the so-called “Deuteronomic double standard.” This has enormous historical consequences.
Christianity inherited the Jewish prohibition and extended it universally. The Council of Nicaea (325 AD) banned clergy from lending at interest; later councils extended the ban to all Christians. By 800 AD, Charlemagne made it secular law across the Frankish empire. The Third Lateran Council (1179) excommunicated usurers.
Islam prohibits riba, stated explicitly in the Quran (2:275–279) as one of the clearest prohibitions in the text. Unlike the Christian prohibition, which eroded over centuries, the Islamic prohibition remains actively enforced — Islamic finance is a multi-trillion-dollar industry built around structures that avoid it.

Why ban it?#

The arguments, across all three traditions, converge on a few points:

Money is sterile. Aristotle argued that money is a tool for exchange; it cannot reproduce itself. Making money from money is “unnatural” because nothing has been produced. This argument was picked up wholesale by Christian theologians like Thomas Aquinas. The engineering version: charging interest treats a number as if it has intrinsic yield, when actually all yield comes from work done in the real economy. Interest is a claim on work you didn’t do.
Exploitation of the vulnerable. In agrarian economies, people borrowed out of desperation — failed harvest, illness, dowry, debt to a bigger creditor. Charging interest on a loan to someone in crisis is a mechanism for transferring land and labor from the desperate to the wealthy. Debt-driven dispossession was common enough that multiple ancient societies had periodic debt jubilees to reset it (Babylonian mīšarum, the biblical jubilee year).
Compounding is unbounded. An interest rate is an exponential. Left unchecked, a small loan becomes an infinite claim. Pre-industrial societies understood this — they’d seen it happen — and considered it a form of enslavement in slow motion.

The engineering framing: interest is unilateral risk transfer. The lender collects regardless of what happens in the real economy. If the borrower’s venture succeeds, the lender gets interest. If it fails, the lender still gets interest (and the collateral). That one-way flow is what the religious bans were trying to prevent.

Workarounds, not principles#

In practice, every society that banned interest found ways around it, because credit is too economically useful to live without. The workarounds are where it gets interesting.

Jewish moneylenders in Christian Europe. This one deserves unpacking because it’s widely misunderstood. The common framing is that Jews “dominated” banking, often with an implication of privilege. The reality is the inverse: in most of medieval Christian Europe, Jews were forbidden from owning land and banned from Christian craft guilds, which closed off farming and most skilled trades. Combined with the Deuteronomic carve-out that let Jews lend at interest to non-Jews, and the Christian ban that forbade Christians from lending at interest at all, this funneled Jewish communities into moneylending as one of the only professions left available. Louis IX of France captured the absurdity in 1254: he outlawed Jewish moneylending and told them to “live from the work of their hands or from trade” — while his own laws forbade them from owning land or joining guilds. It was a closed loop. Christian rulers tolerated (and often coerced) Jewish communities into playing the lender role, then periodically expelled or expropriated them when debts piled up. The Shylock caricature is essentially an anti-Semitic projection of a system Christians themselves designed.
Two later developments are worth noting, because Jewish communities did end up influential in the evolution of international banking, but through a different mechanism than the stereotype implies:
- Court Jews (Hofjuden), 16th–18th centuries: Jewish financiers like Samuel Oppenheimer and Jacob Bassevi handled the finances of Habsburg and German rulers. Their cross-border family networks — an artifact of the segregated community structure forced on them — turned out to be a genuine competitive advantage for international finance, since Christian institutions were fragmented along national lines.
- The Rothschilds (19th century): Mayer Amschel Rothschild started as a court factor in Frankfurt in the 1760s, then sent his five sons to establish coordinated banking houses in Frankfurt, London, Paris, Vienna, and Naples. Arguably the first truly international private bank, using the same “branches across jurisdictions” pattern the Medici had pioneered 400 years earlier — but at a continental scale.
Note, though, that banking as an institutional form — double-entry bookkeeping, the multi-branch firm, the central bank — was mostly a Christian Italian and English development (Medici, Bank of England). Jewish communities dominated moneylending first, and later became disproportionately represented in international finance, but the bank-as-institution was a parallel lineage.
Bills of exchange. Recall the Medici’s bill of exchange — florins in Florence, pounds in London, payable in three months. The exchange rate could be quietly set so that the lender made a profit. This is interest, laundered through a currency conversion. The Church grudgingly tolerated it because it was technically “exchange,” not “lending.” This is the fig-leaf pattern: the protocol is unchanged, only the label is different.
Contractum trinius. An Italian innovation: three separate legal contracts, each individually permitted — an investment, a sale of future profits, and an insurance contract — which, combined, produce the economic effect of a fixed-interest loan. The Church banned this in 1586 once it figured out the trick. The pattern recurs: whenever a system forbids a primitive operation, users will compose it out of three permitted primitives.
Islamic finance: genuinely different structures. The Islamic tradition took the ban more seriously and built actual alternatives. Murabaha (cost-plus sale): the bank buys the asset and sells it to you at a markup, payable in installments. No interest, but economically similar to a loan. Mudarabah (profit-sharing): the bank provides capital, you provide labor, you split profits at an agreed ratio — and critically, if the venture loses money, the bank takes the loss on capital, you take the loss on your time. This is the thing the interest ban was really trying to preserve: shared risk. The lender’s return depends on whether the real economy produced something. It’s equity, not debt.

The engineering version of the Islamic distinction is clean: interest is a fixed return independent of outcome (debt), profit-sharing is a variable return that aligns with outcome (equity). DeFi protocols have independently recreated the equity-style primitive in the form of liquidity pools and staking — you put capital in, your return varies with the performance of the pool. Islamic finance arrived at the same pattern 1,400 years earlier for purely ethical reasons.

Christianity, meanwhile, gradually abandoned the ban. Calvin argued in the 16th century that interest on productive business loans was acceptable, only interest on loans to the desperate was sinful. By the 17th century the distinction had collapsed and interest became simply respectable finance. This ideological shift is what let the Bank of England exist in 1694 as a Christian institution.

Fiat — dropping the backing, keeping the protocol#

Through the 20th century, governments progressively decoupled paper money from gold entirely. Key milestones:

1914–1918: most European countries suspended the gold standard to print money for WWI.
1944 (Bretton Woods): the US dollar becomes the anchor of the global system, pegged to gold at $35/oz. Other currencies peg to the dollar. The system is gold-backed but only indirectly for most countries.
1971 (the “Nixon Shock”): the US suspends dollar-to-gold convertibility. The dollar becomes pure fiat. Every other currency, by extension, does too.

”Fiat” money has no commodity backing. The note used to be a pointer to gold; now it’s a pointer to… nothing physical. Its value comes from two things: forced demand (the government accepts it — only it — for taxes) and Schelling-point trust (everyone accepts it because everyone else does, and nobody wants to be the one holding a currency no one takes).

From an engineering angle, fiat is the moment money becomes pure protocol. There’s no underlying asset. The system is held together by coordinated belief and institutional enforcement. It works because enough participants run the same protocol at the same time. Break the coordination (hyperinflation, loss of confidence in the issuer) and the whole thing unwinds fast, because there’s no physical anchor to land on. Weimar Germany, Zimbabwe, Venezuela are case studies.

This is also what lets central banks manage the money supply in a way that was impossible under a gold standard. Quantitative easing, interest rate targeting, stimulus — these tools exist because money is now software-defined, and software can be rewritten.

Reserve currency — the global settlement layer#

Every country has its own fiat currency, but they’re not equal. For international trade, commodity pricing, and cross-border debt, the world effectively uses a reserve currency — one currency that everyone else holds in large quantities and that most global transactions are denominated in. Today that’s the US dollar. In 2024, the USD made up about 58% of disclosed global central bank reserves, vs the euro at ~20%, yen at ~6%, pound at ~5%, and the Chinese renminbi at ~2%.

Why does anyone need a “reserve”?#

A central bank holds reserves in foreign currencies for the same reasons a company holds cash: to pay bills that come due in those currencies, to stabilize its own exchange rate, and to weather shocks. If your country imports oil, and oil is priced in dollars, you need dollars — even if your domestic currency is rupees or yen or reais. Multiply that across every importer, every cross-border loan, every commodity trade, every international contract, and one currency inevitably accumulates gravitational pull.

Reserve currency status is not something a country declares — it’s an emergent property of network effects. Engineers will recognize the pattern: it’s the same reason TCP/IP won over OSI, English won as the lingua franca of science, or JavaScript ended up running everywhere. The marginal participant adopts the incumbent because everyone else already did. Switching costs are quadratic.

The historical succession#

Reserve currency dominance has shifted roughly once a century:

Portuguese real (1450–1530)
Spanish silver dollar / pieces of eight (1530–1640)
Dutch guilder (1640–1720)
French livre / franc (1720–1815)
British pound sterling (1815–1945)
US dollar (1945–present)

Each transition was tied to a shift in economic, military, or naval dominance. The pound gave way to the dollar because Britain financed WWI and WWII by liquidating its overseas assets and borrowing heavily from the US. By 1944 the US had most of the world’s monetary gold, and Bretton Woods ratified what was already true: the dollar had become the center.

Why the dollar stays on top (for now)#

Four reinforcing layers, all with network-effect dynamics:

Trade invoicing. About half of world trade is invoiced in USD, including almost all oil and most commodities. Exporters want dollars because importers pay in dollars.
Debt markets. Over half of international debt is USD-denominated. Borrowers need dollars to service it; lenders want dollars because that’s what they’re owed.
Reserves. Central banks hold dollars because they need them to intervene in FX markets and service dollar debt.
Safe asset. US Treasuries are the deepest, most liquid bond market in the world. In a crisis, everyone runs to them, which only reinforces their status.

Each layer creates demand for the next. Breaking any single one is hard. Breaking all four at once is the only way a real transition happens.

The “exorbitant privilege”#

French finance minister Valéry Giscard d’Estaing coined this phrase in the 1960s, and it’s the honest name for what reserve status gives the issuer. Because everyone else needs dollars, the US can:

Run persistent trade deficits without the currency collapsing.
Borrow cheaply in its own currency (no exchange-rate risk on its debt).
Print money with less inflationary consequence than any other country, because foreigners absorb a lot of the new supply.
Enforce sanctions globally — anyone cut off from the dollar system is effectively cut off from global commerce.

The last one has consequences. When the US and EU froze Russia’s dollar and euro reserves in 2022 following the Ukraine invasion, it demonstrated to every other country that reserves held in dollars aren’t fully theirs — they’re dollars on permission. That’s the first time in the post-WWII era that the “safety” of dollar reserves was visibly questioned by major non-allied central banks. Since then:

Central bank gold purchases have hit multi-decade highs.
BRICS countries have publicly explored non-dollar settlement arrangements.
China’s CIPS (its alternative to SWIFT) has grown, though still small.
The mBridge project (central bank digital currencies from China, Hong Kong, Thailand, UAE) quietly demonstrated multi-CBDC cross-border settlement without dollar intermediation.

None of these have replaced the dollar. All of them represent hedging. The USD share of reserves has slipped from its 2001 peak of ~72% to ~58% today — not a collapse, but a visible erosion.

Engineering framing: a reserve currency is the default protocol of international finance. Like any dominant protocol, it persists because of network effects, not because it’s technically the best option. And like any dominant protocol, it usually only gets replaced when the incumbent is perceived to have abused its privileged position — which is where we currently are, in slow motion.

Digital money — centralized ledgers at scale#

Most money today is already digital. It’s a number in a bank’s database. Physical cash is a tiny minority of the M2 money supply in most developed economies — single-digit percentages.

When you “send” someone $100, no physical thing moves. Your bank decrements a row, their bank increments a row, and a settlement system (Fedwire in the US, SEPA in Europe, SWIFT for cross-border messaging, CHIPS for large-value USD) reconciles the interbank position later, usually at end of day or across multi-day cycles. Retail speed is an illusion — the user sees instant credit because the bank fronts the money, not because settlement actually happened.

This introduces the double-spend problem in a form engineers immediately recognize. If money is just a number in a database, what stops someone from spending the same number twice? In a centralized system the answer is straightforward: the bank is the source of truth. Every transaction goes through a single authority that maintains one canonical ledger and rejects conflicting writes. It’s basically a database with strong consistency, single-writer authorization, and eventual inter-bank consistency bolted on via settlement networks.

This works, but the trust surface is enormous:

The bank can freeze your account.
The bank can reverse transactions.
The bank can fail (SVB, 2023).
The bank can be coerced by governments.
The bank’s database can be hacked.
The inter-bank settlement layer (SWIFT) can be used as a geopolitical weapon (see: Russia sanctions, 2022).

Every one of these is a “centralization cost” that users pay for the convenience of not carrying metal.

Cryptocurrency — distributed consensus without a trusted issuer#

Bitcoin (2009) asked a question that’s a pure distributed systems question: can you solve the double-spend problem without a central authority?

The answer turns out to be yes, using three pieces of machinery that engineers work with all the time:

Public-key cryptography provides ownership. Your coins are locked to a public key; only the matching private key can spend them. No issuer needs to vouch for you. Your key is your claim.
A public append-only ledger (the blockchain) gives everyone the same view of who owns what. Instead of one bank’s database, there are thousands of copies of the same ledger, all eventually consistent.
Consensus (proof-of-work, proof-of-stake, BFT variants) makes all those copies agree on the order of transactions, even when some participants are malicious or offline. This is the Byzantine Generals Problem, solved at scale. Proof-of-work makes rewriting history economically infeasible by requiring the attacker to redo compute-heavy work faster than the honest network. Proof-of-stake makes it economically infeasible by slashing the deposits of validators who sign conflicting blocks. Different cryptoeconomic trade-offs, same underlying goal.

Scarcity — the thing that makes a currency a currency — is enforced by the protocol itself. Bitcoin’s supply cap of 21 million isn’t a policy, it’s a line of code running on every node. For the first time in 6,000 years, the rules of the money are the software, not a contract you hope an institution honors.

The tradeoff: consensus is slow and expensive compared to a centralized database. Bitcoin settles a few transactions per second. Visa does tens of thousands. Layer-2 systems (Lightning, rollups) are attempts to close this gap by pushing most activity off-chain while keeping the base layer as the settlement root of trust — which, notably, is exactly the same architecture as pre-modern banking, where goldsmith receipts circulated freely and only settled against the vault occasionally. The pattern recurs.

Whether crypto is the next major phase of money or a detour is still being argued. But the engineering shift is real: money went from “trust the issuer” to “verify the protocol.”

The pattern#

Zoom out and every phase of money is solving an engineering problem the previous phase couldn’t handle at scale:

Phase	Problem solved	What’s trusted	Failure mode
Barter	—	Nothing — direct exchange	Doesn’t scale past small groups
Commodity money	Discovery, standardization	The commodity itself	Heavy, hard to verify
Coins	Per-transaction verification cost	The issuer’s stamp (+ metal)	Debasement; Gresham’s Law
Tally sticks	Tamper-evident debt records	Physical pair-matching	Doesn’t scale across strangers
Paper money	Physical weight, transport	Vault + issuer promise	Over-issuance
Banks (Medici, etc.)	Credit, cross-border payments	Institution + bookkeeping	Sovereign default, bad loans
Fractional banking	Capital efficiency	Statistical reserve adequacy	Bank runs
Central banks	Systemic stability	Government + lender-of-last-resort	Inflation, monetary capture
Fiat	Rigid supply tied to metal	Government + institutional trust	Hyperinflation
Reserve currency	Cross-border settlement	Dominant issuer’s stability	Weaponization, erosion
Digital (bank) money	Physical cash handling	Bank + settlement networks	Freezes, hacks, censorship
Cryptocurrency	Single point of trust	Protocol + cryptography	Throughput, key loss, 51% attacks

Six thousand years of humans trying to build a reliable, low-friction, trustworthy protocol for moving value between strangers. The tools keep changing. The problem is the same.

And every generation of money carries the scars of the previous one’s failure mode. Coins exist because commodity money didn’t scale. Paper exists because coins were heavy. Banks exist because paper needed institutions. Fiat exists because gold was rigid. Crypto exists because centralized databases are a single point of trust. Whatever comes next will exist because the current system is failing at something we haven’t fully named yet.

Sources:

General history

Coins and debasement

Medieval and early modern

Tally stick — Wikipedia

Banking history

Usury and religious prohibitions

Jewish moneylending, Court Jews, and Rothschilds

Reserve currency

Cryptocurrency

Bitcoin: A Peer-to-Peer Electronic Cash System — Nakamoto, 2008