Paying a Stranger

A payment is a strange thing. When you tap a phone at a tea stall in Bengaluru, or a card at a café in Lisbon, value moves from you to a stranger in a second or two, and neither of you thinks about how. Yet each payment depends on a chain of institutions: a bank that holds your money, a network that carries the message, a system that settles the debt between banks, laws that say who bears the loss if something goes wrong, and a shared belief that the numbers in an account are as good as cash. Remove any link and the payment fails. The miracle of modern payments is that the chain has become invisible.

The money involved is vast. Businesses using Stripe alone processed $1.9 trillion in 2025, by the company’s count about 1.6 percent of the world’s economic output.¹ In August 2026 India’s Unified Payments Interface, UPI, handled 24.51 billion transactions in a single month, about 790 million a day.² And every one of those payments raises the questions this essay is about. What makes a stranger trust a promise to pay? Who should be charged for the convenience of paying without cash? Who owns the rails? And what happens to privacy, debt and the right to undo a mistake when paying becomes effortless?

To answer them, we need to begin with what money is.

Money before coins

The usual story says that people first bartered goods, found barter inconvenient and invented money, first as precious metal and then as coins and paper. Anthropologists have long doubted that story. In 2011 the anthropologist David Graeber, in Debt: The First 5,000 Years, argued that there is no evidence of an economy based on barter between neighbours, and that credit, records of who owed what to whom, came long before coins. In Mesopotamia from around 3000 BCE, temple and palace administrators recorded obligations on clay tablets, measured in units of silver or barley, while most of those obligations were never settled in actual silver.³ Coins, struck in Lydia around 600 BCE, came much later, and often in connection with paying soldiers.

In 1913 the British diplomat and economist Alfred Mitchell-Innes argued in a pair of articles that money is not a commodity at all but a form of credit: a transferable claim, a promise that someone will accept it in payment. On this view the history of payments is the history of making promises transferable among people who do not know each other. A merchant in medieval Genoa could accept a bill of exchange from a stranger in Bruges because a network of banking houses stood behind it. The Bank of England, founded in 1694, issued notes that were literally promises to pay the bearer, and they became money because people trusted the bank to honour them. Cheques, clearing houses, where banks offset their customers’ payments to one another, and wire transfers by telegraph extended the reach of such promises across countries and oceans.

Each extension of the promise required an institution that both sides trusted more than they trusted each other. That is the pattern the rest of this history repeats.

The charge card: credit at a stranger’s table

The modern card began with a dinner bill. In 1950 Diners Club, founded in New York by Frank McNamara and Ralph Schneider, issued a charge card that members could present at participating restaurants, which billed Diners Club; members paid Diners Club monthly, and the restaurants paid a fee on each bill. The famous story that McNamara thought of it after forgetting his wallet at dinner was later described by an early Diners Club executive as a publicity invention. Whatever its origin, the card solved a problem of trust: a restaurant could serve a stranger on credit because a known company guaranteed the bill. American Express launched its own card in 1958.

Banks followed. In September 1958 Bank of America mailed 60,000 BankAmericard credit cards to residents of Fresno, California, most of whom had not asked for them, in what the industry called a “drop.”⁴ The bank faced a problem that economists would later formalise: merchants would accept the card only if many customers carried it, and customers would carry it only if many merchants accepted it. Mailing the cards to a whole city at once was a way to create both sides of the market overnight. It worked, in the sense that the card spread across California. It also produced heavy losses from fraud and unpaid debts, and mass mailings by many banks in the 1960s caused enough public anger that in 1970 the United States banned unsolicited credit cards and limited a cardholder’s liability for unauthorised use to $50.⁵

The second great invention of the card industry was organisational. Bank of America licensed BankAmericard to other banks, but by the late 1960s the system was chaotic: banks disputed each other’s transactions, fraud was rampant, and clearing took weeks. Dee Hock, a manager at a licensee bank in Seattle, persuaded the banks to form an independent company, National BankAmericard Inc., owned by its members, which would set common rules and run the network while the banks competed for customers. It was renamed Visa in 1976.⁶ Hock later described the structure as neither centralised nor anarchic, a network in which competitors cooperated on the rails and competed on everything else. A rival association of banks, the Interbank Card Association, formed in 1966, became Mastercard.

The card networks also settled a question that shapes payments to this day: who pays? When a customer pays by card, the merchant’s bank pays a fee to the customer’s bank, the interchange fee, set by the network. The merchant pays its bank a larger fee that covers it. The customer usually pays nothing directly, and often receives rewards. Network rules for many years prevented merchants from charging card users more than cash payers. The cost of card payments was therefore built into the prices everyone paid, including people who paid in cash. In 2010 economists at the Federal Reserve Bank of Boston estimated that in the United States the average cash-paying household paid about $149 a year to card users through higher prices, and the average card-using household received about $1,133 a year from cash users, a transfer that ran, on average, from poorer to richer households.⁷

Regulators eventually intervened. In 2011 the US Federal Reserve, implementing the Durbin Amendment, capped the debit card interchange fees of large banks at 21 cents plus 0.05 percent of the transaction. In 2015 the European Union capped interchange at 0.2 percent for debit cards and 0.3 percent for credit cards.⁸ In September 2024 the US Justice Department sued Visa, alleging that it had illegally monopolised debit payments, with more than 60 percent of US debit transactions running on its network; Visa denied the allegations.⁹

Cards also brought machines. In June 1967 Barclays opened the first cash machine in Enfield, north London. Through the 1970s and 1980s, electronic networks connected card terminals, cash machines and banks, and payments became messages: authorisation requests and responses travelling between computers in seconds, followed by settlement between banks later.

Paying online: the problem of the stranger returns

The internet reopened the oldest problem in payments. A buyer and seller who had never met, in different cities or countries, wanted to exchange money for goods neither could inspect.

The first serious answer was cryptographic. In 1982 the computer scientist David Chaum described blind signatures, a technique that let a bank sign a digital coin without seeing it, so that the coin could later be spent without revealing who had withdrawn it. In 1985 he published an article titled “Security without Identification: Transaction Systems to Make Big Brother Obsolete,” warning that electronic payments could create a comprehensive record of people’s lives, and proposing systems in which payments would be verifiable but anonymous.¹⁰ His company, DigiCash, founded in Amsterdam around 1990, launched an anonymous electronic cash called ecash in the mid-1990s. Few banks adopted it, few merchants accepted it, and DigiCash went bankrupt in 1998. The more practical answer turned out to be the credit card, typed into a web form and protected by encryption, with the card networks’ rules for disputes and chargebacks standing behind it.

That still left people who could not or would not take cards. PayPal grew from two start-ups, Confinity, founded by Max Levchin and Peter Thiel, which wanted to let people beam money between handheld computers, and X.com, an online bank founded by Elon Musk. They merged in March 2000.¹¹ The handheld idea went nowhere. What took off was payment by email between individuals, especially buyers and sellers on eBay, who adopted PayPal on their own because it let anyone accept money without a merchant account. PayPal followed its users.

It nearly died of fraud. A system that let anyone send money to anyone by email attracted organised criminals, and by 2000 PayPal was losing millions of dollars a month. Its response was to treat fraud as an engineering problem. Levchin and his team built software that analysed patterns of transactions to detect fraud, and developed one of the earliest tests asking users to read distorted text that computers could not, to block automated sign-ups. Fraud losses fell, and the company’s fraud models became one of its main advantages. eBay bought PayPal in 2002 for $1.5 billion. The lesson, repeated by every successful payment company since, was that the ability to manage fraud is not a cost of the business. It is the business.

In China, the problem was even more basic: shoppers did not trust sellers to send goods, and sellers did not trust shoppers to pay. In 2004 Alibaba’s Taobao marketplace launched Alipay, which held a buyer’s payment in escrow until the buyer confirmed that the goods had arrived, and only then released it to the seller.¹² The platform became the trusted third party. Alipay grew into one of the world’s largest payment systems, and Tencent’s WeChat Pay, launched in 2013, competed with it; in 2014 WeChat let users send digital red envelopes of money during the Chinese New Year, and hundreds of millions did. In China, payments moved from cash to QR codes on phones, largely bypassing cards.

Money by text message: M-Pesa

In Kenya the problem was different again. Most adults had no bank account, but many had mobile phones. In 2007 the mobile operator Safaricom, part-owned by Vodafone, launched M-Pesa (pesa is Swahili for money), which let customers store value on their phone account and send it to anyone else by text message. The key was a network of agents, ordinary shopkeepers, who converted cash to electronic value and back.¹³ The service had been piloted to let customers of a microfinance institution repay loans. Users had other ideas. They used it above all to send money home, from workers in cities to families in villages, a journey that had previously meant entrusting cash to a bus driver or travelling in person.

M-Pesa spread with astonishing speed and reached most Kenyan households. In 2016 the economists Tavneet Suri and William Jack published a study in Science estimating that access to M-Pesa had lifted about 194,000 households, 2 percent of Kenyan households, out of extreme poverty, with the largest effects among households headed by women. Other researchers have questioned how much of the effect can be attributed to M-Pesa alone, and the debate continues.¹⁴ But M-Pesa showed that payment infrastructure could be built from the bottom up, from phones and corner shops, rather than from bank branches and cards.

Payments as code: Stripe

By 2010, accepting card payments online was possible but painful. A new internet business had to apply for a merchant account with a bank, sign up with a payment gateway, satisfy security standards and integrate complicated software. Patrick and John Collison, two Irish brothers who had already sold one start-up, founded Stripe to make this as simple as adding a few lines of code to a website. Stripe handled the merchant accounts, the card networks, the security standards and later tax, fraud detection and compliance in many countries, behind a programming interface designed for developers.

The decision about whom to sell to mattered as much as the technology. Incumbent payment processors sold to finance departments of established merchants. Stripe sold to programmers, at the moment they were building a product, before anyone in finance had been consulted. Start-ups adopted it by default, and as some of them grew into large companies, Stripe grew with them. In February 2026 the company reported that businesses on Stripe had processed $1.9 trillion in 2025, up 34 percent, and that it served more than five million businesses.¹ Payments, for a large part of the internet economy, had become an abstraction: a function call whose complexity was someone else’s problem.

Phones then absorbed the card itself. Apple Pay, launched in 2014, stored a substitute number, a token, on the phone instead of the card number, so that a merchant never saw the real one, and authorised payments with a fingerprint or a face. Google and Samsung offered similar systems. The card network remained underneath, and so did its fees.

India’s experiment: payments as public infrastructure

India took a different path, and it became the most important payments story of the decade.

In 2008, at the urging of the Reserve Bank of India, Indian banks set up the National Payments Corporation of India (NPCI), a not-for-profit company owned by banks, to run the country’s retail payment systems. It built a domestic card network, RuPay, and an instant transfer service, IMPS, launched in 2010, that let people send money between banks at any hour. Alongside, the government built what came to be called the India Stack: Aadhaar, a biometric identity number issued to over a billion residents from 2010; a national programme from 2014 that opened hundreds of millions of basic bank accounts; and electronic know-your-customer checks that let banks verify identity in seconds.

On top of these, NPCI designed UPI. It was launched as a pilot in April 2016 and went live that August.¹⁵ UPI let anyone with a bank account and a smartphone send money instantly to anyone else, using a simple address such as name@bank or a QR code instead of account numbers. Crucially, it was interoperable and open: any bank could join, and any licensed company, including non-banks such as Google, Walmart’s PhonePe and Paytm, could build an app on top of it. A payment from a PhonePe user to a Google Pay user moved over the same public rail. Competition happened in the apps; the rail was shared.

Then came a shock. On the evening of 8 November 2016 the prime minister announced that ₹500 and ₹1,000 notes, about 86 percent of the currency in circulation by value, would cease to be legal tender at midnight. The government said the aim was to destroy unaccounted wealth, counterfeit notes and the financing of terrorism, and later also emphasised the move towards digital payments. For weeks, cash was scarce across the country. Digital payments rose sharply. In 2018 the Reserve Bank reported that 99.3 percent of the withdrawn notes had been returned to the banking system, which undercut the expectation that large amounts of unaccounted cash would be destroyed.¹⁶ How much demonetisation contributed to UPI’s later growth is still debated; adoption rose after the shock, but UPI’s monthly value remained modest until late 2017, and the steepest growth came years later.

In January 2020 the government made UPI payments free for merchants, setting the merchant discount rate, the fee merchants pay, to zero by law.¹⁷ For small merchants who had resisted card terminals because of fees, a printed QR code that cost nothing to accept was transformative. Street vendors, auto-rickshaw drivers and village shops adopted UPI in vast numbers. The pandemic accelerated the shift.

The results are unlike anything else in payments. ACI Worldwide estimated that India accounted for 49 percent of all real-time payment transactions in the world in 2023, about 129 billion.¹⁸ By 2025 UPI was handling around twenty billion transactions a month, and in August 2026 it set a record of 24.51 billion.² India has exported the model, linking UPI with payment systems in other countries.

UPI’s design also created problems. Because the rail was free, banks and app providers earned little from payments directly, and the costs of running a system of this scale, and of securing it against fraud, had to be met from elsewhere: government subsidies, cross-selling of loans and insurance, and the providers’ own budgets. Two private apps, PhonePe and Google Pay, came to handle most UPI transactions, nearly four in five as late as May 2026, despite a rule NPCI adopted to cap any app’s share at 30 percent, whose deadline has been extended repeatedly, most recently to the end of 2026.¹⁹ In September 2026 NPCI announced a merchant fee of 0.4 percent on larger merchant payments, above ₹2,000, with lower flat fees for essential sectors, which it estimated would apply to only about 4 percent of merchant transactions, and in October it deferred the fee until after the festive season.²⁰ The question of who pays for a free public rail had come due.

Others followed India’s model. Brazil’s central bank launched Pix in November 2020, an instant payment system it both regulates and operates. It handled nearly 80 billion transactions in 2025 and is used by most Brazilian adults. In July 2025 the United States opened a trade investigation into Brazil that included Pix among its concerns, arguing that a central bank acting as both regulator and operator disadvantaged American payment companies; Brazil’s government described Pix as a public good and a matter of sovereignty.²¹ The US launched its own instant payment service, FedNow, in 2023, though adoption has been slow.

Money without banks, and money from platforms

A different idea of digital money, money without trusted institutions at all, arrived in 2008 with Bitcoin, which used cryptography and a public ledger maintained by anonymous participants to make payments final without a bank. Bitcoin became an asset more than a means of everyday payment, but its descendants, stablecoins pegged to the US dollar and backed by reserves, became a significant way to move dollars across borders. In 2019 Facebook announced Libra, a global stablecoin backed by a consortium of companies; facing opposition from regulators and central banks worried about a private company issuing money to billions of people, the project was renamed Diem and abandoned in 2022. In July 2025 the United States passed the GENIUS Act, establishing a federal framework for regulating dollar stablecoins.²² Stripe’s annual letter put stablecoin payment volume in 2025 at about $400 billion, double the year before.

The cost of speed: fraud and finality

Instant payments brought a new kind of fraud. Card payments could usually be reversed through a chargeback if a merchant failed to deliver or a card was stolen. Instant bank transfers were final. Fraudsters learned to persuade people to send money themselves, by pretending to be their bank, the police, a romantic partner or a supplier. These authorised push payment scams grew rapidly in countries with fast payment systems. In the United Kingdom, regulators required payment firms, from October 2024, to reimburse most victims of such scams up to £85,000, splitting the cost between the sending and receiving firms. In the first year, firms reimbursed 88 percent of the money lost in eligible claims, and an independent evaluation in 2026 found that losses within the scheme had fallen by about a fifth, though some fraud appeared to have moved to channels the rules did not cover.²³ In India, as UPI grew, so did fraud using fake payment requests and impersonation, and authorities and banks expanded warnings, limits and verification.

The tension is fundamental. The faster and more final a payment is, the more useful it is to honest users, and the more useful it is to thieves.

What payments are today

Today payments are mostly invisible. In many countries a card, a phone or a QR code has replaced cash for most purchases. Behind the screen sit very different kinds of systems: private card networks funded by fees on merchants; platforms such as Alipay and PayPal that hold money themselves; developer-facing companies such as Stripe that wrap the old systems in new interfaces; public, instant rails such as UPI and Pix run by central banks or bank-owned bodies; and a growing layer of stablecoins outside the banking system. The common thread is that a payment is now a message between computers, verified in seconds, recorded permanently and settled behind the scenes by institutions most users never see.

How digital payments changed the way we live and decide

Several shifts follow from the move from cash to digital payments.

From the purse to the account

Cash is held. Digital money is a balance in an account kept by an institution. Paying digitally means trusting the institution, and the institution can freeze, delay or reverse the balance. For most people most of the time, that trust is warranted. When it fails, through an outage, a frozen account or a bank failure, the difference becomes vivid.

From visible cost to hidden cost

Cash has no fee at the till, and its costs, handling, security and the risk of theft, are borne visibly by those who hold it. Card fees are hidden in prices. Free instant payments hide their costs in subsidies, data and cross-selling. In each case someone pays, and the person choosing how to pay rarely sees the bill.

From the local to the anonymous stranger

Payments now flow constantly between strangers across the world. That has expanded trade enormously, letting a small seller in one country find customers in many others. It has also made fraud a global industry, and made the institutions that judge which payments to trust far more powerful.

From the moment to the record

Every digital payment leaves a record. Taken together, those records describe where people go, what they buy, whom they pay and what they believe. They are valuable to lenders, marketers, tax authorities and governments, and they make a cashless society also a recorded one.

From the slow to the irreversible

Older payment systems were slow, and slowness gave time to catch errors and fraud. Instant payments are final in seconds. The burden of caution has shifted from the system to the person pressing the button.

What the history suggests

Payments repeat patterns found across this site. Networks that need two sides at once, cardholders and merchants, buyers and sellers, solved the problem with bold launches and hidden subsidies. Networks that rivals had to join, such as Visa and UPI, worked best when no single rival owned them. Users repeatedly decided what payment products were for: eBay sellers chose PayPal, Kenyan migrants chose to send money home with M-Pesa. Abstraction won, as Stripe hid the complexity of banking behind a simple interface. And the costs of convenience moved to whoever could not see them: cash users, merchants or the taxpayers funding a free rail.

But payments also show something more particular. Every advance in payments has been an advance in trust between strangers, built by an institution that both sides trusted more than each other: a card company, a platform, an agent network, a public utility. The design of that institution, who owns it, who pays for it and whom it protects when something goes wrong, determines the character of the whole system. That is where the philosophy begins.

Understanding the philosophy

The rest of this essay examines the ideas beneath digital payments, one at a time. Each is explained in plain terms, connected to the tradition of thought that has examined it, traced in the history above, tested against its strongest objection and related to how people and organisations decide today. The connections to philosophers, anthropologists, economists and sociologists are interpretations offered to make the ideas clearer. The builders of payment systems were solving practical problems and did not, as a rule, set out to answer these thinkers.

1. Money is a promise that strangers keep

The idea in plain terms

A banknote is a piece of paper; an account balance is a number in a database. Neither has value in itself. They are valuable because almost everyone expects that almost everyone else will accept them. Money is a shared belief, sustained by institutions, that a claim will be honoured. Every payment system is a machine for keeping that belief alive among people who do not know each other.

The tradition

Aristotle noticed this in the Nicomachean Ethics. The Greek word for coin, nomisma, comes from nomos, law or custom, he observed, because money exists not by nature but by convention, and it is in our power to change it or make it useless. Money is a measure that makes unlike things comparable, so that a builder and a shoemaker can exchange, and it exists because people agree to use it.

Two centuries of economists later disagreed about where that agreement came from. The Austrian economist Carl Menger argued in 1892 that money emerged spontaneously from barter, as traders converged on the most saleable commodity, usually a precious metal. The credit theorists, Mitchell-Innes among them, argued that money is a record of debt, a transferable promise, and that its value comes from the creditworthiness of the issuer and the obligations, including taxes, that it can settle.

In 1900 the German sociologist Georg Simmel published The Philosophy of Money, a long study of what money does to social life. Simmel described money as a claim upon society: when someone accepts money for goods, they trust that the wider community will honour it later. Money, he argued, makes relationships more impersonal and more extensive at once. It lets people deal with strangers on the basis of a number, without needing to know or trust them personally, and in doing so it widens the circle of exchange to the whole society and beyond.

In 1995 the American philosopher John Searle offered a general account of such things in The Construction of Social Reality. Some facts, such as the height of a mountain, exist independently of us. Others, which he called institutional facts, exist only because we collectively accept them, according to rules of the form “X counts as Y in context C.” A piece of paper counts as a ten-pound note in the United Kingdom; an entry in a bank’s database counts as a deposit. Money is the clearest example of an institutional fact: real, consequential, and dependent entirely on continued collective acceptance.

Where it shows up

Every institution in this history exists to make a monetary promise believable to a stranger. The Bank of England’s notes promised to pay the bearer. Diners Club guaranteed a member’s bill to a restaurant. Visa’s rules guaranteed that a card issued in one country would be honoured in another. PayPal and Alipay held money themselves and stood between buyer and seller. M-Pesa’s agents converted cash into a balance that Safaricom guaranteed. UPI is backed by the banks and the Reserve Bank. Each new system succeeded when it gave strangers a reason to accept its numbers. DigiCash failed largely because too few banks and merchants were willing to.

Stablecoins show the same structure in a new setting. They are valuable only to the extent that holders believe the issuer really holds dollar reserves. Facebook’s Libra failed not because the technology was weak but because governments refused to accept a private company as a new issuer of the promise.

The strongest objection

The objection is that this view makes money sound fragile, when in practice it is extremely robust. People keep using money through inflation, crises and bank failures. Moreover, the commodity view captured something real: gold and silver had value apart from any promise, and metallic currencies circulated for millennia among people with no shared state. These points have force. But even metallic coins circulated at values above their metal content when issued by trusted authorities, and modern money has no metal behind it at all. Its robustness is the robustness of a deeply entrenched institution, not of a natural substance.

What it means today

Every payment product is, at bottom, a promise backed by an institution, and its users are trusting that institution whether they know it or not. For businesses choosing payment partners, the question is whose promise they are relying on and what happens if it fails: a payment company’s insolvency, a stablecoin issuer’s reserves, a bank’s outage. For governments, the history shows that the right to issue and guarantee money is one they guard carefully, as Facebook discovered and as the regulation of stablecoins confirms.

2. Credit, debt and obligation

The idea in plain terms

The credit card turned borrowing into an everyday habit. Each tap is a small loan, and many people carry balances for years. Debt is not only a financial arrangement. It creates a relationship of obligation between people, and it has always carried moral weight: we speak of owing, of paying back, of settling accounts. When debt becomes easy, invisible and continuous, it changes how people live.

The tradition

In his Genealogy of Morals (1887), the German philosopher Friedrich Nietzsche argued that the moral idea of guilt grew out of the economic relation of debt. The German word Schuld means both. The creditor’s right to punish a defaulting debtor, he suggested, was the origin of the idea that wrongdoing must be paid for. Whatever one thinks of Nietzsche’s speculative history, he saw that debt binds the debtor’s future to the creditor’s claim.

Graeber developed a related argument in Debt. He argued that throughout history the language of debt has been used to make relationships of power look like fair exchanges, and that debt crises recurred so often that ancient rulers periodically cancelled debts altogether, in Mesopotamian proclamations of “clean slates” and the biblical jubilee. Historians and economists have criticised Graeber’s sweeping claims, and some of his historical generalisations are disputed. But his central point, that debt is a social and moral relationship and not only a contract, is widely shared.

Religious traditions have long regulated debt. Christianity, Judaism and Islam all at times prohibited or restricted interest on loans, and Islamic finance still structures credit to avoid it. These rules reflect a common worry: that lending at interest to people in need can trap them.

Where it shows up

The credit card transformed debt from an occasional, deliberate act, such as a mortgage or a loan for a car, into a continuous background condition of everyday spending. The Fresno drop distributed credit to people who had not asked for it. Interchange fees and rewards made cards attractive to use; interest on revolving balances made them profitable for issuers. In recent years “buy now, pay later” services have extended the same logic to online checkout, splitting purchases into instalments with little friction. Digital payments make borrowing as easy as paying.

The strongest objection

The objection is that credit is one of humanity’s most useful inventions. It lets people smooth consumption over time, start businesses, handle emergencies and buy homes. Credit cards give consumers protection that cash does not, such as the ability to dispute a fraudulent charge. Many people use them responsibly and pay their balances every month. Treating debt as a moral danger can be paternalistic. All of this is true. The concern is not credit itself but credit made so effortless that the decision to borrow disappears into the decision to pay.

What it means today

For individuals, the most useful habit is to keep the act of borrowing visible: to know, at the moment of paying, whether one is spending money or borrowing it. For product designers, the lesson is that making credit frictionless is a choice with moral weight, especially for customers who are least able to manage it. For organisations, the same applies to their own finances: easy access to credit, whether through supplier terms or corporate cards, changes spending behaviour in ways budgets rarely capture.

3. Trust between strangers at a distance

The idea in plain terms

In a village, people trade with people they know, and reputation enforces honesty. In a modern economy, most of our transactions are with strangers we will never meet. We trust them not because we know them but because we trust the systems that stand between us: card networks, platforms, banks, courts. Modern life depends on this kind of trust in abstract systems, and payments are where it is tested most often.

The tradition

Simmel wrote a short essay in 1908 called “The Stranger,” describing the trader as the archetypal stranger, someone who is in a community but not of it, and whose relations with others are therefore more objective and more dependent on general rules. In the 1970s the German sociologist Niklas Luhmann argued in Trust and Power that trust is a way of reducing complexity: we cannot check everything, so we rely on others and on systems, accepting the risk of being let down. He distinguished personal trust in individuals from system trust in institutions such as money and law.

The British sociologist Anthony Giddens developed this in The Consequences of Modernity (1990). He argued that modern societies are built on disembedding mechanisms, which lift social relations out of local contexts and stretch them across time and space. He identified two kinds: symbolic tokens, of which money is the main example, and expert systems, such as engineering, medicine and law. Both require trust in abstract systems that ordinary people cannot inspect. Giddens noted that this trust is renewed at access points, the moments when we deal with the representatives or interfaces of such systems, and that it can be shaken when they fail.

Where it shows up

The history of payments is a history of building trust between strangers. Diners Club stood behind its members’ bills; Visa’s rules made a card from a Seattle bank good in a shop in Paris. Online, PayPal’s fraud models decided which strangers could be trusted, and Alipay’s escrow removed the need to trust the seller at all, by having the platform hold the money. M-Pesa’s agents were local, trusted shopkeepers who became access points to an abstract system. Chargebacks gave card users a way to recover money if a merchant failed them, which made them willing to pay strangers online. Each of these is a disembedding mechanism in Giddens’s sense, and each succeeded by making the abstract system trustworthy at the point where people touched it.

The strongest objection

The objection is that trust in systems simply moves the problem: we must now trust the platforms and networks, which have their own interests and their own failures. Those who run the systems gain enormous power, including the power to exclude people deemed untrustworthy, often by opaque algorithms. Escrow and fraud models protect the platform as much as the customer. This objection is right, and it explains much of the regulation of payments. Trust in systems is necessary for an economy of strangers, but it requires the systems themselves to be accountable.

What it means today

Any business that sells to strangers is also in the business of trust, and the payment experience is one of its main access points. Disputes, refunds, fraud handling and the clarity of charges are not back-office functions; they are where customers decide whether to trust the business again. For those designing platforms, the history suggests that whoever solves the trust problem between strangers gains great power over both, and should expect to be held accountable for how they use it.

4. Who pays for convenience

The idea in plain terms

Paying by card or phone feels free to the customer. It is not. Someone pays the networks, banks and processors that make it work. In card systems, the merchant pays and passes the cost on in prices to everyone, whether they use cards or not. In free public systems, taxpayers, banks or the users’ data pay. The question of who pays is never neutral: it decides who uses the system, how much, and who subsidises whom.

The tradition

Economists who studied card networks in the early 2000s found that they were examples of what Jean-Charles Rochet and Jean Tirole called two-sided markets. A platform serving two groups, here cardholders and merchants, must get both on board, and it typically charges the side that cares less about price and subsidises the side whose participation is harder to win. For card networks, that meant charging merchants and rewarding cardholders. The site’s pattern who pays shapes what gets built collects other cases.

The deeper ethical question is one of fairness in hidden transfers. The principle that those who benefit from a service should pay for it, and that costs should not be shifted onto people who did not choose them, has roots in the economics of externalities developed by Arthur Pigou in The Economics of Welfare (1920). An externality is a cost or benefit that falls on someone outside a transaction. When a shopper pays by a rewards card and the merchant raises prices for everyone to cover the fee, the cash-paying shopper bears an externality.

Where it shows up

The interchange system worked exactly as two-sided market theory predicts. Merchants paid, cardholders were rewarded, and cards spread. The Boston Fed study estimated the resulting transfer from cash users to card users, and its regressive pattern. The rules that stopped merchants charging card users more kept the cost hidden. Regulators in Europe and the United States capped interchange, partly to limit the transfer. India’s zero-fee UPI took the opposite approach: neither side paid at the point of sale, and the costs moved to subsidies and to the business models of the apps. By 2026 the payments body was proposing a modest fee on larger merchant payments, an acknowledgement that someone had to pay.

The strongest objection

The objection is that card fees pay for real benefits, such as fraud protection, credit, convenience and faster checkout, which merchants also enjoy through higher sales. Caps on interchange in some places led banks to cut rewards and raise other fees, so consumers may not have gained overall. And cash has hidden costs too, borne by merchants and society. These points are valid; the evidence on the effects of interchange caps is mixed. But they do not dissolve the basic point that the person choosing the payment method usually does not see its cost, and that the people who pay are not always the ones who benefit.

What it means today

For businesses, payment costs are among the largest and least examined expenses, and the choice of which methods to accept, and whether to steer customers towards cheaper ones, is a strategic decision. For policymakers, making payment costs visible to those who choose them is usually better than hiding them. And for anyone designing a payment system, the history shows that “free” is a decision about who pays, not an absence of cost.

5. The pain of paying

The idea in plain terms

Handing over banknotes feels different from tapping a phone. Watching cash leave your hand creates a small sensation of loss, which makes people think about whether a purchase is worth it. Digital payments remove that sensation. As a result, people tend to spend more, and to think less about what they spend.

The tradition

Classical economics assumes that money is money, whatever form it takes. Behavioural economists showed otherwise. In 1998 Drazen Prelec and George Loewenstein proposed a theory of mental accounting in which paying causes an immediate psychological pain that is coupled to, and reduces, the pleasure of consumption. Payment methods that decouple payment from consumption, such as credit cards billed later, reduce the pain and increase spending.

In 2001 Prelec and Duncan Simester ran an experiment in which students bid in a real sealed auction for tickets to a sold-out Boston Celtics game. Half were told they would have to pay by cash and half by credit card. Those paying by card bid, on average, more than twice as much.²⁴ Later studies found the effect varies, but the general finding, that more abstract forms of payment increase willingness to spend, has been widely replicated.

The sociologist Viviana Zelizer, in The Social Meaning of Money (1994), showed that people do not treat money as fungible. They earmark it for different purposes, such as housekeeping money, gifts and savings, often using physical separation, such as envelopes and jars. Cash makes earmarking easy. Digital balances blur it.

Where it shows up

Every stage in the history of payments has reduced the pain of paying. The charge card deferred payment to the end of the month. Contactless cards removed the signature and the PIN for small amounts. Apple Pay reduced payment to a glance at a phone. UPI turned payment into a scan and a tap. Online, stored card details and one-click checkout removed nearly all friction. Buy-now-pay-later services split prices into small pieces. Each innovation was marketed as convenience, and each made spending easier.

The strongest objection

The objection is that digital payments also make spending more visible, not less. Banking apps can categorise and display every transaction, set budgets and send alerts, which cash never could. For disciplined users, digital money may make budgeting easier. And convenience has real value: queues are shorter and transactions safer. This is true, and some studies find that the effect of payment method on spending shrinks as people grow used to it. But the default experience of digital payment is designed to remove friction, and tools to restore awareness depend on users seeking them out.

What it means today

For individuals, it is worth deliberately restoring some of the friction that digital payments remove, through spending limits, regular review or separate accounts for different purposes. For businesses, the temptation to increase spending by reducing the pain of paying is strong, and the line between convenience and manipulation is worth watching. For organisations, corporate cards and frictionless procurement have the same effect on spending as consumer cards; controls that once seemed bureaucratic may be restoring a useful pause.

6. Payments as public infrastructure

The idea in plain terms

Roads, water and electricity are often built or regulated by the state because everyone needs them and private monopolies over them would be dangerous. Are payment rails the same? For most of modern history, retail payments were run by private banks and card networks. India and Brazil chose to treat payment rails as public infrastructure, built by public or not-for-profit bodies, open to all, and free or nearly free to use, with private companies competing on top. The results have challenged the assumption that private networks are the natural form of payments.

The tradition

In 1954 the American economist Paul Samuelson defined public goods as those that are non-rival, one person’s use does not reduce another’s, and non-excludable, people cannot easily be prevented from using them. Markets tend to underprovide such goods because people can benefit without paying. Infrastructure that has strong network effects and natural monopoly characteristics, where one network is more efficient than several, has long been a candidate for public provision or regulation.

The economist Mariana Mazzucato argued in The Entrepreneurial State (2013) that governments have often been the source of fundamental innovations later commercialised by private companies, and that public investment can shape markets rather than merely fix their failures. The concept of digital public infrastructure, promoted by India’s government and international bodies in the 2020s, applies these ideas to identity, payments and data sharing: shared, open systems on which public and private services can be built.

Where it shows up

UPI is the clearest example. NPCI, owned by banks and not-for-profit, built an open, interoperable rail. Any bank and many non-banks could connect. Payments were made free by law. Private apps competed for users on top. The result was adoption on a scale no private network had achieved, from street vendors to large businesses. Brazil’s Pix, run directly by the central bank, followed the same logic. By contrast, card networks are private, owned by shareholders since Visa and Mastercard became public companies, and financed by fees. The United States’ trade complaint against Pix, and the concentration of UPI traffic in two private apps despite the market-share cap, show the tensions: public rails threaten private businesses, and private platforms can still capture public infrastructure.

The strongest objection

The objection is that public infrastructure can stifle innovation, depend on subsidies that become unsustainable, and give the state control over a vital system, including the ability to monitor and block payments. Free payments do not cover costs; someone pays through taxes or cross-subsidies, and underfunded systems may be less resilient and secure. Private networks, funded by fees, invested heavily in fraud prevention and global reach. These concerns are real. UPI’s cost problem and the 2026 debate over merchant fees illustrate them. The answer of the public-infrastructure tradition is not that the state should run everything, but that the shared rail should be open and governed in the public interest, with competition above it.

What it means today

For businesses, public payment rails lower costs and open markets, but they also change the competitive landscape: when the rail is free, advantage moves to the apps, the data and the services built on top. For policymakers, the choice between public and private rails is a choice about who captures the value of a network everyone needs, and about how to fund its security and resilience. For other countries, India’s and Brazil’s experience shows that the design of payment infrastructure is a policy decision, not a fact of nature.

7. Every payment is a record

The idea in plain terms

Cash leaves no trace. Digital payments leave a permanent one: what was bought, where, when, from whom and for how much. Taken together, a person’s payment records reveal their movements, habits, health, relationships and beliefs. A cashless society is also a recorded society, and whoever can read the records has a kind of power over the people they describe.

The tradition

The English philosopher Jeremy Bentham designed in the 1780s a prison he called the Panopticon, in which a single guard could watch all the inmates without being seen, so that inmates, never knowing when they were watched, would behave as if they always were. In Discipline and Punish (1975) the French philosopher Michel Foucault took the Panopticon as a symbol of modern power, which works less through force than through observation, records and the internalised expectation of being watched.

Chaum’s 1985 article applied this concern directly to payments. He warned that as more transactions became electronic, organisations would be able to compile dossiers on individuals’ lives, and he proposed cryptographic systems that would let people pay without being identified, making payments verifiable without making them traceable. Many later privacy advocates, and the designers of Bitcoin, drew on his ideas.

On the other side, the economist Kenneth Rogoff argued in The Curse of Cash (2016) that large-denomination banknotes facilitate tax evasion, corruption and crime, and that phasing them out would bring substantial benefits. Governments have long sought to trace payments for taxation and law enforcement, and anti-money-laundering rules require banks to know their customers and report suspicious transactions.

Where it shows up

The move from cash to cards, platforms and UPI has made most payments traceable. Card networks, banks, payment apps and platforms hold detailed records, used for fraud detection, credit scoring, marketing and, under legal process, law enforcement. India’s demonetisation was justified partly as an attack on untraceable wealth. Chaum’s DigiCash failed commercially; anonymous digital payment remains marginal. Central banks designing digital currencies have had to decide how much privacy to give users, and the debate over who could see the records has been one of the main obstacles to them.

The strongest objection

The objection is that the record serves the public good. Traceable payments reduce tax evasion, corruption, money laundering and the financing of crime. Fraud detection, which depends on analysing payment records, protects users. Honest people have little to fear from records that are held securely and accessed lawfully. These are serious points; Rogoff’s argument is not easily dismissed. The counter-argument is that records outlast the purposes they were collected for, that access rules can change, and that the knowledge of being recorded changes how people behave, as Bentham intended.

What it means today

For individuals, it is worth knowing which payment records exist and who can see them. For businesses, payment data is valuable and dangerous: it supports fraud prevention and personalisation, and it creates obligations and risks. For societies, the disappearance of cash is a decision about privacy as much as convenience, and it deserves to be made deliberately. Keeping some form of private payment available, whether cash or privacy-protecting digital money, preserves a freedom that is hard to recover once lost.

8. Undoing a payment

The idea in plain terms

Mistakes and frauds happen. A payment sent to the wrong person, a purchase from a merchant who never delivers, money handed to a scammer: can it be undone? Card systems built in a right of reversal, the chargeback. Instant bank transfers and cryptocurrencies are designed to be final. The ability to undo a payment protects the payer and burdens the payee; finality does the opposite. How a system answers this question decides who bears the risk of human error and deceit.

The tradition

The political philosopher Hannah Arendt, in The Human Condition (1958), argued that human action has two fundamental predicaments. It is irreversible, since we cannot undo what we have done, and it is unpredictable, since we cannot foresee its consequences. Each has a remedy that is itself a human capacity. The remedy for irreversibility is forgiveness, which releases people from the consequences of what they did; the remedy for unpredictability is promising, which binds the future. Without forgiveness, she wrote, we would be confined to a single deed from which we could never recover; without promises, we could never rely on one another.

Law has long recognised related ideas. Contracts can be rescinded for fraud or mistake. The common law of restitution requires those who receive money by mistake to return it. And in decision theory, the value of being able to reverse a decision is a recognised consideration: irreversible decisions deserve more care, an idea explored in the site’s idea Irreversibility and the Value of Waiting.

Where it shows up

The history of payments can be read as a long negotiation between Arendt’s two remedies. Every payment is a promise; the question is whether it can be forgiven. The US limit on cardholder liability and the chargeback rules of the card networks built forgiveness into cards, which is part of why people were willing to use them with strangers online. Bitcoin was designed to make payments final, without any institution able to reverse them, which is attractive to those who distrust institutions and also to thieves. Instant bank payments such as UPI and Britain’s Faster Payments are final by design, so when scams grew, regulators had to construct forgiveness after the fact, as with the UK’s mandatory reimbursement, which reversed the loss for most victims by making banks bear it.

The strongest objection

The objection is that reversibility creates its own injustices. Chargebacks can be abused by buyers who receive goods and then dispute payment, imposing losses on honest merchants. Reimbursement schemes may reduce the care people take, a problem economists call moral hazard, and shift costs to banks that then restrict services or raise fees for everyone. Finality gives merchants certainty and lowers the cost of payments. These are real trade-offs, and different systems strike different balances. The point is not that every payment should be reversible but that the balance should be chosen deliberately, with an eye to who can bear the risk.

What it means today

For individuals, the question to ask of any payment method is: if this goes wrong, can I get my money back, and who decides? For businesses, the reversibility of the payment methods they accept shapes their exposure to fraud and disputes, and is worth understanding before choosing them. For designers of payment systems, the faster and more final a payment, the stronger the safeguards needed before the button is pressed, because there will be little forgiveness after.

Bringing the threads together

Payments rest on a promise that strangers keep: a shared belief, sustained by institutions, that a number will be honoured. Credit made that promise flexible, and debt made it binding. Abstract systems, from card networks to platforms to public rails, made strangers trustworthy to each other, and gained power over both. The question of who pays for convenience was answered, for decades, by hiding the cost in prices that everyone paid; India answered it by making payments free and has now had to ask it again. Digital payment removed the pain of paying, which made commerce easier and spending less deliberate. It turned every purchase into a record. And it forced each system to choose between finality and forgiveness, between protecting the payee and protecting the payer.

Paying a stranger

When Bank of America mailed sixty thousand cards to Fresno, it was trying to solve a problem as old as trade: how to persuade strangers to trust a promise to pay. The answers since then have been remarkable. A shopper can pay a merchant on the other side of the world in seconds; a migrant worker can send money home by text message; a street vendor can accept payment with a printed square. The institutions that made this possible, card associations, platforms, agent networks and public rails, are among the most consequential inventions of the last century.

The history suggests some working principles. Know whose promise you are relying on when you pay or are paid. Ask who pays for the convenience, because someone always does. Keep the act of spending, and especially of borrowing, visible enough to be a decision. Treat the record every payment leaves as something to protect. And when designing or choosing a payment system, decide deliberately who bears the risk when something goes wrong, because speed without forgiveness places that risk on the people least able to carry it. A payment is a small act of trust between strangers. The systems that carry billions of them a day deserve to be designed as if that trust were the thing being built.

Selected bibliography

• Aristotle. Nicomachean Ethics, Book V. Fourth century BCE.
• Bentham, Jeremy. Panopticon; or, The Inspection-House. 1791.
• Nietzsche, Friedrich. On the Genealogy of Morals. 1887.
• Menger, Carl. “On the Origin of Money.” Economic Journal 2, no. 6 (1892): 239–255.
• Simmel, Georg. The Philosophy of Money. 1900.
• Simmel, Georg. “The Stranger.” In Soziologie, 1908.
• Mitchell-Innes, A. “What Is Money?” Banking Law Journal 30 (1913): 377–408.
• Pigou, A. C. The Economics of Welfare. Macmillan, 1920.
• Samuelson, Paul A. “The Pure Theory of Public Expenditure.” Review of Economics and Statistics 36, no. 4 (1954): 387–389.
• Arendt, Hannah. The Human Condition. University of Chicago Press, 1958.
• Foucault, Michel. Discipline and Punish. 1975.
• Luhmann, Niklas. Trust and Power. Wiley, 1979.
• Chaum, David. “Security without Identification: Transaction Systems to Make Big Brother Obsolete.” Communications of the ACM 28, no. 10 (1985): 1030–1044.
• Giddens, Anthony. The Consequences of Modernity. Polity, 1990.
• Zelizer, Viviana A. The Social Meaning of Money. Basic Books, 1994.
• Nocera, Joseph. A Piece of the Action: How the Middle Class Joined the Money Class. Simon & Schuster, 1994.
• Searle, John R. The Construction of Social Reality. Free Press, 1995.
• Prelec, Drazen, and George Loewenstein. “The Red and the Black: Mental Accounting of Savings and Debt.” Marketing Science 17, no. 1 (1998): 4–28.
• Prelec, Drazen, and Duncan Simester. “Always Leave Home Without It: A Further Investigation of the Credit-Card Effect on Willingness to Pay.” Marketing Letters 12, no. 1 (2001): 5–12.
• Rochet, Jean-Charles, and Jean Tirole. “Platform Competition in Two-Sided Markets.” Journal of the European Economic Association 1, no. 4 (2003): 990–1029.
• Hock, Dee. One from Many: VISA and the Rise of Chaordic Organization. Berrett-Koehler, 2005.
• Schuh, Scott, Oz Shy and Joanna Stavins. “Who Gains and Who Loses from Credit Card Payments? Theory and Calibrations.” Federal Reserve Bank of Boston Public Policy Discussion Paper 10-3, 2010.
• Graeber, David. Debt: The First 5,000 Years. Melville House, 2011.
• Mazzucato, Mariana. The Entrepreneurial State. Anthem, 2013.
• Rogoff, Kenneth S. The Curse of Cash. Princeton University Press, 2016.
• Suri, Tavneet, and William Jack. “The Long-Run Poverty and Gender Impacts of Mobile Money.” Science 354, no. 6317 (2016): 1288–1292.
• Regulation (EU) 2015/751 on interchange fees for card-based payment transactions.
• Reserve Bank of India. Annual Report 2017–18. August 2018.

Research limitations

The origin story of Diners Club is disputed, and its founders’ roles vary by account. Figures for the Fresno drop come from later histories; Bank of America’s early losses are widely reported but not quantified here. The Boston Fed estimates come from a calibrated model, and a later study funded by the banking industry disputed them. Prelec and Simester’s experiment was small, and later studies have found the effect varies by good and context. PayPal’s early fraud losses are reported in participants’ accounts. The estimate of M-Pesa’s effect on poverty is from a single influential study whose interpretation other researchers have challenged. Stripe’s volume figures are the company’s own. UPI volume figures are from NPCI as reported by the press; the calendar-year totals vary slightly by source. ACI Worldwide’s estimate of India’s share of real-time payments covers all real-time systems in India, not only UPI. Demonetisation figures are from the Reserve Bank of India; the relationship between demonetisation and UPI’s growth is contested, and this essay does not resolve it. The merchant fee on UPI announced in September 2026 had been deferred, with no new date, at the time of writing. The status of the US case against Visa and of the US trade investigation into Brazil may have changed. The UK reimbursement figures are from the regulator and an independent evaluation as reported by secondary sources. The philosophical connections to Aristotle, Menger, Mitchell-Innes, Simmel, Searle, Nietzsche, Graeber, Luhmann, Giddens, Rochet and Tirole, Pigou, Prelec, Loewenstein, Simester, Zelizer, Samuelson, Mazzucato, Bentham, Foucault, Chaum, Rogoff and Arendt are the author’s interpretation; their works are paraphrased apart from short, attributed phrases. The decisions in the decisions box are judged by the author, separating what could be known at the time (process) from what happened (call).

Footnotes

1. Stripe, “Stripe publishes 2025 annual letter,” 24 February 2026. ↩ ↩²

2. “UPI hits new record with 24.51 bn transactions in August,” Entrackr, September 2026. ↩ ↩²

3. David Graeber, Debt: The First 5,000 Years (Melville House, 2011), chapters 2 and 3; A. Mitchell-Innes, “What Is Money?” Banking Law Journal (1913). ↩

4. “Credit cards: as American as apple pie,” PaymentsJournal, quoting a 1994 Washington Post account; Joseph Nocera, A Piece of the Action (1994). ↩

5. Public Law 91-508 (1970), amending the Truth in Lending Act; see Encyclopedia.com, “Visa International”. ↩

6. “Dee Hock, credit card visionary, is dead at 93,” Boston Globe, August 2022; Britannica, “Visa Inc.”. ↩

7. Scott Schuh, Oz Shy and Joanna Stavins, “Who Gains and Who Loses from Credit Card Payments? Theory and Calibrations,” Federal Reserve Bank of Boston, 2010. ↩

8. Regulation (EU) 2015/751; see Osborne Clarke. On the US cap, Wikipedia, “Durbin amendment”. ↩

9. “DOJ accuses Visa of monopolizing debit card market,” Fortune, 23 September 2024. ↩

10. David Chaum, “Security without Identification,” Communications of the ACM 28 (1985); Wikipedia, “DigiCash”. ↩

11. Wikipedia, “Timeline of PayPal”. ↩

12. “How Alipay is killing cash in China,” KrASIA; Wikipedia, “WeChat red envelope”. ↩

13. CGAP, “Why does M-PESA lift Kenyans out of poverty?”. ↩

14. Tavneet Suri and William Jack, “The long-run poverty and gender impacts of mobile money,” Science 354 (2016); summary by FSD Kenya; critique by Bateman, Duvendack and Loubere. ↩

15. “UPI’s solutions are UPI’s problems,” Catalyst Analyst, on the 2016 pilot and launch and early adoption. ↩

16. “99.3% of demonetised currency returns to RBI,” The Quint, August 2018; The Week, on the share of currency in circulation. ↩

17. NPCI circular on zero MDR, reported by YourStory, February 2020. ↩

18. ACI Worldwide, Prime Time for Real-Time 2024, press release, 29 April 2024. ↩

19. “NPCI extends 30% UPI market share cap deadline to December 2026,” Business Today, December 2024; “PhonePe, Google Pay combined UPI market share drops below 80%,” Outlook Business. ↩

20. “MDR on UPI explained,” Manorama Yearbook, 14 October 2026. ↩

21. “US investigation into Brazil’s Pix payment system,” Payment Expert, July 2025; Payment Expert, August 2026. ↩

22. “Trump signs law to create regulatory regime for stablecoins,” CBC, 18 July 2025. ↩

23. “APP fraud reimbursement: the evidence after one year,” Bratby Law; “PSR reveals 88% of money lost to APP scams reimbursed,” FStech. ↩

24. Drazen Prelec and Duncan Simester, “Always Leave Home Without It,” Marketing Letters 12 (2001); see MIT Spectrum. ↩