Chapter 1: The Exchange That Cannot Quite Happen
Prologue: The Number at the Bottom of the Page
Beatrice can put a number on the repair before she knows whether she can obtain the part. The customer wants to know what the job will cost. The supplier wants a specification before confirming availability. Between them sits an estimate that looks more settled than the work behind it.
Fallow Repair is a hypothetical business. Its owner, Beatrice, repairs household equipment that customers would prefer to keep using. A familiar fault can conceal an unfamiliar complication: a component has changed, an older fitting needs an adapter, or the replacement is available only in a batch larger than the job requires. She can inspect carefully without knowing everything she will find once the equipment is open. The uncertainty belongs to the service she sells. Her quotation decides how much of it she promises to bear.
One customer needs the equipment back for a particular occasion. Another can wait but has little money to spare. Offering both the same number may seem evenhanded, yet the promises they need are different. A dependable completion date can require a different source for the part, time reserved in the workshop, or a willingness to decline another job. A flexible date can make the work easier to fit around delays. Those differences need not appear on the first line of an estimate. They still affect what Beatrice can responsibly offer.
The customer sees the total. Beatrice sees obligations attached to it: the supplier's terms, the time she will spend checking compatibility, the possibility of reopening the repair, and the explanation she owes if her first judgment proves wrong. Neither view is complete. Her account of the work may overlook an easier method another repairer has found. The customer's comparison may overlook that another quotation excludes a necessary step. Agreeing on the number will not settle these questions by itself.
She could protect herself with qualifications. Parts subject to availability. Additional faults charged separately. Completion subject to inspection. Each reservation has a reason; together they can leave the customer unable to tell what has been promised. A quotation that transfers every uncertainty to the buyer may preserve the business's freedom while giving the buyer very little to accept. At the other extreme, a firm price can be so ambitious that keeping it requires rushed work or an argument about what was included. Clarity comes from deciding which uncertainties can be bounded, who is able to manage the others, and when the customer will be asked to decide again.
The price is part of that decision. It can support a workable promise or disguise an unresolved one.
Beatrice also has to consider people who have not yet come through the door. Reserving the workshop for this repair affects the next customer's waiting time. Accepting a difficult job at a familiar charge can leave too little capacity for routine work that pays the rent. Raising the charge may protect that capacity, or merely lose a customer whose repair would have fitted well. She cannot read the missing demand from an empty appointment. She has to judge what the absence means.
This gives a price two lives. Before the transaction it is an offer, built from expectations about costs, customers, alternatives, and the value of time. Afterward it becomes a record, apparently precise: someone paid this amount for this service. The record usually preserves less than the decision contained. It may omit the urgency, the assurance, the options refused, and the condition of the equipment. Comparing two recorded prices without those details can create an impressive account of a bargain that neither participant would recognize.
The same difficulty grows with scale. A platform can charge one group so that another will participate. A theatre can alter an offer as the performance approaches. A program can revise a price after observing another program's revision. These arrangements produce numbers faster than a workshop owner could manage, but speed leaves the underlying questions intact. What is being offered? Who can respond? Which costs enter the calculation, and which fall on someone outside it? What has the seller learned from a purchase, and what has it merely assumed?
We will follow these questions through exchanges that cannot quite happen, signals that arrive after commitments have been made, and offers whose apparent simplicity depends on conditions the customer may never inspect. Some chapters concern a seller's calculation. Others concern what happens when many such calculations meet. A useful account has to move between those scales without supposing that the logic of one settles the other. A shop's sensible attempt to cover its costs does not establish that every customer can afford its service. A buyer's pleasure at a discount does not reveal whether the offer can last.
There are also things a price cannot resolve. Someone who lacks the money to bid remains outside a comparison of bids. A person who receives a service without paying a bill may surrender time or accept interruptions. A dataset can be copied without exhausting the original, while the work that makes it reliable still needs support. In each case, attending to the number is necessary and insufficient. We need to know how the exchange was made possible and what it leaves unfinished.
Back at Fallow Repair, Beatrice can improve her estimate without claiming to know the customer's entire valuation or every future cost. She can separate the work she has inspected from the complication she has not. She can explain when an additional decision would be needed, and make the proposed date consistent with the part she can actually obtain. The customer may still decline. Another repairer may make a better offer. A clear refusal would tell her more than an acceptance founded on a misunderstanding.
For now, the total remains at the bottom of the page. Above it, she has room to say what she is willing to do. The usefulness of the number depends on whether those words describe a service someone can recognize, compare, and reasonably expect to receive.
Chapter 1: The Exchange That Cannot Quite Happen
Ada has fruit to sell and a machine that needs attention. In the hypothetical Morrow Orchard, a sorting conveyor must be serviced before the next harvest reaches the packing shed. Felix, the technician able to do the work, has time before the harvest but little use for fruit in the quantity Ada can offer. His household cannot consume it quickly enough. His suppliers expect payment for replacement parts, and his own working day cannot be stored until Ada finds a better use for it. Both can recognize the benefit of a repair. That recognition does not yet produce an exchange.
Offering more fruit may leave the problem intact. Felix would acquire a larger disposal task, while Ada would sacrifice stock she could otherwise sell. The difficulty concerns the form and timing of payment before it concerns the generosity of either party. A seemingly simple disagreement over terms contains several separate questions: what is being transferred, when it becomes usable, who must find the next buyer, and who carries the risk before that buyer appears.
It is useful to begin here because a price can look like a property attached to an object. We see the number on a quotation and forget how much coordination allows anyone to write it. Ada and Felix need a way to make unlike obligations comparable. They also need a way to discharge those obligations. Agreement about the first does not guarantee agreement about the second. Their orchard is an imagined analytical case, not a reconstruction of how money began.
A surplus in the wrong place
Suppose Ada offers Felix a share of the coming crop. This is a claim on future fruit, not fruit available today. Felix must decide whether the conveyor service can wait for payment, whether Ada can deliver the promised quality, and whether he can use or transfer the claim when it arrives. Even if Ada is entirely reliable, the crop may disappoint. A payment arrangement can fail without anybody cheating. The promise puts weather, storage, grading, and collection between the technician and the purchasing power he needs.
Ada might instead offer stored fruit from the previous picking. Felix would then face less uncertainty about delivery but more urgency about disposal. A crate available now may be less useful to him than a smaller claim deliverable when a buyer needs it. Physical possession and economic usefulness do not arrive together automatically. Transport to his workshop, space in his vehicle, the need to inspect the contents, and the effort of finding customers all affect what he can accept. The fruit is valuable, yet its value does not release him from these tasks.
The two parties could search for someone who wants the fruit and can provide something Felix wants. An imagined local caterer might accept produce and offer prepared meals. But Felix needs parts and fuel as well as meals. Adding participants increases the possible matches and the number of agreements that must remain aligned. If the caterer will accept fruit only after a large event is confirmed, Ada cannot treat that acceptance as secure. If Felix hopes to exchange the meals for parts, he must find a supplier willing to accept that further bargain. A sequence of individually plausible exchanges can remain collectively impractical.
There is no reason to assume that every link will occur at the right time. Ada might deliver fruit before learning that the caterer cannot fulfill the next obligation. Felix might complete the repair before the final supplier agrees to participate. Someone must advance resources while the chain is incomplete. That person effectively finances the arrangement and bears the danger that a later agreement fails. The cost is not abolished because nobody has written an interest charge. It appears as waiting, exposure, or a refusal to take the first step.
This search problem differs from a judgment that fruit is worth little. Ada may have excellent fruit and Felix may admire it. What he lacks is a sufficiently dependable route from this particular payment to his own obligations. The same asset could be easy payment for a buyer with storage and established customers, difficult payment for a technician already travelling between jobs. Acceptability depends partly on the recipient's position in a network of further exchanges.
There is a further problem if Ada and Felix try to compare several possible exchange chains. One route may promise a more desirable final payment but require more people to agree. Another may yield less while allowing immediate settlement. Comparing only the final goods ignores the reliability of the route. Felix could reasonably prefer a smaller, certain receipt to a larger claim dependent on several unconfirmed participants. Ada could reasonably reject an arrangement that asks her to make an irreversible delivery before learning whether the service will happen.
The parties also need to know which failed link releases the others. If a caterer declines Ada's fruit, is Felix still obliged to reserve the maintenance day? If the repair cannot proceed because a part is missing, must Ada retrieve produce already delivered? A chain can create obligations whose cancellation is harder than their creation. The work of unwinding an arrangement belongs in the assessment of whether it is practical, especially when perishable goods are involved. Fruit can lose usefulness during negotiations even when every participant eventually agrees to reverse their part.
Pieces, grades, and units
Divisibility introduces another obstacle. The maintenance visit includes inspection, adjustment, and the possibility of replacing a component. Ada can divide fruit into smaller lots, but the technician cannot always divide his journey or his preparation proportionately. A shorter repair does not release an equally smaller amount of travel time. If the inspection reveals a fault requiring another visit, the original offer must specify whether that return is included. Otherwise the parties may agree to a quantity of fruit while imagining different quantities of service.
Counting crates seems to supply a unit until Ada and Felix consider their contents. Crates differ in weight, condition, variety, and expected storage life. A promise of a crate must identify which differences are allowed. If Felix receives fruit suitable only for immediate processing, he cannot assume that a household buyer will value it as fresh produce. The unit of payment therefore includes a standard, however informal. Without one, a dispute about quality can masquerade as a dispute about the agreed quantity. Service requires its own standard. A machine that runs at the end of the visit may still perform poorly under a full load. Felix can promise a defined inspection and adjustment more confidently than an uninterrupted season. Ada needs to know which promise she is buying. If she treats the payment as securing a successful harvest while Felix treats it as compensation for specified work, no common payment medium can rescue their misunderstanding. A price needs an object, and an object in commerce is often a description of duties rather than merely a physical thing.
They could create a common unit for discussion without selecting a common object for payment. Ada might express the service and the fruit in an agreed accounting unit, then settle the balance through several deliveries. That would simplify comparison. It would not remove the need to inspect the deliveries, record credits, or decide what happens after a short crop. A unit of account helps people state a debt. A means of settlement allows them to close it. Treating these functions as interchangeable makes the proposed solution look more complete than it is.
Consider the effort of revising terms after Felix discovers an additional fault. If every item must be compared directly with every other item, the parties must renegotiate relationships among labour, parts, fruit grades, and delivery dates. A shared unit lets them identify the additional obligation separately from the original bargain. This is a considerable convenience. Yet it also creates a new question: does the unit retain a dependable relationship to what either party will need later? A stable label is not enough if the goods obtainable under that label change sharply before settlement.
The person who accepts what he does not want
A widely accepted medium offers Felix another route. He can receive something he does not want to consume because others will accept it. Ada no longer needs to identify his suppliers or solve their preferences. Felix can separate earning from spending and choose his purchases after completing the job. The medium reduces the number of personal coincidences needed for exchange. It does so by relying on a broader expectation of acceptance, an expectation neither participant can manufacture alone.
This expectation has a practical geography. A token accepted at the orchard's own stall might buy fruit, but Felix already has enough access to fruit. A voucher accepted by several nearby businesses would be more useful if those businesses supply things he needs. The number of participants matters less than the range of obligations he can discharge. Many willing recipients concentrated in one narrow activity may provide less useful acceptance than a smaller network spanning fuel, food, parts, and rent. The relevant network follows payments as well as preferences.
Acceptance can also be conditional. A supplier may take a voucher only for stock already on hand, refuse it for special orders, or require the holder to spend it before a deadline. Felix then receives a purchasing option with restrictions rather than an unrestricted substitute for payment. Ada may sincerely describe the voucher as equivalent to its stated amount while Felix discounts it because the restrictions do not fit his business. Their disagreement concerns future opportunities, not simply confidence in Ada's honesty.
If an intermediary promises to convert the voucher into a more acceptable medium, the conversion promise becomes part of the exchange. Felix needs to know when conversion is available, who funds it, and what happens when many holders ask at once. The intermediary must maintain resources or arrange incoming payments. A convenient transferable claim can conceal a demanding balance between obligations due now and assets available later. Writing a transferable promise is easy; sustaining the ability to honor it requires continuing coordination.
A recipient may accept an unfamiliar claim because a trusted customer offers it, then discover that others do not share that trust. Personal confidence has travelled less far than the paper. Ada's reputation with Felix cannot automatically become Felix's reputation with a distant supplier. Each additional transfer asks someone to rely on an issuer or an arrangement they may not know. The friction helps explain why a payment medium can be more useful than another asset of similar apparent worth. Its usefulness includes the reduced work of persuading the next recipient.
An accepted medium reduces dependence on this particular chain. Ada's sale need not involve Felix's eventual supplier, and Felix need not wait for the exact buyer who prefers her crop. Their transactions remain connected through a wider system of acceptance and settlement. They gain freedom to choose partners and dates, while relying more heavily on that system's continuity. This is a change in the structure of dependence. It deserves attention precisely because the resulting payment can feel routine. The routine works when many other people keep accepting, recording, and settling obligations that Ada and Felix never have to examine individually.
An observed exchange economy
R. A. Radford's 1945 account of prisoner of war camps describes cigarettes becoming a currency alongside continuing barter. Public notices and organized exchange helped communicate terms; interruptions to cigarette supplies disturbed trading. His account concerns a confined setting supplied largely through rations and parcels, with limited production. It is an eyewitness account, not a controlled experiment or evidence that all societies passed through barter before acquiring money.[1-1]
The orchard should not be mistaken for a miniature version of that camp. Ada produces goods for sale, Felix buys productive inputs, and both have commitments outside their immediate locality. The value of bringing the historical account beside this imagined case is to keep several functions visible at once. People need to compare goods, communicate terms, transfer something acceptable, and complete settlement. These tasks may be handled together or by different arrangements. Their coexistence does not establish a universal historical sequence.
Nor should ease of exchange be confused with a complete account of an institution's origin. An arrangement can solve a coordination problem without having been invented consciously for that purpose. It can also be imposed, inherited, revised, or supported by authorities whose reasons extend beyond trade. To understand Ada's present difficulty, we do not need a story about the first money. We need to identify what has to happen between her offer and Felix's ability to meet his own commitments.
The repair financed by a promise
Ada and Felix could dispense with immediate payment and create a debt. Felix performs the service now; Ada settles after selling the crop through ordinary channels. The mismatch between fruit and parts disappears from the immediate transaction. It returns as a question about Ada's future receipts. Felix must assess whether he can wait, whether she will have sufficient funds, and what priority his claim has among her other obligations. Credit changes the location of the difficulty. It may still be the best available solution.
A repeated relationship can make this possible. Felix may understand Ada's operating calendar and recognize that a delayed receipt is routine rather than alarming. Ada may know that keeping his confidence will matter at the next maintenance visit. These expectations can support a promise without requiring constant verification. But repeated dealings do not remove exposure to a common shock. A disappointing harvest can reduce Ada's receipts precisely when several other growers also delay payment, leaving Felix short of funds across his customer list.
The practical burden includes records. Ada needs to distinguish work completed from work still contingent on inspection. Felix needs a receivable he can reconcile with materials purchased and visits made. A dispute about whether a payment was partial or final can damage a relationship that began with sincere intentions. A written record does not require distrust; it allows the parties to remember an obligation without relying on identical recollections. The record also helps a bookkeeper or successor understand an agreement they did not personally make.
Suppose Ada offers to pay earlier in exchange for Felix accepting less. The offer may be sensible if he values immediate purchasing power. It may be impossible if Ada has no funds before the crop sells. Conversely, Felix might agree to a later payment if Ada supplies a deposit sufficient for parts. Dividing payment by stages can align resources with the sequence of work. Such arrangements require more administration than a single exchange, but the additional detail can make a viable bargain out of a failed simple offer.
Neither party can infer fairness from the existence of agreement alone. Ada may accept difficult credit terms because the conveyor is essential, while Felix may extend credit because losing the customer would leave a gap in his schedule. Outside options influence what each can refuse. There can be real gains from exchange and an uneven distribution of those gains. A common medium makes the distribution easier to express, but it does not decide whether the underlying dependency is acceptable.
What a quoted number presupposes
Once Ada can pay in an accepted medium, Felix can quote a service charge. That number gathers several obligations into a manageable form, but it does not make their boundaries disappear. Does the charge cover only the scheduled visit or an emergency return? Must Ada provide access to the machine? Who pays for an unavailable part ordered in advance? A quotation that leaves these matters unresolved may appear simple because the difficult decisions have been postponed until someone is already committed.
The date matters too. A quote valid before the harvest may rely on Felix being able to combine the trip with nearby work. If Ada delays acceptance until his schedule fills, the same service can require a separate journey or displace another customer. The changed charge need not express a changed opinion about Ada or the intrinsic worth of the repair. It can reflect a changed set of feasible arrangements. A price is often attached to an opportunity that expires, even when the physical object remains unchanged.
The payment medium allows these differences to be stated without reconstructing a chain of barter each time. Ada can compare the repair with other demands on her funds; Felix can compare this job with others. That convenience makes more choices visible. It can also encourage careless comparison when unlike promises share a number. A cheaper visit without testing under load may not substitute for a more complete service. The accounting unit makes arithmetic possible before it makes judgment sound.
A recorded price also leaves some contributions outside the bill. Ada may prepare the machine, clear the shed, and supply information about its recent behavior. Felix may spend time diagnosing the problem before deciding whether to accept the work. These activities affect the feasibility of exchange even when they are not separately charged. A transaction can therefore move substantial costs between parties while leaving its quoted amount unchanged. Anyone comparing quotations needs to know which participant is expected to perform the unpriced work.
When direct exchange is enough
The obstacles at Morrow Orchard do not make direct exchange inherently foolish. If Felix wants a modest quantity of fruit, is already visiting nearby, and can complete a clearly bounded adjustment, a direct arrangement might involve little trouble. It could spare both parties a separate sale and purchase. The practical judgment depends on the fit between their needs and on the cost of handling the exchange. A device that helps across many transactions can be unnecessary for one well matched transaction.
Even a wider orchard arrangement could work if a cooperative already handles produce deliveries and records members' balances. Felix could accept a claim the cooperative can readily transfer or settle. Yet the cooperative would be supplying services: grading, storage, bookkeeping, collection, and perhaps credit. Its involvement would not demonstrate that these tasks had vanished. It would demonstrate that someone was equipped to perform them at a lower burden than Ada and Felix could manage separately.
There is also a difference between reciprocal help and a payment intended to close an account. Ada might help Felix at another time without the parties specifying an exact equivalence. That can be a durable and worthwhile relationship. Converting every favor into a priced balance might impair what makes it work. Conversely, leaving an ordinary commercial obligation indefinitely undefined can expose one party to demands they never accepted. The relevant question is what kind of relationship the participants intend and whether their expectations correspond.
A community can sustain several forms of coordination at once. Ada may receive unpaid help from a relative, buy parts for cash, owe Felix after a service visit, and exchange a small produce surplus with a neighbor. These arrangements need not be stages on one ladder of development. They answer different problems of scale, familiarity, timing, and enforceable commitment. Their boundaries can shift as circumstances change. A method that works within a known household may become cumbersome when transferred to strangers carrying unrelated obligations.
The unfinished route to payment
Return to the point at which the conveyor still needs servicing. Ada's first offer failed because it placed too much conversion work on Felix. She can improve it by finding a buyer for the fruit, arranging credible delayed settlement, providing something more widely acceptable, or narrowing the task to match what he can readily receive. Each route assigns work and risk differently. No single route is automatically superior regardless of their circumstances.
The order of action is decisive. If Felix must buy a part before Ada's funds arrive, someone finances the interval. If Ada must deliver fruit before the service is complete, someone carries performance risk. If a third party accepts her produce, someone verifies the grade and finds a subsequent buyer. A transaction's apparent simplicity is often the result of these tasks being performed elsewhere. Their absence from the quotation does not mean they have no cost.
Ada can now ask a more useful question than how much fruit will persuade Felix. She can ask which remaining obligation prevents him from accepting the offer. He, in turn, can distinguish unwillingness to work from inability to use the proposed payment. This does not guarantee agreement. It gives each a clearer account of why agreement has not occurred and which change could make a difference.
Felix still needs purchasing power with which to meet his supplier bills. Ada still has fruit whose usefulness will diminish while they negotiate. A quoted amount gives them a common comparison only after they specify how it will be paid and what work it covers. Their next step is to choose a route Felix can use in time, rather than keep adding fruit to an offer that asks him to solve somebody else's selling problem. The number can then describe a repair that both are in a position to undertake.
Notes and Sources
[1-1] R. A. Radford, “The Economic Organisation of a P.O.W. Camp,” *Economica* 12, no. 48 (November 1945), 189–201. Original article scan, accessed 26 October 2026: https://gwern.net/doc/economics/1945-radford.pdf. Eyewitness account of constrained camp exchange; not a universal origin history of money.