The Sound a Market Makes
Haute Lumière · The Embedded Market · Chapter One
Chapter One
The Sound a Market Makes
Haute Lumière · The Embedded Market · Chapter One
Every organisation learns about itself in two ways. Somebody leaves, or somebody speaks.
The first is the signal economists trust. A customer stops buying, a supplier stops supplying, an employee hands in a letter, a member lets a subscription lapse. The number moves. It is clean, it is impersonal, and it arrives whether anyone wanted it or not. The second is the signal people trust, and are often afraid of. A customer writes to say what went wrong. A supplier's engineer rings to say the part will not hold. A nurse tells a ward manager that the call bells are going unanswered. A child at the kitchen table says that something at school is not right. It is messy, it costs the speaker something, and it can be ignored.
In 1970 the economist Albert O. Hirschman gave these two signals their names, exit and voice, and added a third word, loyalty, for the attachment that keeps a person in place long enough for their voice to work. Twenty-six years earlier, the economic historian Karl Polanyi had written a book about what happens to a whole society when the first signal is allowed to replace the second: when land, labour and money are treated as if they were ordinary goods, and the relationships that used to carry information about them are reduced to prices. He called the condition the market had lost embeddedness, and he called his book The Great Transformation.
This book is about those two ideas, and about one claim that joins them. A market, a firm or a household is always held inside something: a web of relationships, laws, customs and obligations. Whether it is held well can be read in a single place. It can be read in what happens when somebody has something to say.
It would be possible to write a book about embeddedness entirely in the past tense. Polanyi wrote about the nineteenth century and the catastrophe that followed it. Hirschman wrote about a declining railway in Nigeria and the logic of firms and states in the 1960s. Both books are old enough to be quoted more often than read.
They are worth reading now for a plain reason. Most of the relationships in modern economic life have been rebuilt, over the past forty years, to carry exit well and voice poorly. A purchase is a click; leaving is another click. A supplier is chosen by an auction that will run again next year. Work is arranged in contracts that are easy to end and hard to speak inside. A customer who has a complaint is routed to a form, and the form has a field for a reference number and no field for what actually happened. None of this was designed to silence anybody. Much of it was designed, quite reasonably, to make exchange faster, cheaper and less dependent on any particular person. What it changed, along the way, was the ratio of the two signals. The organisations most people deal with, and most people work for, now learn about trouble later than they used to, and from the people least able to tell them why.
Polanyi's word for the result was disembedding. Hirschman's description of it was a firm that loses its most demanding customers first, and declines quietly because those who remain cannot or will not speak. Ammanuel Santa Anna's Ecstatic Economics gives the inside view of the same condition, in a sentence about what a healthy organisation does instead: "Vitality-centered organizations cultivate feedback richness — multiple channels, rapid loops, and the psychological safety to deliver and receive honest information without punishment." Every phrase in that sentence is a design choice, and every one of them can be built or neglected.
This book began as five articles in Ecstatic Economics, the Haute Lumière lineage on money and markets read as living systems. The articles traced the two ideas to their sources, set out the fair critiques of each, read them through the house's living-systems lens, worked them through one public case, and joined them to the rest of the series. Those five are here, revised into chapters 2 to 5 and chapter 13, with the repetition between them removed and each keeping the practical tool it ended on.
The rest is new, and it is the reason for the book. Three chapters take public cases further than an article could: the plant at Fremont, California, where General Motors and Toyota ran a factory together and a cord on the assembly line changed what workers were willing to say (chapter 6); the fire that destroyed the only source of a brake part for every Toyota car in 1997, and the network of firms that rebuilt production in days because they had been talking to each other for years (chapter 7); and the hospital in Staffordshire, England, whose patients and families complained for years before a public inquiry found that nobody had listened (chapter 8). Then a field manual, written so a leader can run the whole sequence over thirteen weeks (chapter 9). A chapter on measurement, on what to count and how to read it honestly (chapter 10). A chapter on the hard conversations, with boards, staff and their representatives, suppliers, families and sceptics, with words a person can say aloud (chapter 11). A chapter for one household and its money (chapter 12). A closing reading of the whole (chapter 14), and a workbook that gathers every tool in the book into one place, with a ninety-day plan (chapter 15).
It is not a book against markets. Neither Polanyi nor Hirschman was against markets, and the book takes care to say where a price carries everything that needs to be known and an arm's-length relationship is the right one. It is not a book of investment advice, and nothing in it promises a result. It describes a pattern that research has found in particular places, and it gives a reader the instruments to find out whether the pattern holds where they are.
Who it is for
Haute Lumière · The Embedded Market · Chapter One
It is written for three readers, and most people are at least two of them.
The leader who runs something: a firm, a team, a school, a hospital ward, a practice, a board. For this reader the book is practical. Its question is whether the organisation hears what it is being told, and how to tell. A leader who suspects that they are learning about problems from customers, auditors or resignations rather than from the people closest to the work will find the instruments here.
The buyer and the partner, the person who manages relationships with other organisations: suppliers, contractors, distributors, funders. For this reader the book's centre is chapter 5 and chapter 7, and the protocol that keeps a problem conversation apart from a price conversation.
The household: a person, a couple, a family, deciding which of the dozen relationships that hold their life together deserve to be held closely, and making it safe for the people at their own table to say when something is wrong. Chapter 12 is theirs, and most of the tools in the book have a household form.
The argument moves in four steps.
The first is the origin. Chapter 2 reads Polanyi and Hirschman from the books themselves, corrects three things commonly said about them, and argues that embeddedness and voice are one idea seen from two distances: Polanyi from outside, at the scale of a society; Hirschman from inside, at the scale of an organisation and the people who depend on it.
The second is the critique. Chapter 3 takes the strongest objections seriously, from the historians who showed that Polanyi had trusted a partial report to the researchers who showed that voice is expensive, unevenly distributed and easy to silence, and finds in each objection a better way to use the idea.
The third is the reading. Chapter 4 reads the whole through the house's lens: the embedding as a membrane that holds an organism together and tells it what is happening outside; exit and voice as the out-breath and in-breath of one rhythm; loyalty as a root system. The reading is an instrument for noticing, and the chapter says plainly what it cannot measure.
The fourth is the practice, which is most of the book: three cases, the manual, the measurement, the conversations, the household, and the workbook.
A note on the voice of the book
Haute Lumière · The Embedded Market · Chapter One
Santa Anna's Organizational Soul Work contains a sentence that could stand at the head of every chapter that follows: "Organizational silence is not the absence of speech." Organisations are full of speech. Meetings, reports, surveys, newsletters, town halls. What goes missing is particular: the truth that would have changed a decision, said early enough to change it. The same book puts the cost of that absence in ten words: "The truth you cannot speak becomes the architecture you inhabit."
This book tries to be useful about that architecture. It names its sources. Every figure in it comes from a published study or a public report, and the notes say which. Where the research is thin, or where a source could not be checked word for word, the text says so. Composite examples, assembled from ordinary situations, are labelled as composites in the paragraph where they appear. The living-systems reading is offered as a lens. The tools are offered as tools: each one can be run in an afternoon or an hour, and each says how to read what it shows.
The wider argument behind the book, that an economy is a living system that lives in the body as well as in the ledger, is set out at full length in Santa Anna's Ecstatic Economics, part of the Luminous Developmental Canon. This book is one of its companions, alongside volumes on the commons, on gift and reciprocity, on employee ownership and on gainsharing, and it refers to them where they meet. All of them are kept at hautelumiere.life.
Four scenes recur in the chapters that follow, and it may help to have them in mind from the start.
A railway in Nigeria in the 1960s, losing its freight to lorries, and not improving, because the shippers who cared most about service were the ones who left first.
A cord on an assembly line in California, above each work station, that any worker could pull to stop the line when something was wrong, installed in a plant that had once had a backlog of more than a thousand grievances.
A fire on the first of February 1997, in a factory in Kariya, Japan, that made a small valve every Toyota car needed, with two or three days of stock on hand.
A hospital ward in Stafford, England, where relatives wrote and spoke and were not heard, until a public inquiry found that complaints from patients, relatives and staff had not been listened to, and made two hundred and ninety recommendations.
Each scene is a membrane: the edge where an organisation meets the people it depends on. In two of them the edge carried only exit, and the organisation learned too late. In two it carried voice, and the organisation learned in time. The difference was not intelligence, money or goodwill. It was whether somebody with something to say had a place to say it, and whether anybody had to listen.
That is the sound a market makes when it is held well. Not silence, and not noise. Somebody saying, in time, that something is wrong.
Chapter Two
The Market Was Always Inside Something
Haute Lumière · The Embedded Market · Chapter Two
In the middle of the 1960s an economist travelling for a study of development projects watched a railway decline. The Nigerian Railway Corporation was losing freight to lorries on the roads that ran beside its lines. That much was ordinary. What held the economist's attention was what did not happen next. In the textbook, a firm that loses customers to a competitor feels the loss and improves. The railway lost customers and stayed as it was.
The economist was Albert O. Hirschman, and the puzzle stayed with him for several years. The answer he arrived at, published by Harvard University Press in 1970 as Exit, Voice, and Loyalty, was that the customers who left were the very customers whose complaints the railway most needed. The shippers who cared most about speed and reliability, and who had the means to switch, switched. The ones who remained were the least able to leave and the least likely to make a fuss. Competition had not disciplined the railway. It had removed its most informative critics and left it a quieter, easier decline.
That is a small story about freight. It is also the clearest way into a question this book will spend its length on: what is a market inside of, and what happens to it when it forgets?
Twenty-six years before Hirschman's book, a Hungarian economic historian teaching at a small college in Vermont had asked the same question at the scale of a civilisation. His answer was that a market is always inside something, a web of law, custom, family, land and obligation, and that the nineteenth-century attempt to lift it out, to make it the thing everything else sat inside, was the root of the catastrophe his generation was living through. He called his book The Great Transformation.
This chapter traces both ideas to their sources, corrects three things commonly said about them, and argues one claim of its own: that Polanyi's embeddedness and Hirschman's voice are one insight seen from two distances. Polanyi describes the membrane from outside, at the scale of a society. Hirschman describes it from inside, at the scale of a single organisation and the people who depend on it. Both are describing how a living economy hears itself.
The book's own front matter settles its origin more precisely than most summaries do. The 1944 first edition, printed in New York for Farrar & Rinehart, carries a note that Karl Polanyi "spent the years 1940 to 1943 in the United States where he was on the staff of Bennington College, and lectured widely." His acknowledgments thank, among his Bennington colleagues, Horst Mendershausen and Peter F. Drucker, and they add a detail that deserves to be better known. Drucker and his wife, Polanyi writes, "were a source of sustained encouragement, notwithstanding their wholehearted disagreement with the author's conclusions."
That sentence is a fair portrait of the book's reception ever since. It was written by a man who had lived through the collapse of the Habsburg world, the Vienna of the 1920s and the rise of fascism, and who wanted to know why the liberal order of the nineteenth century had ended in war. The sociologist Fred Block, who has studied the manuscript's development closely, shows that Polanyi's argument changed while he wrote it, and that the finished book contains tensions he did not fully resolve.1 It is worth reading as what it is: one forceful historical argument, written fast, by a careful mind under pressure.
The argument, stated plainly, runs like this. For most of human history, Polanyi says, economic life was organised by reciprocity, redistribution and the household, with markets present but subordinate. In his own words, which can be checked against the first edition, "Economic systems, as a rule, are embedded in social relations; distribution of material goods is ensured by noneconomic motives." The nineteenth century tried something new: a self-regulating market system in which land, labour and money were themselves bought and sold as if they had been produced for sale. They had not been, and Polanyi calls their commodity description "entirely fictitious." Labour is human life. Land is nature. Money is a token of purchasing power created by banks and states.
What followed, in his most quoted passage, was a reversal. "Instead of economy being embedded in social relations, social relations are embedded in the economic system." The market was no longer a room in the house. The house had become a room in the market.
His verdict on that reversal is unsparing, and it is worth reading at full length rather than in the fragment that circulates:
Our thesis is that the idea of a self-adjusting market implied a stark utopia. Such an institution could not exist for any length of time without annihilating the human and natural substance of society; it would have physically destroyed man and transformed his surroundings into a wilderness.
The word that matters is utopia. Polanyi does not say the self-regulating market was a successful system with bad side effects. He says it was an impossible project, and that society protected itself from it before it could be completed. He called that protection the double movement: "the market expanded continuously but this movement was met by a countermovement checking the expansion in definite directions." Factory acts, public health law, tariffs, central banks, trade unions. Some of these protections were wise and some were disastrous. All of them were society pulling the market back inside.
Ammanuel Santa Anna's Ecstatic Economics opens its survey of alternatives on the same nerve. "There is a peculiar arrogance embedded in the modern economic imagination," he writes, describing the assumption that the rational self-maximiser is "the natural, inevitable, and universal form of human economic being." Polanyi is one of the scholars he names in that chapter, alongside Mauss, Sahlins, Graeber and Hyde, as part of the long anthropological correction to that assumption.
Embeddedness has become a word people use without reading the book that made it famous. Three errors are common enough to correct here.
The first myth is that Polanyi said markets were new. He did not. He says markets existed in almost every society he studied, often for long periods. His claim is about their rank. "Never before our own time," he writes, "were markets more than accessories of economic life." A market town, a fair, a merchant guild: all present, all governed by custom and authority. What was new was the attempt to make the market the governor rather than the governed. A reader who comes away thinking Polanyi was against markets has read the summary, not the book.
The second myth is that the free market arose by itself and regulation was imposed on it later. Polanyi argues close to the opposite, and he puts it in one of the most compact sentences in economic history: "Laissez-faire was planned; planning was not." The free market in labour, land and money required deliberate acts of state: the New Poor Law of 1834, the enclosure acts, the gold standard. The protective countermovements, by contrast, arose piecemeal, from many quarters, often from people who agreed on nothing else. Whether every detail of that history holds is the subject of the next chapter. The structure of the claim is clear, and it reverses the usual story.
The third myth is that embeddedness means one thing. The word owes much of its present currency less to Polanyi than to the sociologist Mark Granovetter, whose 1985 paper in the American Journal of Sociology used it to mean something narrower: that economic action is shaped by concrete networks of personal relations, trust and obligation, in modern economies as much as ancient ones.2 Granovetter thought Polanyi and his followers had overstated how embedded older economies were and how disembedded modern ones had become. The sociologist Greta Krippner later showed that the field had come to use one word for two different ideas, a society-wide claim about institutions and a network-level claim about relationships, without always noticing.3 Both are useful. They answer different questions, and this book keeps them apart.
The question that joins them is the one a leader actually faces: what is this market, this firm, this household budget, held inside, and is the holding healthy?
Hirschman's three words
Haute Lumière · The Embedded Market · Chapter Two
Hirschman's book is shorter than Polanyi's and gentler, and it began as a problem in economic theory rather than a history. Its subtitle names its scope: Responses to Decline in Firms, Organizations, and States. Every organisation, Hirschman starts, slips. Quality falls, service decays, a party loses its way. The interesting question is how the slip is noticed and reversed.
He describes two mechanisms. Exit is the economist's mechanism: the dissatisfied customer or member leaves, revenue or membership falls, and management is warned. Voice is the political mechanism: the dissatisfied person stays and says so, through complaint, petition, protest or argument, and management is told what is wrong. Exit is clean, impersonal and quick. Voice is messy, costly and informative. Exit tells you that something is wrong. Voice can tell you what.
The railway story is the hinge. Where exit is easy for the most demanding customers, it can drain away exactly the voice an organisation needs, and leave behind a clientele too captive or too tired to speak. Hirschman's later reflections, published in 1980, extend the point to public services, where the families most able to leave a failing school are often the ones who would have pushed hardest for its repair.4
Loyalty is the third word, and the one most often misread. It is not, in Hirschman's account, passive devotion. It is the attachment that holds exit back long enough for voice to work. A loyal member is someone who stays a little longer than the arithmetic says they should, because they believe the organisation can recover and that their voice may help it. Loyalty, in this reading, is the reserve that makes voice possible. Without it, every disappointment is a departure.
Read beside Polanyi, Hirschman's three words describe the inside of the embedding. A market embedded in society is one in which voice has somewhere to go: a court, a parliament, a union, a customer who is known by name, a supplier who is called rather than replaced. A market lifted out of society is one in which exit is the only signal left. Prices move, people leave, and nobody says why.
Santa Anna's book names that loss in a sentence about the commons that could stand as a gloss on both thinkers. Of enclosure he writes: "When shared resources become private property, the relational web of mutual obligation that governed their use is severed. The resource becomes a commodity. The community becomes a market." A community can speak. A market, in the narrow sense, can only clear.
The claim of this chapter is that embeddedness and voice are one idea at two distances, and it is worth testing rather than asserting.
Polanyi's double movement is, at the scale of a society, an enormous act of voice. Workers, farmers, landowners, physicians, even bankers, each in their own interest, refused to let the market treat their lives, their land or their currency as mere commodities, and said so through every institution they could reach. Where that voice found working channels, the countermovement produced labour law, public health and central banking. Where it did not, Polanyi argues, it produced the authoritarian movements of the 1930s, which offered protection at the price of freedom. Voice that has nowhere to go does not disappear. It finds a worse door.
Hirschman's organisation is a market society in miniature. It has members who can leave and members who can speak, and its health depends on the balance. Too little exit and it becomes what he called a lazy monopoly, safe from complaint because nobody can go anywhere. Too much exit and it loses the people who would have told it the truth.
Santa Anna's book offers a third description of the same mechanism, written from the inside of a person rather than a society or a firm. Of the healthy organisation he writes that "Vitality requires agency," and of the unhealthy one that "Organizations that suppress feedback — through hierarchy, blame cultures, or information hoarding — create the economic equivalent of a body that can't feel pain: they lose the ability to adapt and self-correct." A body that cannot feel pain is a precise image for a market that hears only exit. It can be badly hurt and keep walking.
In the house's Luminous 100, the upgrade of Stafford Beer's Viable System Model describes the same structure as a nesting of living wholes: "holons all the way up and all the way down, each simultaneously autonomous and embedded." That phrase is this book. A firm is autonomous enough to decide, and embedded enough to be told when it is wrong. Lose either quality and it stops being viable. Chapter 4 returns to this reading in full.
What the founders cannot show
Haute Lumière · The Embedded Market · Chapter Two
It would be easy to make these two books carry more than they can. Three limits are worth naming at the start.
The Great Transformation is a historical argument, not a measurement. Its account of the English Poor Law relied on sources that later historians found to be partial, as the next chapter sets out. Its larger claim, that a fully self-regulating market is impossible, cannot be tested directly, because by Polanyi's own argument no society ever completed one.
Exit, Voice, and Loyalty is a framework, not a finding. It tells you which questions to ask, and later researchers have spent fifty years testing its predictions, with results that are encouraging in places and complicated in others. One of the best tested is in the labour market, where Richard Freeman found in 1980 that union members, who had a formal voice channel through grievance procedures, quit their jobs significantly less often than comparable non-members.5 That is one setting, a particular era of American labour, and a correlation analysed with care. It is evidence for the mechanism, not proof of it everywhere.
And neither book tells a leader what to do on Monday. Polanyi wrote about civilisations. Hirschman wrote about the logic of decline. The finer points of implementation, how to tell whether your own organisation is hearing its voice or only counting its exits, are the work of this book, and they begin with a simple audit.
Every organisation and every household sits inside something. The useful question is not whether, but which relationships are carrying information and which have been reduced to price.
Start with the three things Polanyi called fictitious commodities, because they are the three places where a market most easily forgets what it is holding. Labour: the people who work with you. Land: the places and materials you depend on. Money: the credit, capital and cash flow that keep you running. For each, ask whether the relationship is one in which the other party can tell you something, or one in which they can only leave.
A supplier you have dealt with for twelve years, who calls when a part is about to go short, is embedded. A supplier chosen by reverse auction every quarter, who will never call, is not. An employee who raised a concern last month and saw it acted on is embedded. One who has stopped raising concerns is giving you Hirschman's quietest signal, the one that comes before exit. A household that knows its bank manager, its landlord or its neighbours well enough to ask for time is embedded. One that meets every institution through an app is mostly not.
None of these is good or bad on its own. Granovetter's work and the research that followed it show that embedded ties carry trust and information, and that they can also carry obligation, favouritism and blindness.2 The point of the audit is not to maximise embeddedness. It is to know where your voice channels are, which are working, and where you are learning about trouble only when someone leaves.
Two finer points, learned the hard way by organisations that have tried this.
First, count silence as data. An embeddedness audit that records only complaints will show the healthiest numbers in exactly the places where voice has already died. The railway had very few complaints in its last years. That was the problem.
Second, watch who exits first. Hirschman's sharpest prediction is that the most quality-sensitive members leave earliest. If the customers, staff or partners you would least like to lose are the ones going, your exit signal is also a voice signal, and it is telling you about quality before your averages do.
For a household the same audit is smaller and just as telling. Where does the family learn that something is wrong: from a conversation, or from a bill? Santa Anna's High Frequency Workbook puts the household version of Polanyi's point in seven words: "Every financial decision is a values decision." A budget that has become only arithmetic has, in a small way, been disembedded.
Go back to the station. The lorries are on the road beside the line, carrying the freight that used to travel by rail. The railway's ledgers show the loss, and the loss is real. What the ledgers cannot show is the conversation that never took place: the shipper who would have walked into the office and explained precisely what was wrong with the schedule, and who instead made one phone call to a haulier and was gone.
Polanyi would have recognised the scene. A market relation had replaced a social one, and the information that used to travel inside the relationship now travelled only as a price. Hirschman would have recognised it too, because he was standing in it. Santa Anna's book, writing of the gift economies Polanyi studied, gives the reason in one line: "An economy that destroys relationships in order to produce commodities is destroying wealth in order to create the illusion of wealth."
The railway's real loss was not the freight. It was the voice that had travelled with it.
An afternoon's work for a leader, an hour's for a household. Use a single sheet of paper, landscape, with five columns.
Reading the sheet. Lines with a recent voice, a named channel and no exits are working embeddings; leave them alone and thank the people in them. Lines marked exit only among your most important relationships are where you are learning about trouble last. Repeat the audit in ninety days, and compare the second column. The measure of progress is not fewer complaints. It is more lines where someone told you something in time.
Your page — run the The Embeddedness Audit here, dated.
Chapter Three
Where Embeddedness Goes Wrong
Haute Lumière · The Embedded Market · Chapter Three
In 1832 the British government appointed a Royal Commission to inquire into the Poor Laws, the parish system that had relieved the poor of England for more than two centuries. Its report, published in 1834, was among the most influential documents of the century. It described a countryside demoralised by relief: labourers paid from the rates whatever their wages, employers cutting pay because the parish would make up the difference, families multiplying on public money. The report's diagnosis produced the New Poor Law of 1834 and the workhouse.
Karl Polanyi built a chapter of The Great Transformation on this history. In his account, the allowance system that magistrates in Berkshire introduced at Speenhamland in 1795 prevented a free market in labour from forming for a generation. "During the most active period of the Industrial Revolution, from 1795 to 1834," he writes, "creating of a labor market in England was prevented through the Speenhamland Law." The allowance, in his reading, was an attempt to keep the economy embedded that went badly wrong, and its collapse cleared the ground for the labour market.
In 1963 the economist Mark Blaug published a paper with a title that said what it meant: "The Myth of the Old Poor Law and the Making of the New."7 Blaug went back to the evidence the Commission had gathered and argued that it had been collected to support a conclusion already reached. The allowance system, he found, was less widespread than the report implied, often a response to low wages and seasonal unemployment rather than their cause, and much of the damage attributed to it could be explained by other things that were happening at the same time. Later historians have argued about the details since. The sociologists Margaret Somers and Fred Block, writing in 2005, traced how the Commission's story of perverse incentives became a template for welfare debates two centuries later.8
Polanyi had taken the Commission's account largely at face value. That is the first fair critique of his book, and it is a serious one.
This chapter sets out the strongest objections to both founding texts of this book, and to the way their ideas are used now. Each is stated as fairly as the evidence allows. Each is followed by what a careful reader can do with it, because a critique that leaves the idea unusable has usually been aimed at the wrong target.
The Speenhamland chapter is the best-known case, but the wider objection is that Polanyi wrote history in the service of an argument. His account of pre-market societies leaned on anthropology from the 1920s and 1930s that later fieldwork complicated. His picture of a single nineteenth-century attempt to create a self-regulating market compresses a long, uneven and contested process into a clean story.
The objection lands. It does not, however, reach the structure of the argument. Polanyi's central claims, that land, labour and money are not produced for sale, that treating them as if they were requires deliberate institutional work, and that societies resist the treatment, do not depend on Speenhamland being what the 1834 report said it was. Somers and Block's reading is, if anything, an extension of his point: the Commission's story was itself an act of institution building, a narrative that helped make the labour market by describing the alternative as a failure.8
The way through is to read Polanyi as a framework for asking questions, and to check every historical example he gives against later scholarship before relying on it. For a leader, the practical form of this is simple. When someone tells you that an older arrangement, a guaranteed hours scheme, a long-standing supplier relationship, a mutual structure, "failed", ask who wrote the report that says so, and what they were trying to justify. Blaug's paper is a model of the question.
The second objection is conceptual. If every economy is embedded, as Polanyi sometimes seems to say and as Fred Block argues he came to believe, then the word stops distinguishing anything.9 If only pre-modern economies were embedded, the claim is historically doubtful, as Mark Granovetter argued in 1985 when he wrote that modern markets are also shot through with personal ties and that older ones were more calculating than the anthropologists allowed.10 Greta Krippner's review of the field found that sociologists had come to use embeddedness as a way of studying everything around the market while leaving the market itself unexamined, as if it were the one thing that was not social.11
This is a fair hit, and the house takes it seriously because it applies to the house's own language. A word that can describe everything explains nothing.
The way through is to use embeddedness as a matter of degree and of channel, not as a property something has or lacks. The useful questions are specific. Through which relationships does information about quality travel? Which of those relationships can carry a complaint, and which only a price? How quickly does a problem raised at the edge reach someone who can act? Those questions have answers, and the answers differ between two firms in the same industry. That is the version of the idea this book uses, and it is closer to Hirschman than to Polanyi.
Ammanuel Santa Anna's Ecstatic Economics holds itself to the same discipline when it draws on the traditions Polanyi studied. Of gift economies, it says plainly that they "were not utopias", that the potlatch "could be coercive", and that gift obligations "could create hierarchies of debt and dependence." Then, in four words: "We are not romanticizing." An embedded economy is not automatically a kind one. The question is what the embedding carries.
Three: embeddedness can suffocate
Haute Lumière · The Embedded Market · Chapter Three
That leads to the third objection, which is empirical and comes from inside the sociology that made embeddedness popular. In a study of twenty-three apparel firms in New York, published in 1997, Brian Uzzi found that firms whose relationships with their partners were embedded, built on trust, fine-grained information and joint problem-solving, did better than firms that dealt with everyone at arm's length. But he also found a threshold, beyond which the benefits of embeddedness reversed. Firms that were embedded with too few partners, or too exclusively, became vulnerable: closed to new information, dependent on a small circle, exposed when a key partner failed.12 He called it the paradox of embeddedness.
Every leader who has run a family firm, a long-standing partnership or a team that has worked together for a decade will recognise it. The relationships that carry trust can also carry blindness. The supplier you have known for twenty years may be the last to tell you that a competitor has a better process, because neither of you is looking.
The way through is to treat embeddedness as a portfolio, not a virtue. Uzzi's finding points to a mix: some deep relationships that carry trust and fine information, and some arm's-length relationships that keep you exposed to what the wider market knows. The practical test is whether any of your deep relationships would tell you something that embarrassed both of you. If none would, the depth has become comfort.
The fourth objection is to Hirschman, and it is the one leaders most need to hear. Voice costs something. It takes time, it takes standing, and it carries risk. The person who complains may be labelled difficult. The employee who raises a problem may be blamed for it. The member who speaks up may be the first to be let go.
The research on organisational voice has spent thirty years documenting these costs. Elizabeth Morrison and Frances Milliken, writing in 2000, described organisational silence as a collective state, not an individual choice: a climate in which people share the belief that speaking up is futile or dangerous, and stay quiet together.13 James Detert and Ethan Burris, studying 3,149 employees and 223 managers in a restaurant chain, found that the leader behaviour most consistently related to employees offering improvement ideas was managerial openness, that the relationship ran through employees' sense of psychological safety, and that it was strongest among the best-performing employees.14 That last finding is Hirschman's railway in a restaurant. The people with the most to say are the ones most attentive to whether it is safe to say it, and among the first to stop.
Santa Anna's canon work Organizational Soul Work names a form of this that is easy to miss because no one chooses it. He calls it structural silencing: meeting formats with no space for reflection, feedback forms with no open-text field, decisions that move faster than dissent can form. "Structural silencing is particularly insidious because it appears neutral," he writes. "No one is deciding to suppress truth. The structures simply don't create space for it."
The way through is to lower the price of voice deliberately and to measure whether you have. Detert and Burris's study suggests where to start: what leaders visibly do with what they are told matters more than how inspiring they are. A leader who acts on one small concern within a week, and says so publicly, has reduced the cost of the next concern. A leader who thanks people for raising issues and then does nothing has raised it. The same applies to a household. A child who mentions a worry and is met with a lecture learns the price of voice in that house.
There is a finer point here that leaders often miss. The price of voice is not set by the leader's intentions, which are invisible, but by the last three or four occasions on which someone spoke, which everyone remembers. A single public dismissal of a concern can raise the price for a year. That is why the tool at the end of this chapter looks backwards at what actually happened to concerns, rather than forwards at what the leader hopes will happen next time.
Hirschman's loyalty holds exit back long enough for voice to work. The objection is that the same attachment can be used to hold people in place after voice has stopped working, or to keep them from using it at all. "We're a family here" is a sentence that can mean you can tell us anything or you would not betray us by complaining. Organisations that invoke loyalty most often are not always the ones that reward voice most.
Later researchers made this concrete by adding a fourth response to Hirschman's three. In 1983 Dan Farrell, studying how employees respond to dissatisfaction, found a pattern that was neither leaving, speaking nor staying in faith: neglect, the slow withdrawal of effort and care by people who have stayed but stopped trying.15 Caryl Rusbult and colleagues, testing the expanded model in 1988, found that high satisfaction and high investment in a job encouraged voice and loyalty, while better outside alternatives encouraged exit and, perhaps surprisingly, voice as well.16 People who could leave were more willing to speak, because speaking cost them less.
That finding turns the loyalty objection into an instruction. An organisation that makes leaving very hard, through non-compete clauses, punitive vesting, or sheer dependence, may reduce exit. It does not thereby get voice. It is as likely to get neglect.
The way through is to stop treating loyalty as something the organisation is owed and to treat it as something it earns by honouring voice. In practice that means two things a leader can do. Make it safe to leave well: generous notice, decent references, no punishment for departure. And make it visible that staying and speaking changes things. People who know they could go, and who see their concerns acted on, are the ones who stay and tell you the truth.
Six: the double movement can go to dark places
Haute Lumière · The Embedded Market · Chapter Three
The last objection is the gravest, and Polanyi made it himself. Society's countermovement against the market, in his account, did not always produce wise protection. In the 1930s it also produced fascism, which promised to protect people from the market by abolishing their freedom. A theory that celebrates society pulling the market back inside has to say what kind of society, pulling in what direction.
Santa Anna's book holds the same line in its chapter on governance, and it is the right warning to carry into any application of this book. "Policy without consciousness change," he writes, "produces bureaucratic structures that reproduce the old patterns in new forms." A protective structure built by people in fear tends to protect fear. Re-embedding the market is not automatically liberating. It can mean capture by incumbents, protection for the comfortable, or a closed circle that calls itself a community.
The way through is to judge every re-embedding by whether it widens voice or narrows it. A licensing rule that protects patients and lets them complain is a re-embedding worth having. One that protects practitioners from competition and from complaint is a closed shop. A supplier partnership that shares information both ways is embedded. One that locks a smaller firm into dependence on a larger one is a hold. The test is Hirschman's, applied to Polanyi: after the change, can the people inside it still say what is wrong, and does anyone have to listen?
Fairness runs both ways. Blaug's revision of the Poor Law history weakens one of Polanyi's examples. It does not show that labour can be treated as an ordinary commodity without consequence, and nothing in the research since has shown that. The voice literature shows that speaking up is costly and unevenly distributed. It does not show that organisations do better without it, and Freeman's finding that grievance channels are associated with lower quit rates points the other way. The paradox of embeddedness shows that relationships can become too close. It does not show that arm's-length dealing is better on average, and Uzzi's own firms suggested the opposite.
Each critique, read carefully, narrows the idea and makes it more usable. That is what good criticism does to a living idea. It is also, as chapter 4 argues, how a living system stays alive.
Go back to 1834. The Commission had a question, and it gathered a great deal of evidence. What it did not do was listen to the labourers whose lives the allowance system touched, except as objects of the inquiry. The voice that might have complicated its story was not in the room. Blaug's correction, a hundred and thirty years later, was in part an act of returning that missing voice to the record.
Every organisation writes its own reports of 1834: the exit interview summarised for the board, the engagement survey reduced to a single score, the customer churn analysis that lists reasons nobody was asked. Santa Anna's Organizational Soul Work describes what a leader can do instead when a hard truth is finally said aloud, and it is harder than it sounds: resist the reflex to explain or fix, and for a moment, simply receive it. For that moment, he writes, "the only appropriate response to a truth that has been suppressed for months or years is to let it land."
The critics of Polanyi and Hirschman have, between them, let a great deal land. The ideas are better for it.
A tool for a leader of a team of any size, or for a household. It measures what it costs, in practice, for someone to tell you something is wrong. Allow one hour to set it up and ten minutes a month to keep it.
Reading the ledger. Average cost per concern should fall over three months. The number of concerns may rise; that is the measure working, not failing. If the missing voices start to appear, you have lowered the price of voice where it was highest. If cost stays high and concerns fall, you are moving toward silence, and the next exits will tell you what the concerns would have.
Your page — run the The Voice Cost Ledger here, dated.
Chapter Four
The Membrane That Listens
Haute Lumière · The Embedded Market · Chapter Four
Under a microscope a single cell looks like a small, self-contained world. It is not. Everything that keeps it alive happens at its edge. The membrane around it is a few molecules thick, and it does two jobs at once. It holds the cell together, so that what is inside stays organised. And it is constantly letting things through, nutrients in, waste out, signals in both directions, so that the cell knows what is happening around it and can respond. A cell with no membrane dissolves into its surroundings. A cell with a membrane that lets nothing through starves, and dies without knowing why.
That image is the best short description of what this book has been calling embeddedness. Karl Polanyi's argument in The Great Transformation, traced in chapter 2, was that a market needs a membrane of social relations around it: law, custom, obligation, the claims of land and of human life. Remove the membrane, try to make the market the thing everything else lives inside, and the society it serves begins to come apart. Albert Hirschman's Exit, Voice, and Loyalty described, at the scale of one organisation, the two ways that membrane senses the world: people leave, or people speak.
This chapter reads both through the house's own lens, the living-systems reading set out in Ammanuel Santa Anna's Ecstatic Economics and in the Luminous 100. The claim it makes is specific. Exit and voice are not alternatives a market chooses between. They are two phases of one sensing rhythm, like breathing in and breathing out, and a healthy economy needs both, in proportion, all the time. The failures described in chapter 3 are, almost without exception, failures of rhythm: one phase has stopped, or the other has taken over.
A note on method first. The cell is an analogy, and analogies between organisms and economies have a long and uneven history. The house uses them as instruments for noticing, not as proofs. Where the analogy suggests something a leader can check, this chapter says how to check it.
Ecstatic Economics begins where Polanyi ends. Polanyi argued that labour, land and money are not commodities, because none of them was produced for sale. Santa Anna's book takes the next step and asks what happens to the people inside an economy that treats them as if they were.
Its answer is that the economy lives in the body. The book argues that "economic exchange at its best is not a transaction between isolated agents. It is a relational event between nervous systems." That is a strong claim, and the book grounds it in the somatic research it surveys, particularly the polyvagal framework, which describes how human nervous systems shift between states of safety and connection, mobilisation, and shutdown. The book's reading of Polanyi follows from it. A disembedded market is not only a different institutional arrangement. It is an arrangement that keeps the people inside it in a state of alarm, because every relationship has become a transaction that could end at any moment.
The book is careful about how far it takes this. Its chapter on the anxiety that money produces names a pattern it calls relational numbing, in which people for whom every relationship might be financially motivated find that "the nervous system learns to dampen relational signals." It is describing, in the language of the body, what Hirschman described in the language of the firm. A system that can only hear exit stops hearing anything else.
The book's chapter on governance applies the same reading to institutions. It begins with a sentence political science seldom writes: "governance has a somatic dimension." The experience of being governed, it says, "registers in the body before it registers in the mind." A citizen who votes every few years and is otherwise managed has what the book calls "a felt sense of being governed at rather than governing together." That is a description of a membrane that lets very little through. The citizen can leave, by emigrating or disengaging, and can occasionally speak, by voting. In between, nothing travels.
Read with the book's lens, Polanyi's embedding is a membrane with three properties, and each one is something a leader can observe.
It is selectively permeable. A healthy membrane does not let everything through. It lets through what the cell needs and keeps out what would damage it. An embedded market works the same way. Some things are rightly priced and traded freely; others, a person's safety, the integrity of the land, the stability of the currency, are held back from the market's full logic by law and custom. Polanyi's fictitious commodities are the places where the membrane needs to be least permeable. A firm that lets the market's logic run straight through its treatment of people, with hours, safety and dignity all set by the cheapest available bid, has a membrane with a hole in it.
It carries signals. This is the property Hirschman describes. Exit is a signal: a customer leaves, a member resigns, a supplier stops supplying. Voice is a signal: someone stays and says what is wrong. A membrane that carries only exit signals is like a cell that registers only when something has already been lost. A membrane that carries voice registers the problem while there is still time to respond.
It is alive. A cell membrane is not a wall. It is constantly being rebuilt from the inside. The equivalent in an economy is that institutions, relationships and customs have to be maintained or they decay. Santa Anna's book, writing of enclosure, describes what happens when that maintenance stops: "the relational web of mutual obligation that governed their use is severed." The web was not a fixed structure. It was the continuous practice of people who knew each other.
The Luminous 100 makes the same point about stakeholders in the language of strategy. Its upgrade of the conventional stakeholder map, which sorts people by power and interest and assigns each a management strategy, proposes asking instead: "Who are the living beings and systems affected by this decision? What is the quality of our current relationship with each?" It goes further and rewrites the verbs, "replacing "manage closely" with "deepen relationship" and "monitor" with "extend awareness and care."" Read beside Hirschman, that is a description of opening the membrane. A stakeholder who is managed can only exit. One who is in relationship can speak.
The nervous system
Haute Lumière · The Embedded Market · Chapter Four
If the membrane is where the economy meets the world, voice is how the signal travels inward. The book's workplace chapter states the principle plainly: "Vitality requires information." And it describes what happens without it in an image that has already appeared: organisations that suppress feedback become "the economic equivalent of a body that can't feel pain."
The research witness here is Amy Edmondson's work on teams. In a study of fifty-one work teams in a manufacturing company, published in 1999, she found that teams whose members shared a belief that it was safe to take interpersonal risks, to ask a question, admit a mistake or raise a concern, engaged in more learning behaviour, and that learning behaviour was associated with better team performance.18 The finding has been widely replicated in its broad shape, though how large the effect is, and how it varies across settings, remains an active question. It is a description of a nervous system that conducts. A team where people speak is a team where signals reach the place where action is possible.
Santa Anna's Organizational Soul Work describes what happens when a leader opens that conduction deliberately. When a person with positional power names an uncomfortable truth, especially one about their own contribution, he writes, "it sends a signal through the entire organizational nervous system: it is safe to be honest here." The phrase is not loose. A nervous system is a set of pathways along which a signal either travels or does not. A leader's behaviour sets the resistance on those pathways.
Loyalty, in this reading, is the root system. It is the part of the organism that is not visible, that holds it in place through a bad season, and that allows it to draw on reserves when the surface is under stress. Hirschman's loyalty does exactly this. It holds a member in place long enough for voice to work, and it lets an organisation survive a decline that would otherwise have emptied it.
Roots can also hold a plant in poor soil long after it should have moved. That is the loyalty critique from chapter 3, seen from the ground. The Luminous 100's upgrade of the Net Promoter Score names the distinction precisely. The conventional instrument, it says, "measures transactional loyalty but misses relational depth and developmental impact." A customer who will recommend you because switching is inconvenient has shallow roots. A customer who stays because the relationship has changed something for them has deep ones. The upgrade replaces the follow-up question with a better one: "describe the moment when our work most mattered to you." That question measures roots, not inertia.
Here is the chapter's central claim, stated as plainly as possible. Hirschman's book is often read as describing a choice: a dissatisfied member either exits or uses voice. Read as a living system, exit and voice are not a choice between two tools. They are two phases of one rhythm, and the health of a market or an organisation depends on both phases being present, in proportion, over time.
Exit is the out-breath. It is how a living system releases what no longer fits: a customer moves on to a provider who serves them better, a worker leaves for work that suits them, capital moves to where it is more needed. Without exit, a system becomes a lazy monopoly, in Hirschman's phrase, holding on to members who should have gone and protected from the most honest signal of all.
Voice is the in-breath. It is how a living system takes in what it needs to know: a complaint, a warning, an idea from the edge. Without voice, a system releases members without learning why, and the decline Hirschman watched on the Nigerian railway sets in.
The Luminous 100's upgrade of Polarity Management describes the shape of the relationship exactly. Apparent opposites in organisational life, it says, are "breathing rhythms of a single living system," and "the creative power in any polarity lives not in either pole but in the dynamic movement between them." Applied here, the question for a leader is never whether to have exit or voice. It is whether the rhythm between them is working: whether people who leave are leaving for good reasons, and whether people who stay are speaking.
Every failure in chapter 3 is a broken rhythm. The Speenhamland Commission heard the exit of rising poor rates and never took in the voice of the labourers. The restaurant chain in Detert and Burris's study lost the voice of its best people first, which meant its next exits would be its most costly.19 The organisation that locks people in with punitive terms has stopped the out-breath, and gets neglect, a kind of held breath, instead.
Three signs the membrane is failing
Haute Lumière · The Embedded Market · Chapter Four
The reading earns its place only if it helps a leader notice something earlier than they otherwise would. Three signs, each observable without special instruments, suggest a membrane that has stopped listening.
The surprise departure. When someone leaves and the people closest to them say they had no idea, the membrane has been carrying exit without voice. In a healthy system, departures are usually preceded by conversations: a concern raised, a request refused, an alternative explored. When a leader is regularly surprised by who goes, the signals were travelling somewhere else, or nowhere.
The uniform report. When every review, survey or supplier scorecard comes back in the same narrow band, mildly positive and without detail, the membrane is filtering out everything that does not fit the form. Santa Anna's Organizational Soul Work calls this structural silencing, and it is the easiest of the three signs to mistake for good news. Real systems are uneven. A report with no unevenness in it is a report that is not reaching the edge.
The late fix. When problems are routinely found by customers, auditors or regulators rather than by the people who work closest to them, the signal path from edge to centre has broken. Chapter 5 follows one industry in which this pattern cost a great deal.
None of these signs is proof on its own. Each is a reason to look more closely, using the map at the end of this chapter. Taken together, they describe a cell whose membrane still holds it together but no longer tells it what is happening outside. The organism looks intact. It has stopped learning.
The living-systems reading is a lens. It is good at three things. It makes visible what a conventional analysis misses: that silence is a signal, that the most valuable members leave first, that loyalty can be deep or shallow. It turns abstract arguments into things a person can feel: the difference between being governed at and governing together is something most people recognise in their own bodies. And it joins ideas that are usually kept apart, Polanyi's history, Hirschman's theory, the research on psychological safety, into a single picture that a leader can hold.
It cannot measure. The membrane is an analogy, and its permeability cannot be read off a meter. The claims about nervous system states in Ecstatic Economics draw on a body of research whose application to whole economies is still largely interpretive, and the book presents them as a framework for practice rather than settled findings. And the reading does not, by itself, say what a specific organisation should do. That is the work of the next chapter, which takes one public case and works it through.
The book is clear about this limit in its own terms. Its practices are invitations, not prescriptions, and its chapter on governance warns against assuming that any structure, however well designed, works without the people inside it changing too.
Back to the slide. The cell is still there, its membrane a few molecules thick, letting through what it needs and holding back what it does not, registering every change outside and carrying the signal inward. It is not a wall and not an open door. It is a listening edge.
Ecstatic Economics closes its account of the gift economies Polanyi studied with a line about the oldest form of that edge. In those economies, the book says, "Wealth was not something you had. It was something you moved." The membrane is where the moving happens. A market that keeps its membrane alive goes on moving: taking in what it needs to know, letting go what no longer fits, and remembering, through its roots, why it is there.
That is what it means for a market to be embedded. Not walled in, and not dissolved. Listening.
A one-page map of where your organisation or household meets the world, and how it senses what is happening there. Allow ninety minutes the first time. It can be drawn on paper or a whiteboard.
Reading the map. A healthy membrane shows mostly Vs, a few Es on relationships that are rightly arm's-length, and few or no Ss. Silences are the most important marks on the page, because they are the places where voice once worked and stopped. Redraw the map every six months. A map that has more Vs and the same or fewer departures is a membrane that is listening.
Your page — run the The Membrane Map here, dated.
Chapter Five
Two Carmakers and Their Suppliers
Haute Lumière · The Embedded Market · Chapter Five
Somewhere in a car's development there is always a part that does not quite fit. A bracket is a millimetre off. A seal wears faster than the specification predicted. A component that was cheap to make in the design is expensive to make at volume. The engineers at the supplier usually see it first, because they are the ones making it.
What happens next is the whole subject of this chapter. In one kind of relationship, the supplier's engineer picks up the phone to the carmaker's engineer, they talk it through, and the design changes before the problem reaches a customer. In another, the supplier says nothing, because raising a problem is likely to be heard as an excuse, or as an opening for the carmaker to demand a lower price, and because the contract will be put out to bid again next year anyway. The part goes into production as it is, and the problem surfaces later, at a greater cost, in someone else's budget.
The first relationship is voice. The second is a relationship in which the only signal left is exit: the supplier quietly stops investing, or the carmaker changes supplier when the problems become visible. This chapter works the difference through one of the best documented public cases in modern industry. The figures come from published research, and each one names its source. Where the research cannot show something, the chapter says so.
In a 2014 paper in the Journal of Economic Perspectives, the economists Susan Helper and Rebecca Henderson set out the scale of what happened to General Motors. The company had once been regarded as the best-managed firm in the world. Between 1980 and 2009, they write, GM's share of the United States market fell from 46 to 20 per cent, and in 2009 the firm went bankrupt.22
The usual explanation for the decline is cost: high legacy pension and healthcare obligations to retired workers that competitors did not carry. Helper and Henderson do not dismiss those costs. They argue that the explanation is seriously incomplete. GM's share collapsed, in their account, for many of the same reasons that other dominant American firms of the 1960s lost their markets: a failure to understand the nature of the competition it faced, and an inability to respond effectively once it did. At the centre of that failure they place one thing in particular. GM had great difficulty building the relational contracts that modern design and manufacturing depend on.22
A relational contract, in the economists' sense, is an agreement that cannot be fully written down. It rests on both parties' shared understanding of what each will do in situations the written contract does not cover, and on their trust that the other will keep to it. It is, in the language of this book, an embedded relationship. It is where voice lives.
Helper and Henderson's working paper, the longer version of the journal article, sets out what the two kinds of buying looked like in practice.22 American carmakers' relationships with outside suppliers from the 1950s to the early 1990s, they write, were characterised by short-term contracts, usually of one year, arm's-length relationships, and a reliance on as many as six to eight suppliers per part. Purchase decisions were driven overwhelmingly by price and governed by written contracts that specified the part precisely. There was little communication between suppliers and the engineers who designed the parts or the plants that used them. They quote a senior GM purchasing manager, speaking in 1993, in an unpublished interview with Helper: "GM doesn't need to understand the technologies that our suppliers use—we let the market tell us."
That sentence is the whole of chapter 2. A firm that lets the market tell it has chosen exit as its only signal.
The Japanese carmakers, in the same account, did something different. They entered long-term relationships and were much less likely to switch suppliers over small differences in quoted price. For many parts they specified only the exterior dimensions and the performance required, and let a specialist supplier design the part to fit its own production process. Engineers on both sides came to understand each other's products and processes. When a design had to change, which Helper and Henderson note happened to most parts at least once in a model's development, an American supplier often required a renegotiated contract and could charge well for the change; a Japanese supplier was simply asked to make it, and both parties trusted each other to sort things out later. They quote a supplier who sold to several carmakers: "Honda cares about making the part fit the car, while Ford cares about making the part fit the blueprint."
None of this was soft. Toyota's relationship with its suppliers was, in the words of one supplier's manager quoted in the paper, not "a cozy relationship." Toyota pushed suppliers very hard on cost and defects, reduced the business of suppliers who missed their goals, and left the relationship altogether if improvement did not come. It collected more data on supplier performance than GM did. The difference was in how the data was used: as an aid to finding the root causes of problems, alongside judgement and conversation, rather than as a scoreboard for rewards and penalties.
Read with Hirschman's terms, Toyota kept exit and voice both alive, and put voice first. GM, for most of the period, relied on exit.
The figures Helper and Henderson gather from earlier research show what was at stake. Purchased parts, they report, made up about seventy per cent of GM's costs, while direct labour, legacy costs included, was about ten per cent. Other researchers had estimated that supplier contributions accounted for a third of the Japanese carmakers' advantage in the engineering hours needed to develop a new car, and that the defect rates of parts from Japanese suppliers were of the order of one tenth of those from American ones. In a sample of suppliers selling both to Toyota and to an American carmaker, the average supplier reduced defects by fifty per cent for Toyota between 1990 and 1996, and by twenty-six per cent for its largest American customer. The same supplier, in other words, improved nearly twice as much in the relationship that listened.
A second body of research looks inside the network that did the listening. In a study published in 2000, Jeffrey Dyer and Kentaro Nobeoka examined how knowledge moved among Toyota's suppliers.23 Earlier research had suggested that knowledge spread faster inside Toyota's production network than inside its competitors'. Dyer and Nobeoka argued that Toyota had built the processes that made this happen: it gave members reasons to share valuable knowledge openly while protecting them from undesirable spillovers to competitors, it prevented free riders, and it lowered the cost of finding the right knowledge. It did this by creating a strong network identity with rules for participation and entry, and by treating production knowledge as the property of the network rather than of any single firm. They report evidence that suppliers learned faster after joining.
The careful reader will notice what this research does and does not show. It does not show that supplier relationships alone caused GM's decline; Helper and Henderson are explicit that many factors were at work. It does not show that Toyota's model would have worked unchanged at GM, with GM's history and organisation. What it shows is a consistent pattern, documented across independent studies, in which the firm that invested in voice with its suppliers learned faster than the firm that relied on exit.
The most instructive part of Helper and Henderson's account is not the contrast. It is what happened when GM tried to change. Helper surveyed American auto suppliers in 1989 and 1993. By 1989, GM had the highest scores of the three Detroit firms on items such as whether suppliers expected their customer to help them rather than switch to a rival offering a better deal. By 1993, the paper reports, suppliers again felt that GM could not be trusted.
The authors describe this as a problem of credibility. In 1982 GM's executive director for purchasing told a trade reporter, as they quote: "We need to throw off the old shackles of adversarial confrontation and work together in an enlightened era of mutual trust and confidence." In the same article a GM purchasing agent was quoted: "I find it hard to stop thinking that efficient purchasing means having a lot of vendors fighting over a job." Senior managers could announce a commitment to long relationships, Helper and Henderson write, but until the announcement was matched by commitments and incentives at the local level, neither workers nor suppliers appear to have believed that the managers they actually dealt with would keep to it.
That is the finer point of voice that this whole book keeps returning to. A voice channel is not opened by an announcement. It is opened by the last few occasions on which someone spoke and something happened, and it is closed by the last few on which nothing did. A supplier who has been squeezed for a decade does not hear a speech about partnership as an invitation. It hears it as the opening of a negotiation.
What voice cost Toyota
Haute Lumière · The Embedded Market · Chapter Five
It would be misleading to describe Toyota's way as free, and the research does not. Helper and Henderson's account makes the costs visible in its own terms. Engineers on both sides had to spend time learning each other's processes. Design had to be handed to suppliers with only the outline specified, which meant trusting them with work a more controlling buyer would have kept. Design changes had to be made on trust and settled later. Relationships had to be kept through bad years, with business reduced rather than ended when a supplier fell short. Dyer and Nobeoka add the cost of building a network with rules that let knowledge move without being stolen.23
In Hirschman's terms, Toyota paid to lower the price of voice for its suppliers. It made speaking up cheap for them by making it expensive for itself.
That is the finer point a leader most needs from this case. Voice is not costless, as chapter 3 set out. Someone always pays for it. The design question is who pays, and when. In an arm's-length system, the supplier pays to raise a problem, through risk to the relationship, and so problems tend not to be raised. In a relational system, the buyer pays in advance, through time, people and patience, and problems arrive early. The second system costs more at the start and less later. Whether it is worth it depends on how much the buyer depends on what its suppliers know.
For a car, which contains thousands of parts designed and made by other firms, the answer is a great deal. For a business buying office paper, the answer is very little, and an arm's-length market is the right membrane. Part of a leader's judgement is knowing which of their relationships are which.
There is a reason this matters beyond the history of one industry. Arm's-length contracting often looks cheaper at the moment of purchase. Competitive bidding drives the price of a part down. The costs it creates arrive later and elsewhere: in design problems found late, in suppliers who will not invest, in quality failures that reach customers, in the slow loss of the most capable suppliers to customers who treat them better.
That last cost is Hirschman's railway again. The suppliers most able to leave an adversarial customer are usually the most capable ones, who have other customers to go to. A carmaker that squeezes its suppliers hardest may find, over time, that it is left with the suppliers least able to help it, which is exactly the clientele Hirschman's railway was left with. The decline that follows does not appear as a single decision. It appears as a slow drift in quality and speed that no one ordered.
Ammanuel Santa Anna's Ecstatic Economics states the principle behind this in a sentence written about the whole economy, and it fits the supply chain closely: "An economy that destroys relationships in order to produce commodities is destroying wealth in order to create the illusion of wealth." The cheaper contract was the illusion. The relationship that would have carried the warning was the wealth.
The book is equally clear that this is not an argument against business. "The alternative is not anti-business," it says, describing organisations that measure more than one figure. It is business that keeps its relationships able to carry information.
Chapter 2 described Richard Freeman's finding that union members, who had a grievance channel, quit less often than comparable workers without one. The car industry showed the inside version too. Helper and Henderson describe GM's assembly lines of the 1960s and 1970s as places where workers were neither expected nor encouraged to contribute to improving the process, and relationships with local management as actively hostile. Chapter 6 follows what happened when one of those plants was reopened under different rules.
Santa Anna's canon work Organizational Soul Work names what any such channel is up against. "Structural silencing is particularly insidious because it appears neutral," he writes. A supplier review that asks only about price and delivery dates, an employee survey with no open-text field, a quarterly business review that runs out of time before anyone asks what is going wrong: none of these is designed to silence anyone. Each of them does.
A household, working it
Haute Lumière · The Embedded Market · Chapter Five
The same pattern can be seen at the scale of a household, and here the example is a composite, assembled from several ordinary situations and labelled as one.
A family of four has, for nine years, used the same small garage for its two cars. The owner knows the cars, calls when a part is about to need replacing, and once told them, unprompted, that a repair they had booked was not yet necessary. Last year, to save money, they moved to whichever garage had the lowest quote each time, booked through an app. The individual bills were lower. Over the year, two problems that the old garage would have spotted early were found late, and one breakdown cost more than the savings. No one at the new garages did anything wrong. None of them was in a position to tell the family anything, because none of them knew the cars.
This is not an argument that the cheapest quote is always wrong, or that long relationships are always right. It is a small version of the GM story. The family had moved from a relationship that carried voice to one that carried only price, and had lost the information that travelled in the first without noticing until the cost appeared. The same reading applies to a family doctor, a landlord, a bank, or a school. The High Frequency Workbook gives the household version of the principle: "Every financial decision is a values decision." Choosing the cheapest provider each time is a decision about what the household values, and one of the things it gives up is being told.
Most leaders reading this will not be redesigning a global supply chain. They will be responsible for a few dozen relationships that matter: key suppliers, a few major customers, a team, a handful of partners. The GM and Toyota contrast suggests five points of practice, each grounded in what the research describes.
Separate voice from price. A supplier who raises a problem in the same meeting where prices are negotiated will hear every problem as a bargaining chip, and so will you. Helper and Henderson's account of Toyota is of a firm that used hard performance data to find root causes rather than to hand out penalties, which kept the problem conversation from turning into a price conversation. A leader can do the same at small scale: hold the problem-solving conversation separately from the commercial one, and say so.
Pay for the first warning. The supplier's engineer who calls about the bracket is saving you money. Make it visible that this is valued, not punished. A simple rule, that a supplier who flags a problem early is not penalised on price for it, changes what they will tell you.
Keep more than one source, and talk to both. Toyota did not rely on voice alone. Keeping a second supplier keeps exit available as a discipline. The mistake is to treat the second supplier as only a threat. Talk to both.
Ask about the problems you have not heard of. In every supplier review, add one question: What is the thing you are worried about that we have not asked you about? It costs a minute. It is the question that the most capable suppliers are waiting for.
Watch who leaves. If the suppliers or staff leaving you are the ones you would least like to lose, your exit signal is also a voice signal. Find out why, from them, before the next one goes.
Two related books take these points further: employee ownership, which builds voice into the structure of the firm itself, in employee ownership, and the history and evidence of gainsharing, which ties the gains from shared improvement back to the people who make them, in gainsharing. For a value chain reading of where improvement actually happens inside a firm, see the Alpha Paper Alpha Value Chain.
Back to the bracket that is a millimetre off. In one relationship, the supplier's engineer calls. The carmaker's engineer listens, the drawing changes, and the problem never reaches a customer. No one outside the two firms will ever know it happened. Its value appears nowhere in the accounts, because the cost it prevented was never incurred.
In the other relationship, the engineer says nothing. The part goes into production. Months later, warranty claims rise, a recall is considered, and the purchasing department begins looking for a new supplier. The exit signal has arrived. It is accurate, and it is late.
Helper and Henderson's figure, forty-six per cent to twenty over twenty-nine years, is the sum of many such moments, along with everything else that went wrong. It cannot be broken down into brackets. But the pattern is visible in it all the same. A firm that listened to its suppliers learned. A firm that mostly counted them as costs learned only when they left, and by then the knowledge had gone with them.
For any leader responsible for buying from other organisations, from a team lead with three key vendors to a director with a supply base of hundreds. It sets up a voice channel that is kept apart from price. Adapt the word supplier to partner, contractor or provider as fits.
Reading the results. In the first year, expect the number of problems raised to rise. That is the protocol working, not the supply base getting worse. Over two to three years, the measure that matters is whether problems are being found earlier, in design and planning rather than in production and with customers. For a household, the same protocol shrinks to one question asked of the three or four providers you rely on most, once a year: what should we know that we have not asked you about?
Your page — run the The Supplier Voice Protocol here, dated.
Chapter Six
The Cord at Fremont
Haute Lumière · The Embedded Market · Chapter Six
On an assembly line run the Toyota way, a cord hangs within reach of every worker. It is called the andon cord. A worker who sees a problem pulls it, and help comes to that station while the line keeps moving. If the problem cannot be solved in the time the line allows, the worker pulls it again, and the whole line stops.
Read as an object, the cord is a length of rope and a switch. Read as an institution, it is the most concentrated voice channel in modern industry. It gives the person standing closest to a fault the power to halt production for everyone, on their own judgement, at once. The economists Susan Helper and Rebecca Henderson, whose study of General Motors anchors the case in Chapter 5, report an estimate that shutting down the line for a popular model could cost ten thousand dollars in lost profit per minute.25 A firm that hangs that cord above every station is saying something about whom it trusts to speak.
This chapter tells the story of one plant where that cord was hung over people who had been, by every account, at war with their managers, and of the company that owned the building and could not carry what it learned there back home. It is the clearest public case of the argument this book has been making: that voice is not a technique, a tool or a piece of equipment, and that a firm which tries to copy it as one will photograph everything except the thing that matters.
The plant stood in Fremont, California, and General Motors had run it until it closed the plant two years before the joint venture began. As Helper and Henderson report, drawing on the industrial-relations scholars Clair Brown and Michael Reich, it had been one of the least productive plants in the company and was famous for troubled labour relations. Unexcused absenteeism often exceeded twenty per cent. The plant had accumulated a backlog of more than a thousand grievances. A senior union member described the relationship between union and management as an "ongoing war."25
Helper and Henderson describe the wider GM practice of the period in plain terms. Jobs on the line were narrowly defined: a worker would perform the same small set of tasks every sixty seconds for eight to ten hours a day, and was not expected or encouraged to do anything else. Responsibility for improving the process sat with supervisors and engineers. Quality was the job of a separate department that inspected cars as they came off the line. GM's managers, they write, were notorious for believing that blue-collar workers had little to contribute to improving production.25
They quote a worker named Joel Smith, interviewed by the management scholar Paul Adler, describing what that meant on the floor. Workers protected their jobs by working slowly, by refusing to show a neighbour how their own job was done, and now and then by throwing a wrench into the equipment so that it broke down and the repair crew had to come. Management, in his account, answered in kind, with threats and contempt: if you did not do it their way, there were ten more people at the door looking for your job.25
Read with Hirschman's three words, the old Fremont is a textbook case of a system in which voice had died and exit had been blocked. Workers could not leave without losing a well-paid job in a union plant. They could not speak to any effect, because nothing they said changed the work. What remained was the fourth response that later researchers added to Hirschman's three and that Chapter 3 described: neglect, the slow withdrawal of effort by people who have stayed but stopped trying. Sabotage is neglect with its sleeves rolled up. The grievance backlog is the sound voice makes when it is being filed rather than heard: a thousand complaints, each recorded, none of them changing anything.
Ammanuel Santa Anna's Organizational Soul Work has a sentence for what Fremont had become. "The truth you cannot speak becomes the architecture you inhabit." The worker who knew a better way to do the job and kept it to himself was not being lazy. He was living inside a structure where the truth was dangerous, and the structure was built, piece by piece, out of every truth that had gone unspoken.
In 1984 General Motors and Toyota opened a joint venture in the closed Fremont plant. It was called New United Motor Manufacturing, Inc., NUMMI, and Helper and Henderson are precise about why GM entered it: to gain insight into Toyota's capabilities.25
What happened to the workforce is the part of the story that most surprises people meeting it for the first time. Toyota agreed to rehire the entire union hierarchy. When production began, Helper and Henderson report, ninety-nine per cent of the assembly workers and seventy-five per cent of the skilled-trades workers were former GM Fremont employees and members of the United Auto Workers. NUMMI went on to reach levels of productivity and quality comparable to Toyota's plants in Japan.25
Same building. Largely the same people. Same union. A different result. The case is so clean that it closes off several of the usual explanations at once, and Helper and Henderson make the point directly: the successes at NUMMI, and at GM's own Saturn project, make it implausible that GM's troubles came from unique features of Japanese culture or from a uniquely recalcitrant workforce.25 Whatever had changed at Fremont, it was not the workers.
Helper and Henderson describe the Toyota practice in some detail, and its first feature is easy to misread. Toyota's jobs were, if anything, more tightly specified than GM's. Standard work instructions said which hand should pick up each bolt. This was not a looser system. It was a system in which the specification was precise and the worker was also responsible for improving it.25
Each worker was cross-trained to handle six to eight different jobs on the line. Workers were responsible for the quality of the car and for the continual improvement of the process. They took part in teams that looked for ways to make the line faster or more efficient, and they were trained in statistical process control and in experimental design. And each was expected to see quality problems as they occurred and pull the cord.25
Put in this book's language, three things had been done at once.
The membrane was opened at the point of greatest knowledge. The person nearest the fault was given a channel that ran straight to the people who could fix it. Under the old system, information about a defect travelled down the line, through inspection, into a report, and back to engineering, if it travelled at all. At NUMMI it travelled the length of a cord.
The price of voice was lowered deliberately. A worker who pulled the cord was met by a team leader who came to help, not by a supervisor who came to find fault. Helper and Henderson, drawing on Robert Gibbons and Henderson's earlier work, set out what that bargain required on both sides in a table worth describing. The worker's part was to pull the cord when there was a real problem and to offer suggestions for improvement, even suggestions that might make a job redundant. The supervisor's part was to treat a pulled cord as a problem to solve, to implement improvements without automatically cutting jobs, and to accept the authority of work teams over some shop-floor decisions. The ways each side could break the bargain are listed beside it: the worker who never pulls the cord out of fear, the supervisor who punishes a worker for pulling it, the supervisor who cuts the workforce as soon as an improvement reveals a spare person.25
Voice was made useful as well as safe. The training in statistical process control matters here. A worker who can see a problem but cannot describe it in terms an engineer will act on has a voice that is safe and still weak. Helper and Henderson cite John Paul MacDuffie's account of shop-floor problem solving, which found that good process improvement depended on including several perspectives on a problem, on problem categories loose enough to admit what nobody had foreseen, and on a common language for discussing problems.25 A common language is voice infrastructure. It is what lets the edge talk to the centre in words the centre can use.
Ecstatic Economics describes the organisational version of this in a sentence that could serve as a caption for the cord: "Vitality-centered organizations cultivate feedback richness — multiple channels, rapid loops, and the psychological safety to deliver and receive honest information without punishment." The cord is a rapid loop. The team leader who comes running is the psychological safety. The training is what makes the information honest and usable at once.
Why it could not be written down
Haute Lumière · The Embedded Market · Chapter Six
The deepest point in Helper and Henderson's analysis is about why this worked, and it is the point that explains everything that happened afterwards.
The andon system could not be governed by a written contract. It was, they write, very difficult to specify in advance exactly when a worker should pull the cord, or what counted as being an effective team member. Management had to be confident that a worker deciding to stop a line worth thousands of dollars a minute had both the knowledge and the incentive to judge well. Workers, in turn, would only pull the cord if they were confident that the right understanding was in place behind it.25
Economists call an agreement of this kind a relational contract. Helper and Henderson define it as an agreement based on subjective measures of performance that could neither be fully specified beforehand nor verified after the fact, and that was therefore enforced by "the shadow of the future."25 Each side kept its part because both expected to go on dealing with each other, and both knew what would be lost if the understanding broke.
That is the precise sense in which voice is embedded. A grievance procedure can be written down. A suggestion box can be bought. The belief, held by a worker on a Tuesday afternoon, that pulling the cord on this particular doubtful seal will be welcomed and not held against them cannot be written down or bought. It exists only in the relationship, and it is built only by what happened the last few times someone pulled.
What happened next is the part of the story a leader most needs, because it is the mistake leaders most often make.
NUMMI was, from 1984, GM's privileged window into Toyota's methods. Helper and Henderson report that it took General Motors more than two decades to imitate those methods consistently.25 They report, too, a story told by Jeffrey Liker, a professor who had consulted to GM since the 1980s, in a 2010 radio documentary. A GM manager was ordered, from the level of a vice president, to make a GM plant look like NUMMI. He was told to go there with cameras, photograph every square inch, and make his own plant look like the pictures. The manager, Liker recounts, knew at once that this was impossible: you cannot copy employee motivation, or good relationships between union and management, and you cannot even take a photograph of them.25
The photograph is the image this chapter keeps. It captures exactly what a disembedded view of an organisation can see and what it cannot. A camera can record the cord, the layout of the stations, the boards on the walls. It cannot record the belief that makes a worker pull the cord. GM's managers, Helper and Henderson write, appear for a long time to have believed that the essence of Toyota's advantage lay in tools, such as fixtures for changing dies quickly or just-in-time inventory, rather than in the management practices that made it possible to develop and use those tools.25
At GM's Van Nuys plant, managers introduced elements of the practices that had worked at NUMMI. They produced significantly less improvement, and Van Nuys was soon closed.25 The cord, it turns out, can be installed in an afternoon. What sits behind it takes years, and can be undone in a week.
Helper and Henderson give three reasons why GM found these relationships so hard to build, drawing on the economics of relational contracts. Two of them are the finest points of implementation in this book.
The first is credibility. GM's stance towards its blue-collar workforce, they write, had been deeply adversarial, and it was hard both for GM to change its past patterns of behaviour and for its workers to believe they were changing. They give a sharp example. In 1984 the company announced that it wanted to modify its union contract to support teams and joint problem solving, and then a leaked internal memo suggested that it planned to use the new contract to reduce headcount. Through the 1980s, many in the union leadership remained convinced that GM was adopting these methods only to speed up the line and put workers under more pressure.25
Read with Hirschman, this is voice being invited and distrusted in the same breath. An invitation to speak is only as good as the inviter's record. A workforce that has seen a suggestion turned into a redundancy will treat the next invitation as a trap, and it will be rational to do so. Chapter 3 of this book argued that the price of voice is set not by a leader's intentions but by the last few occasions on which someone spoke. Fremont at GM had a thousand such occasions on file.
The second reason is clarity. A Toyota-style relational contract required an immense amount of ambiguous information to be shared before it could work. Everyone involved had to learn which actions counted as cooperation, come to believe those actions were in their own interest, and understand the other side's temptations to defect.25 That is not something a memo can convey. It is learned the way a language is learned, by use, in the company of people who already speak it. NUMMI had Toyota managers on the floor who spoke it. Most GM plants did not.
Helper and Henderson add a point about structure. GM's focus on short-term financial results, its many levels of control and its large size made credibility harder still. Senior management could announce a commitment to long-term relationships, but until that was matched by commitments and changed incentives at the local level, neither workers nor suppliers appear to have believed that the managers they actually dealt with would keep to it.25 They report that a GM manager named Weller, sent to spread what had been learned at NUMMI, found some plant managers responsive and others not, including one who asked him to leave the factory after his presentation. Asked why the chief executive would not remove a manager who resisted a better system, Weller said that it was a big company and did not work that way.25
That sentence is worth holding beside the cord. At Toyota, a worker could stop the whole line. At GM, a chief executive apparently could not stop one plant manager. Voice had been given to the edge in one system and withheld from the centre in the other.
What the case does not show
Haute Lumière · The Embedded Market · Chapter Six
The NUMMI story is often told as a parable, and parables flatten. Four limits keep it honest.
First, the figures in this chapter are reported by Helper and Henderson from other researchers' work: the absenteeism and grievance figures from Brown and Reich, the rehiring percentages from Adler and colleagues and from Brown and Reich, the productivity comparison from the MIT study The Machine That Changed the World. This chapter has not checked those underlying studies independently, and it computes no figure of its own.
Second, Helper and Henderson do not claim that NUMMI's methods alone explain the difference between Toyota and GM, or that the andon cord was the decisive practice. Their argument is about relational contracts as a whole, inside the firm and with its suppliers, and they are careful to present several possible explanations for why GM struggled to build them.
Third, NUMMI was not a utopia. Many in the union leadership, as Helper and Henderson note, suspected that lean methods were a way to intensify work, and that suspicion was not confined to GM's own plants. A system that asks people to improve their own jobs out of existence makes a serious demand. It is only fair if the people who make the improvement share in what it produces and are not the first casualties of it. The supervisor's commitment in Gibbons and Henderson's table, to implement improvements without automatically cutting jobs, is not a courtesy. It is what holds the whole bargain up.
Fourth, the radio documentary quoted for Liker's and Weller's stories is journalism, reported by economists. It illustrates their argument. It does not, by itself, prove it.
What the case does show is a pattern that every other case in this book repeats. The same people, offered a voice that was safe, useful and credible, used it, and the work changed. The organisation that owned the evidence could not carry it home, because it tried to move the visible part and leave the relationship behind.
Most leaders will never hang a cord above an assembly line. Every leader has a version of it: the moment when someone close to the work can stop something, or chooses not to. Six points of practice follow from Fremont.
Give the stop to the person nearest the fault. Ask where in your organisation a problem is first visible, and whether the person who sees it can stop the process, or only report it. A report travels slowly and loses detail. A stop is voice with authority. It need not be literal: the right to delay a release, hold an order, pause a campaign or refuse a patient discharge until a question is answered is a cord.
Meet every pull with help, not an inquiry. The first response to a stop sets the price of the next one. At NUMMI the team leader came to help solve the problem. In your organisation, the person who raises a flag should hear thank you, what do you need before anyone asks whether it was justified. Review false alarms later and privately. A system that punishes false alarms gets silence, and Gibbons and Henderson's table names that defection plainly: the worker who never pulls the cord out of fear.
Never let an improvement cost the improver. If the reward for finding a better way to do a job is losing the job, people will keep better ways to themselves, and Joel Smith's colleagues did exactly that. Decide in advance, and say in writing, what happens to time and people freed by an improvement. Redeploy before you reduce. Where you cannot promise that, say so honestly, and expect less.
Build the common language. Teach people to describe a problem in terms the people who can fix it will use. A short shared vocabulary for problems, a single page for describing one, a habit of naming the evidence, makes voice useful as well as safe. This is training, and it is cheap compared with what it carries.
Match the announcement with the local incentive. A commitment from the top that is contradicted by how a local manager is measured will be believed by nobody. Before announcing that you want people to speak up, find out what happens to a manager whose team raises the most problems. If it counts against them, change that first.
Do not photograph. When you visit an organisation that does this well, resist the temptation to copy its forms, its boards, its meeting rhythms. Ask instead what its people believe will happen when they speak, and how that belief was built. Then ask the same question of your own people, and listen to the answer before you install anything.
Ecstatic Economics has a line on why trust behaves this way inside a firm. Applying the logic of the gift economy to organisational life, it puts the principle in eight words: "trust circulates, and where trust circulates, value flows." The cord carried a signal. The trust behind it carried the value.
The same pattern appears in boardrooms, where the cord is the right of any member to stop a decision and say that something is wrong. Santa Anna's Conscious Governance Models is direct about why that right matters: "This capacity for principled dissent is absolutely essential in governance, where groupthink is perhaps the single most dangerous dynamic." A board on which no director has pulled the cord in a year is not necessarily a board in agreement. It may be a board in which pulling it has become too expensive.
For readers who want to place these practices within the whole of an organisation's design, the Alpha Paper on the Viable System Model reads a firm as a nesting of units, each of which must be able to sense and correct itself: Alpha Viable System Model. The series on employee ownership, employee ownership, takes the bargain at the heart of the cord one step further, into who owns the gains when the people at the line make the work better. And Santa Anna's own Ecstatic Economics, from which this book draws its living-systems reading, is at Ecstatic Economics.
Go back to Fremont. The building is the same one that had a thousand grievances on file. The people at the stations are, almost all of them, the people who used to slow the line down and keep their knowledge to themselves. A worker sees a seal that is seated badly. She reaches up and pulls.
Nothing about that gesture could be photographed in a way that would let another plant copy it. What made it possible was a set of promises nobody had written down, kept often enough that she believed them. The cord was a rope. The voice was the relationship it hung from.
A tool for any leader of a team, department or organisation, and in a smaller form for a household. It finds where your people can stop a process when they see a fault, and what happens to them when they do. Allow two hours for the first inventory and thirty minutes each quarter after.
Reading the inventory. Over a year, the number of pulls should rise and then settle, and the harm the processes cause should fall. A process where pulls stay at zero is not necessarily safe; it may be silent. A process where pulls rise and the people who pull them are thanked is a process with a voice. Repeat the inventory each quarter, and compare step four. The measure of progress is not how few faults you have. It is how early you hear about them, and from whom.
Your page — run the The Stop-Right Inventory here, dated.
Chapter Seven
The Fire at Kariya
Haute Lumière · The Embedded Market · Chapter Seven
On Saturday the first of February 1997, fire broke out at one of the plants of Aisin Seiki, a supplier to Toyota in Aichi Prefecture, Japan, whose works are centred on the city of Kariya. The plant made proportioning valves, known in the trade as P-valves: small brake parts, cheap, unglamorous, and used in every vehicle Toyota made. Aisin, which the management scholars Toshihiro Nishiguchi and Alexandre Beaudet describe as one of Toyota's most trusted suppliers, was the sole source.27
Toyota ran on just-in-time production, which means it held almost no inventory. When the fire stopped the line at Aisin, Nishiguchi and Beaudet report, only two or three days' worth of P-valves was on hand. The machines that made them were destroyed. In the textbook account of supply-chain risk, this is the scenario every lecture warns about: a single source, no buffer, a part in everything, gone overnight. The expected result is weeks of lost production across the whole group.
That is not what happened. "Yet, remarkably," Nishiguchi and Beaudet write, "disaster was averted, and assembly plants were reopened after only two days of shutdown."27
This chapter is about how. The answer is not a contingency plan, a stockpile or a contract clause. It is a set of relationships among dozens of firms that had been built for ordinary days and turned out to carry extraordinary weight on a bad one. The fire at Kariya is the second deep case in this book, and it sits on the other side of the ledger from Chapter 5's account of General Motors. There, a market held at arm's length learned too late. Here, an embedded market learned in hours.
The response came from firms that had no contract to make P-valves and, in many cases, had never made one. Nishiguchi and Beaudet's account names several. Hundreds of employees of Denso, another large member of the Toyota group, worked on P-valve production. About seventy people at Taiho Kogyo did the same. Twenty-five people at Toyota itself joined in, and hundreds more were sent out to Aisin and to other firms in the effort. The first company to complete a working P-valve, by their account, was Koritsu Sangyo, a supplier with three hundred and twenty employees.27
Pause on that last fact. In a crisis that threatened the production of the largest carmaker in Japan, one of the first firms to put a working part on the table was small enough that most of its employees could stand in one room. It had not been engaged to do this. It had, as far as the account shows, no expectation of a large reward for it. It knew enough, and was trusted enough, to try.
The ties among these firms were not all equal or all independent. Toyota, according to the same account, owned about twenty per cent of Aisin, and Toyota made up some sixty-five per cent of Aisin's sales.27 This was a group with real ownership links and real dependence. Those facts matter for reading the case honestly, and the chapter returns to them. But neither ownership nor dependence explains why a small supplier with no stake in Aisin reached for a drawing it had never been asked to read.
The cost to the wider economy was not trivial. Nishiguchi and Beaudet cite an estimate from Japan's Ministry of International Trade and Industry that output in the transportation-equipment industry fell by 8.3 per cent in February 1997.27 The fire was a real shock, and it was felt. What the case shows is how much worse it would have been without the response, and why the response was possible at all.
Nishiguchi and Beaudet put the lesson of the fire in a sentence that turns the usual way of thinking about supply chains on its side: "the relationships among a firm's suppliers are equally important."27
Most strategy treats a supply chain as a set of spokes. The buyer sits at the hub, and each supplier is a line out to it: a price, a quality standard, a delivery schedule, a contract. The relationships that matter are the ones between the hub and each spoke. In that picture, what suppliers think of each other is somebody else's business.
The fire at Kariya shows the rim of the wheel. The firms that saved Toyota's production were connected not only to Toyota but to each other. They knew one another's capabilities well enough to divide up the work of making an unfamiliar part. They shared drawings and know-how. They trusted that the effort would be recognised fairly afterwards, without needing to settle terms first. None of that could have been arranged by Toyota in two days, from the hub. It had to exist already, in the rim.
The economists Jeffrey Dyer and Kentaro Nobeoka, in a study of Toyota's supplier network published in 2000, describe in general terms how such a rim is built. Their abstract reports that knowledge diffuses more quickly inside Toyota's production network than in competing carmakers' networks, and that suppliers learn more quickly after taking part in Toyota's knowledge-sharing processes. Toyota, they write, solved three problems that usually stop firms from sharing knowledge: it motivated members to take part and share openly while guarding against unwanted leaks to competitors; it prevented free riders; and it reduced the cost of finding and reaching valuable knowledge. It did this by building a strong network identity, with rules for participation and entry. Most importantly, in their account, production knowledge was regarded as the property of the network, and the network had developed institutionalised routines that let knowledge flow in many directions among suppliers.28
This chapter uses Dyer and Nobeoka only at the level of that abstract, because the full text could not be checked for this book. At that level they describe exactly what the fire required: many-directional flows of knowledge, a shared identity, and rules that made sharing safe. In Hirschman's terms, they describe an industrial network in which voice had been built into the relationships among the members, not only into the line from each member to the centre.
Reading the fire with exit, voice and loyalty
Haute Lumière · The Embedded Market · Chapter Seven
Hirschman's three words, which this book has carried since Chapter 2, read the case closely.
Exit was available, and nobody used it. In a purely priced market, a fire at a sole supplier is an exit event. The buyer looks for another source, the supplier loses the business, and the relationship ends on the evidence of a single bad day. Toyota could have treated Aisin that way, and other firms could have treated the crisis as Aisin's problem. Neither happened. The first response was not to leave. It was to help.
Voice travelled sideways. The information that made recovery possible did not mainly move up to Toyota and back down. It moved across: from Aisin to the firms that would make the valves, among those firms as they divided the work, back to Aisin as parts were made. A supply chain in which suppliers can speak only to their customer would have been too slow by days. Here, the edge could talk to the edge.
Loyalty held long enough. Hirschman's loyalty is the attachment that holds exit back long enough for voice to work. In the fire, it held more than exit back. It held back the question of terms. Firms committed people and machines before anyone had agreed who would pay for what. That is only rational if each firm believes, from long experience, that the account will be settled fairly afterwards. Loyalty here was not sentiment. It was credit extended in trust, by firms that had seen such credit honoured before.
Ecstatic Economics states the principle beneath all three in a sentence about every economy, not only Japan's: "Every economic act occurs within a web of relationships." The book means it as a description of how economies have always worked. The fire at Kariya is a demonstration of what that web is worth on the one day it is tested hardest.
It is tempting to read the case as a story about Japanese culture, or about one exceptional company, and to conclude that it has nothing to teach anyone else. That reading misses the more useful point. The web that answered the fire was built, and the building is describable.
Dyer and Nobeoka's abstract names the materials: identity, rules for taking part, a shared view of knowledge as belonging to the network, and routines that make knowledge flow.28 Chapter 5 of this book added, from Helper and Henderson's analysis, the relational contract: Toyota's practice of long-term relationships, of letting suppliers earn a reasonable return, and of expecting both sides to sort things out later when a specification changed, rather than reopening the contract.29 Helper and Henderson are careful to note that this was not a cosy arrangement. Toyota pushed its suppliers hard on cost and defects, reduced the share of those who fell short, and left those who did not improve.29 Embedded did not mean soft. It meant that hard standards lived inside a relationship that both sides expected to continue.
Put together, those materials explain the fire. Firms that had shared knowledge for years knew what each other could do. Firms that had seen problems sorted out fairly did not need terms before they acted. Firms that thought of production knowledge as the group's did not hoard the drawings. The response was not improvised from nothing. It was the ordinary practice of the network, run at a speed nobody had needed before.
Ammanuel Santa Anna's Organizational Soul Work gives a warning that belongs beside this case, because it explains why most organisations would not have known whether their own web was there. "Organizational silence is not the absence of speech," he writes. Organisations are full of meetings and memos. What matters is whether the particular truths that count, we can make that part; we have the machine; our people know that process, travel when they are needed. A supply base can be noisy with weekly calls and still silent about the one thing that would save it in a crisis.
The fire at Kariya is a famous story, and it has often been told with more confidence than the evidence supports. Five limits keep this telling within its sources.
First, every fact in this chapter about the fire comes from one source, Nishiguchi and Beaudet's 1998 article in the Sloan Management Review, which drew on interviews and documents gathered after the event. This book has not checked their underlying data. Where they give no total, for example for the number of firms that took part in making P-valves, this chapter gives none.
Second, the Toyota group was not a market of independent strangers. Ownership links and heavy dependence, such as Toyota's stake in Aisin and its large share of Aisin's sales, gave many firms strong reasons to help that had nothing to do with goodwill. A reader should not conclude that loose networks of independent firms would respond in the same way. The case shows what an embedded network can do. It does not show that embeddedness without shared stakes is enough.
Third, the same feature that saved Toyota also created the risk. A sole source with two or three days of stock is fragile. The network's response repaired the consequences of a design choice that a more cautious buyer might never have made. The lesson is not that single sourcing is safe. It is that if you choose it, the relationships around it are your real inventory.
Fourth, Brian Uzzi's study of New York apparel firms, set out in Chapter 3, warns that networks which are too closed become blind to what lies outside them.30 A group that can mobilise in hours around a shared drawing can also share the same blind spots. The fire tested the network's ability to act together, not its ability to see a threat none of its members had noticed.
Fifth, Dyer and Nobeoka's study is cited here at the level of its abstract only. Its detailed account of how Toyota's network routines worked could not be checked for this book, and none of those details is relied on.
The finer points, for a leader
Haute Lumière · The Embedded Market · Chapter Seven
Few leaders run a supply group the size of Toyota's. Every leader depends on a web of others, suppliers, partners, contractors, neighbouring teams, that will one day be asked to carry more than usual. The fire suggests six points of practice.
Map the rim, not only the spokes. List the organisations you depend on most, and then ask a question most supplier reviews never ask: which of them know each other, and how well? A set of suppliers that have never met each other is a set of spokes. On the day one of them fails, nobody else will know how to help.
Introduce your suppliers to one another. Hold one meeting a year, even an informal one, where your most important suppliers meet each other and hear what each does. Dyer and Nobeoka's abstract points to network identity and rules of participation as foundations. A first step towards both is simply that the members know each other's names and capabilities.
Settle accounts fairly after you have been helped. The willingness of firms to act before terms are agreed rests on a record of being treated fairly afterwards. When a supplier or partner goes beyond the contract for you, pay or recognise it in a way they would call fair, promptly, without making them ask. Every such settlement is an investment in the next emergency.
Keep standards hard and the relationship long. Helper and Henderson's account of Toyota is a useful corrective to a soft reading of embeddedness. The firm pushed suppliers hard and dropped those who did not improve. What made it different from an arm's-length buyer was that the pressure came inside a relationship expected to continue, and that it came with help. Hold your partners to high standards, and be the partner who helps them meet them.
Treat know-how as something to circulate. Ask which of your own improvements could be shared with suppliers or partners without real harm to you. A network in which knowledge moves in many directions is one where the knowledge needed in a crisis is already in more than one place. Ecstatic Economics reads the hoarding of information, and its circulation, as one of the clearest signs of whether an economic web is alive.
Rehearse the bad day. Choose one single point of failure in your operations and ask the people who depend on it, and the people who could replace it, what they would do if it failed tomorrow. Listen before proposing anything. Organizational Soul Work gives the posture in a phrase: "listen first, deeply and patiently, before you do anything at all." The answers will show you where your web is real and where it is only a list of contracts.
For readers placing these practices in the whole of a firm's activity, the Alpha Paper on the value chain reads where value is created, inside the firm and across its boundaries: Alpha Value Chain. The history of gainsharing, gainsharing, takes up the question this case leaves open: how the gains from shared effort are returned to those who made them, so that the next time they are asked, they answer.
The case is industrial, and its shape is domestic. Every household runs on a few single sources: one earner's income, one car, one person who knows how the boiler works or where the papers are kept, one grandparent who collects the children on Wednesdays. Most families never draw these out, for the same reason most firms never draw their rim. On an ordinary day the spokes are enough.
The bad week arrives the way the fire did, on a Saturday, without notice. An illness, a lost job, a car that will not start. What decides how the week goes is rarely the family's savings alone. It is whether the people it depends on know each other: whether the neighbour knows the grandparent, whether the friend who could drive knows where the school is, whether anyone besides one tired adult knows what needs doing. A family whose helpers have never met is a set of spokes. A family whose helpers know each other has a rim, and it will recover in days rather than weeks.
The practice is the same at both scales. Introduce the people you rely on to each other before you need them. Settle kindly and promptly with anyone who carries you through a bad week. Let what you know circulate. The tool below works for a supply base and for a street.
Go back to Kariya on that Saturday. The plant is burning. In the account of the fire, a small supplier with three hundred and twenty people on its payroll is about to become one of the first to put a working P-valve on a bench. Nobody has signed anything. A drawing has crossed from one firm to another because both believed it belonged, in a sense that no contract defined, to all of them.
A spreadsheet of that supply chain, drawn the day before, would have shown one supplier for one part, and two or three days of stock. Every line on it would have been accurate. It would have missed the only thing that mattered: the lines it did not draw, between the suppliers themselves. The inventory that saved Toyota was not on any shelf. It was in the web.
A tool for any leader who depends on other organisations, from a manager with five key vendors to a director responsible for a whole supply base. In its household form, it maps the people a family could call on in a bad week. Allow two hours the first time; repeat once a year.
Reading the map. A healthy map has rim lines around every square and bad-day answers that name other spokes. A map that is all spokes, however many there are, belongs to an organisation that will face its first fire alone. Redraw it each year and count two things: squares without rim lines, and bad-day answers that name someone else on the page. The first should fall. The second should rise.
Your page — run the The Rim Map here, dated.
Chapter Eight
The Complaints Nobody Read
Haute Lumière · The Embedded Market · Chapter Eight
In the autumn of 2007 a woman named Julie Bailey sat beside her mother in a hospital in Stafford, in the English Midlands. What she saw on the ward over the following weeks, and what she saw happen to other patients around her, made her do something most relatives in that position do not have the strength for. She complained, and then she kept complaining. When the complaints went nowhere inside the hospital, she found other families who had seen the same things, and together they formed a campaign group with a plain name: Cure the NHS.
This chapter is about what happened to those complaints, and to the many others that came before and after them, and about why a public institution with every formal channel for hearing them did not hear. It is the book's third case, and the one in which the cost of a failed membrane is counted not in market share or warranty claims but in the care of people who could not leave.
That last point is the reason the case belongs in a book about Polanyi and Hirschman. Albert Hirschman's railway lost its most demanding customers to the lorries. A hospital patient, in most cases, cannot take that route. Illness does not wait for a comparison of providers, and an elderly person on a ward at night is about as far from a free market participant as a human being can be. When exit is closed, voice is the only signal left. Mid Staffordshire is what happens when that last signal is also lost.
A word on method before the case. The public record on Mid Staffordshire is very large: two independent inquiries, a public inquiry, regulators' reports, parliamentary debates and a decade of commentary. This chapter draws on a small number of facts from it that are well established and widely reported, and it does not reproduce any estimate of deaths, because the figures that circulated were contested and are not needed to make the argument. Where the record cannot show something, the chapter says so.
The fullest account is the report of the public inquiry chaired by Robert Francis QC, published on 6 February 2013 and commonly called the Francis Report. It was the second inquiry Francis chaired into the trust; the first, an independent inquiry, had reported in 2010. The public inquiry's report made 290 recommendations, addressed to almost every part of the system around the hospital: the Department of Health, the trust and other providers, commissioners, regulators, professional bodies, local authority scrutiny committees, and the bodies that were meant to carry the views of patients and the public.
Among its central findings was one that could stand as the summary of this chapter. The trust, the inquiry found, had failed to listen to the complaints it received from patients, relatives and staff, and as a result serious problems were neither identified nor acted upon.
It is worth reading that sentence slowly, because each part of it matters.
Patients, relatives and staff. The information was arriving from all three directions a hospital can hear from: the people being cared for, the people who loved them, and the people doing the caring. This was not a case of an organisation with no voice channels. The voices existed, and they were using the channels.
Failed to listen. The failure was not that nobody spoke. It was that what was said did not travel to a place where it changed anything.
Neither identified nor acted upon. The consequence of the listening failure was a knowledge failure. The organisation did not know what was happening on its own wards, in a sense that matters for action, because the information that would have told it had been received and then lost.
Julie Bailey's own exchange with the inquiry carries the same point in a single moment. When Francis first visited Stafford, he recorded later, she asked him what his inquiry could bring out that she and the other members of Cure the NHS did not already know. The relatives knew. The knowledge was already in the town. What was missing was a membrane that would carry it in.
Hirschman himself turned his framework to health services. In his 1980 reflections on Exit, Voice, and Loyalty, published in a health policy journal ten years after the book, he surveyed how the idea had been taken up and gave particular attention to public services, where the people most able to leave a failing provider are often the ones who would have pushed hardest for its repair.32 His argument, paraphrased here because the passage could not be checked word for word, was that services of this kind depend unusually heavily on voice, because exit is either unavailable to many users or, where it is available, removes precisely the users whose complaints carry the most information.
Read with that in mind, the structure of the Stafford failure is clear.
Exit was closed to the people with the most at stake. A frail patient admitted through the emergency department does not choose a ward. Many of the patients whose care the inquiry examined were elderly, and many could not speak for themselves. The market signal that disciplines a shop, a customer walking out, was simply not available to them.
Voice was therefore carried by proxies. Relatives spoke for patients. Staff spoke for the standards of their profession. Both are what Hirschman would recognise as loyal voice: people staying and saying what was wrong, because leaving was not a real option for the patient and not, for many staff, a choice they wanted to make.
The voice reached the membrane and stopped. The complaints were received. They entered a process. And then, in the inquiry's finding, they were not listened to in a way that led to the problems being identified or acted on. In the language of chapter 4, the membrane registered the signal and failed to carry it inward.
Exit returned in its most expensive form. When voice fails inside an organisation that people cannot leave, the information does not vanish. It finds a worse door, as chapter 2 put it. At Stafford the doors were the press, the campaign group, the regulators and, finally, public inquiries. That is exit at the scale of a society: the public's confidence leaving the institution, and the state stepping in from outside. It is accurate, and it is very late.
How a complaint dies
Haute Lumière · The Embedded Market · Chapter Eight
The inquiry's report runs to several volumes, and this chapter does not try to summarise its account of how the system around the hospital failed. What is useful for a leader is the general shape of how a complaint dies inside an organisation that has a complaints process. Ammanuel Santa Anna's Organizational Soul Work names the shape precisely in a list of the ways truth is handled when an organisation cannot bear it. One of them is this: "The truth is acknowledged in principle but deferred indefinitely — always valid, never urgent."
That sentence describes, in thirteen words, the most common fate of a complaint in a large organisation. It is not refused. It is logged, acknowledged, answered with a letter, and placed in a queue that has no end. Each complaint, taken alone, is valid. None is urgent. And because each is handled alone, the pattern that would have been visible across fifty of them is never seen by anyone.
Four mechanisms do most of the work, and every one of them can be found in organisations that have nothing to do with health.
The complaint is treated as a case, not as data. A complaints process designed to resolve individual grievances will resolve them, one at a time, and never ask what they have in common. The relative who complains about a mother left without water receives an apology. The pattern of patients left without water is never counted.
The complaint is routed away from the people who could act. In many organisations the complaints function sits beside the line of management rather than inside it. It can investigate and respond; it cannot change how a ward is staffed. The signal reaches a part of the membrane that has no connection to the muscles.
The complaint is read as a threat. An organisation under pressure to meet external targets, financial or otherwise, has every reason to read complaints as risks to be managed rather than information to be used. Chapter 3 set out what this does to the price of voice. Once complaining is treated as hostility, the people who complain are treated as hostile, and the next person thinks twice.
Staff stop speaking first. This is the quietest mechanism and the most dangerous. Relatives, for all the difficulty, are outside the organisation and cannot be disciplined by it. Staff are inside. When a nurse's concern about staffing is met with indifference or worse, she learns what this book has called the price of voice, and so do her colleagues. Luminous Moral Development asks the question that applies to every ward and every office in this condition: "How many people stay silent in the face of wrongdoing because the power dynamics in the room make speaking up feel dangerous?"
None of these mechanisms requires anyone to want patients to suffer. Santa Anna's phrase for this, quoted in chapter 3, is structural silencing: a silence produced by the design of the system rather than the intent of anyone in it. Organizational Soul Work puts the same insight in a sentence that could be carved over the door of any complaints office: "Organizational silence is not the absence of speech." At Stafford there was a great deal of speech. What was absent was hearing.
There is a finer point in the case that deserves its own space, because it is the one most leaders get backwards. The people who keep speaking to a failing institution longest are, almost by definition, its most loyal members. Relatives who return to the ward every day. Nurses who have worked there for years and remember when it was better. These are Hirschman's loyalists: people who stay a little longer than the arithmetic says they should, because they believe the place can recover and that their voice may help.
Loyalty of that kind is a reserve, and like any reserve it can be spent. Each complaint acknowledged and not acted on draws it down. The relative stops writing letters and starts writing to a newspaper. The nurse stops raising concerns and starts looking at job advertisements, or, if she cannot leave, stops caring quite so much, which is the response later researchers called neglect. By the time an institution notices that its loyal voices have gone quiet, the loyalty that produced them has usually gone too.
That is why silence from a group that used to speak is the most important signal in any review of voice. A complaints log that shows falling numbers from a captive group may describe an improving service. It may equally describe a group that has concluded speaking is useless. The only way to tell the difference is to ask, directly and in person, and to listen to the answer without defending anything. Organizational Soul Work calls this the hardest skill in its whole repertoire, and it is: to receive a truth that implicates you without explaining it away.
The response to the Francis Report is still argued over, and this chapter does not try to judge it. One element of it bears directly on the subject of this book. The government accepted the report in general terms and committed to a statutory duty of candour, together with other recommendations that required legislation.
A duty of candour is, in Hirschman's terms, an attempt to reverse the direction of voice. Most voice channels depend on the person who has been harmed noticing the harm and finding the strength to speak. A duty of candour places an obligation on the organisation to speak first: to tell the patient, or the family, when something has gone wrong in their care. It is the institutional form of the deposit described in chapter 13, information given before it is asked for.
Whether such a duty works in practice depends on everything this book has been describing: on whether staff feel safe enough to report what they have seen, on whether the organisation treats the disclosure as information or as liability, on whether the people told have anywhere to take what they hear. A law can require the first sentence of a conversation. It cannot require the listening that should follow it. That is not a criticism of the duty. It is a description of what a membrane needs in addition to rules.
What the case cannot show
Haute Lumière · The Embedded Market · Chapter Eight
Three limits are worth stating.
First, this chapter has used a small part of a very large record. The Francis Report attributes the failures at Stafford to many causes, across many institutions, over several years, and this book's reading, that a membrane received voice and did not carry it inward, is one way of seeing a story with many strands. It is a useful way, because it points to things a leader can check. It is not the whole account.
Second, the case does not show that complaints processes are worthless, or that more complaints would have prevented what happened. The relatives were complaining. The question was never the volume of voice. It was what happened to it once it was inside.
Third, the case is from a public health system in one country, and public services differ from firms in ways that matter. The patients could not leave; most customers can. But the mechanism, voice received and not carried, is the same one the book found on a Nigerian railway, in a restaurant chain and among an American carmaker's suppliers. What differs is the cost of failure, and the cost is highest where exit is least available.
Most readers of this book do not run hospitals. Every reader who runs anything has some version of Stafford's position: a group of people who depend on the organisation and cannot easily leave it. In a firm these are often employees with few alternatives, small suppliers who depend on a single large customer, tenants, members of a scheme, or customers locked in by contracts or by the cost of switching. For each of these groups exit is closed or expensive, and so voice is the only signal that will arrive in time.
Three practical rules follow from the case, and they apply at any scale.
Find the people who cannot leave, and put them first. Most organisations instinctively pay most attention to the customers and staff who could leave, because their departure shows up in the numbers. The Stafford case argues the reverse. The people who cannot leave are the ones whose voice carries the information that will never arrive by exit. A leader should know who they are by name or by group, and should know how their voice reaches the top.
Read complaints as a set, not a queue. At least once a quarter, someone with authority to act should read every complaint received, together, looking for what they have in common. Not a summary prepared by the complaints function; the complaints themselves, or a sample large enough to hear the words. The pattern that is invisible in a case file is often obvious in a pile.
Close the distance between the complaint and the muscle. Whoever receives a complaint should be one step, not five, from someone who can change the thing complained about. If the structure of the organisation makes that impossible, the leader should carry the complaint the remaining distance personally, at least for the ones that matter most.
Conscious Governance Models names the condition that has to hold underneath all of this. Speaking up, it says, cannot be left to policy alone: "It must be cultural — a genuine collective commitment that truth-telling is valued even when it is uncomfortable." A complaints process is a channel. A culture is the willingness to hear what comes down it.
Go back to the ward in Stafford. A daughter is sitting beside her mother, watching something she knows is wrong. She speaks to the nurse, who is overstretched and who may well know it is wrong too. She writes a letter. The letter is answered. Nothing changes.
The information that would have changed things was in the room the whole time. It was in the daughter, and in the nurse, and in the patient, who could not leave. Organizational Soul Work has a sentence for what an organisation becomes when that information has nowhere to go: "The truth you cannot speak becomes the architecture you inhabit." The hospital at Stafford became, for a time, the architecture of every truth its patients, families and staff had spoken and nobody had carried in.
The membrane at Stafford was not missing. It was there, with forms and processes and an address for letters. It received. It did not listen. And for people who could not leave, a membrane that receives without listening is the same as no membrane at all.
---
For the gift relationship that makes speaking first possible, see Ammanuel Santa Anna's sibling book in this series on gift, reciprocity and trust, gift and reciprocity, and the reading of governance and the somatic life of institutions in Ecstatic Economics, Ecstatic Economics. For the Alpha Paper on mapping the people around an organisation, including those who cannot leave, see Alpha Stakeholder Map. More at hautelumiere.life.
For a leader responsible for any group of people who depend on the organisation and cannot easily leave it: patients, residents, tenants, small dependent suppliers, staff with few alternatives, members locked in by contract. Allow two hours the first time and one hour each quarter after.
Reading the review. Over a year, the measure is not fewer complaints. It is shorter paths in step 2, more recurring themes acted on in step 3, more staff voices in step 4, and at least four quarters in which the organisation spoke first. A captive group that hears from you before it has to complain has a membrane that is beginning to listen.
Your page — run the The Captive Voice Review here, dated.
Chapter Nine
Thirteen Weeks: A Field Manual
Haute Lumière · The Embedded Market · Chapter Nine
A leader has read this far and is persuaded. The organisation, or the part of it she is responsible for, hears mostly exit. She has filled in an audit on the train and found more lines marked exit only than she expected. She wants to change it, and she has a quarter.
What she does not have is a sequence. The earlier chapters each end with a tool, and each tool is useful on its own. Used in the wrong order, or all at once, they become a programme: a launch, a slide deck, a survey nobody trusts, and a quiet return to the old rhythm by the summer. Every organisation has watched that happen to some initiative or other, and the people inside it remember. The memory is part of what this work has to overcome, because the price of voice, as the third chapter set out, is set by what happened the last few times somebody spoke.
This chapter is the sequence. It runs for thirteen weeks, one quarter, because a quarter is long enough for a changed habit to be seen and short enough for a leader to hold in her head. It names the roles, the decision at the end of each phase, and what most often goes wrong at that point. It is written for a leader of a team, a department, a small firm or a division. A household can run a smaller version of it, set out at the end of the chapter, and the chapter on the kitchen table takes that version further.
One principle governs all of it. Ammanuel Santa Anna's Organizational Soul Work gives it in eleven words: the living-systems posture is to "listen first, deeply and patiently, before you do anything at all". The first four weeks of this manual contain almost no changes. That is deliberate. An organisation that has learned to hear mostly exit will treat any sudden invitation to speak as a test, and it will be right to. The listening has to be seen to be real before anyone will risk much on it.
The work needs four people, or four functions held by fewer people in a small organisation. Name them in week one and write the names down.
The sponsor. The person with the authority to change how decisions are made in the area covered. Usually the leader herself. The sponsor's job is not to run the work. It is to do three things visibly: act on what is heard, say what was acted on, and protect the people who speak. Detert and Burris's study of 3,149 employees and 223 managers found that managerial openness, what leaders actually do with what they are told, was more consistently related to employees' improvement-oriented voice than transformational leadership was, and that the relationship ran through employees' sense of psychological safety.34 The sponsor's behaviour is the instrument. Everything else measures it.
The convenor. The person who keeps the sequence, books the conversations, keeps the ledger and the notebook, and chases the follow-up. In a team of twelve this may be the leader's deputy. In a firm of three hundred it is a half-time role for a quarter. The convenor must not be the sponsor, because the sponsor's attention will be pulled elsewhere in week six and the sequence will stall if it depends on her diary.
The listeners. Between two and six people, depending on scale, who will hold the voice conversations in weeks three to ten. They are chosen for two qualities: people trust them, and they can keep a confidence. They are not chosen for seniority. In many organisations the best listeners sit one or two levels below the top, close enough to the work to understand what they are told and far enough from the power to be told it.
The second reader. One person, inside or outside the organisation, who has no stake in the result and whose job is to read the record at each gate and disagree with it. The second reader asks the questions the sponsor would rather not: who did we not hear from, what did the numbers not look at, is the improvement we see real or did we change how we counted. The house practice of a second reading exists because the author of a piece of work is the one person who cannot see its blind spots, and this work is no exception.
Write down, before week one ends, what each of the four will do and how much time it will take. A sponsor who cannot give two hours a week for thirteen weeks should not begin. It is better to start next quarter than to start and stop, because a stopped initiative raises the price of voice for a year.
What happens. The sponsor and the convenor complete the Embeddedness Audit from the second chapter for the whole area covered. Every relationship the area depends on goes on the sheet: people, places and materials, money. For each, the second column records the last time that party told you something you did not already know; the third, the channel it came through; the fourth, any exits in the past year; the fifth, one small act that would open a channel.
The audit is done twice, independently. The sponsor completes it from her own knowledge. The convenor completes it by asking: walking round, checking the records, talking to the people who manage each relationship day to day. The two versions are then laid side by side.
What it shows. The difference between the two sheets is usually the first real finding. Where the sponsor believes a channel is working and the convenor finds it silent, the sponsor has been reading a structure as if it were a practice. A monthly supplier meeting that exists on the calendar is not a voice channel if nothing has been said in it for a year.
The gate. At the end of week two the sponsor, the convenor and the second reader meet for an hour. The decision is which relationships the quarter will cover. Not all of them. Choose between five and fifteen lines from the audit: the ones where the relationship carries something Polanyi would call fictitious, people's working lives, the places and materials you depend on, or the credit you run on, and where the channel is marked exit only or none recalled. Those lines are the scope. Write them down.
What goes wrong. The commonest failure here is scope. A sponsor who has just seen how much of her organisation hears only exit wants to fix all of it, and chooses forty lines. Forty lines cannot be served by four listeners in a quarter, follow-up slips, and the people who were invited to speak and then heard nothing are worse off than before. The second reader's job at this gate is to cut the list.
The second failure is defensiveness. The audit will show that some relationships the sponsor is proud of are quieter than she thought. The right response is the one Organizational Soul Work describes for any hard truth: receive it before explaining it. The book is precise about why: "When a truth is named and immediately followed by a solution, the system doesn't fully receive the truth." Weeks one and two are for receiving.
What happens. The sponsor completes the Voice Cost Ledger from the third chapter for her own conduct over the past six months. The last ten concerns raised with her; for each, who raised it, how, how long she took to respond, what she did, and whether she told them. Each concern is scored for cost. The missing voices are listed: the people who report to her, supply her or depend on her who appear nowhere among the ten.
The convenor does the same for each listener, in a short version: the last five concerns each of them received and what happened.
What it shows. The ledger is the sponsor's own record, and it is usually uncomfortable. Its value is that it is specific. "We need to listen more" is an intention. "Six of my last ten concerns took more than seven days to answer, and I told only three of the people what I did" is a finding, and it points at the habit to change.
The gate. No meeting this week. The sponsor chooses one cost from the ledger, the one that appears most often, and commits to changing one habit for the rest of the quarter. The habit is written down and given to the convenor, who will check it at every later gate. A common choice is a rule to acknowledge every concern within three working days, even if only to say when a decision will come.
What goes wrong. Sponsors skip the ledger, or delegate it. It is the one tool in this manual that cannot be delegated, because its subject is the sponsor. A ledger completed by someone else measures the organisation's opinion of the sponsor, which is useful, but it is not the same instrument.
Week four: the map
Haute Lumière · The Embedded Market · Chapter Nine
What happens. The sponsor, the convenor and the listeners draw the Membrane Map from the fourth chapter together, in one ninety-minute session. The circle in the centre is the area covered; outside it, every party in scope. Each meeting point is marked V for a working voice channel, E for exit only, S for a voice that has gone silent. Every E and S on a relationship that carries people's time and safety, land or premises, or the credit you run on is circled.
What it shows. The map turns the audit into a picture the listeners can hold. Its most important marks are the Ss, because they are the places where voice once worked and stopped, and each has a story. The listeners will often know the story. Someone raised something, something happened to them, and others drew the conclusion.
The gate. At the end of week four, the sponsor, convenor, listeners and second reader meet for ninety minutes. Two decisions. First, assign each circled party to a listener. Second, decide which of the circled parties are suppliers or external partners, who will be served by the Supplier Voice Protocol in weeks five to ten, and which are internal, who will be served by listening conversations.
What goes wrong. The map is drawn by the sponsor alone. A sponsor's map shows the membrane as she believes it to be. The listeners' map shows it as it is. They are different pictures, and the difference is the work.
This is the long middle of the quarter, and it is where most initiatives die. Six weeks is long enough for the sponsor's attention to drift, for a crisis to arrive, and for the people who were invited to speak to start wondering whether anyone is still listening.
What happens, with external partners. For each supplier, contractor or partner in scope, the assigned listener runs the Supplier Voice Protocol from the fifth chapter. The conversation is held apart from price, and the separation is stated in writing: nothing said in it will be used in commercial negotiation. The four questions are asked in the same words each time. What is not working well for you in how we work with you? What problem do you see coming that we have not raised? What would you change in our specification, process or timing if you could? What have you stopped telling us, and why? The answers are written down in the partner's words. Within thirty days, each answer is returned to the partner marked acted, not acting with the reason, or investigating with a date.
What happens, inside. For each internal group in scope, a team, a shift, a function, the assigned listener holds a forty-five minute conversation every fortnight with a small group, between four and eight people. The same four questions, adapted. The listener writes down what is said, without names unless the speaker asks for their name to be attached. The convenor collects the notes each Friday.
The weekly rhythm. Every Friday, the convenor brings the week's notes to the sponsor for thirty minutes. The sponsor decides, for each item, act, explain why not, or investigate by a date. The convenor records the decision. Every Monday, the sponsor says, in whatever forum the area already uses, a team meeting, a written note, a short message, what was changed in the past week because someone spoke. One sentence per change. No names, unless the person wants to be named.
That Monday sentence is the single most important act in the quarter. It is the moment the organisation learns whether voice is heard. Detert and Burris's finding that openness, what a leader is seen to do, matters more than inspiration is a finding about exactly this kind of public, repeated, small evidence.34
The gate, week seven. A mid-point meeting of all four roles, one hour. The convenor brings three numbers: how many items have been raised, how many have been closed with a response, and the median days from raising to response. The second reader asks who has not been heard from. The decision is whether to continue as planned, or to change the listener assignment for any party that has stayed silent.
What goes wrong. Three things, in rough order of frequency.
The first is a gap in the Monday sentence. One week the sponsor is travelling, or nothing feels worth reporting, and it is skipped. The organisation notices. A second gap confirms what people suspected. If there is genuinely nothing to report, the sentence says so: nothing was changed this week; these three items are still being looked at and will be answered by Friday. An honest small sentence is better than silence.
The second is the problem that cannot be solved in a quarter. Someone raises a question about pay, or structure, or a long-standing decision that the sponsor cannot change. The temptation is to treat it as out of scope and say nothing. The protocol's answer is the not acting response, with a reason, given promptly. People can accept a refusal they understand. They do not forget being ignored.
The third is retaliation, usually unintended. A manager below the sponsor hears that a concern came from his team and treats the team differently. Morrison and Milliken describe organisational silence as a collective state built on shared beliefs that speaking up is futile or dangerous, and a single visible consequence for a speaker can rebuild that belief across a whole group.35 The sponsor must make it clear, before week five begins, that any adverse treatment of someone who spoke through these channels will be treated as a serious matter, and then act if it happens. Conscious Governance Models puts the requirement in a sentence: "It must be cultural — a genuine collective commitment that truth-telling is valued even when it is uncomfortable."
What happens. The listeners and the convenor redraw the Membrane Map for the parties in scope. Each E or S from week four is checked: is there now a working channel, has the party used it, and did they see what happened to what they said? The sponsor redoes her Voice Cost Ledger for the concerns raised since week three. The convenor prepares the record for the gate: items raised, items closed, median response time, and a list of parties still silent.
What it shows. In most first quarters, the number of concerns raised goes up. This is the most commonly misread result in the whole manual. A rise in concerns is not a sign that things have got worse. It is the measure working: problems that existed all along are now arriving where someone can act on them. The next chapter takes this up at length, because it is where a measurement system most often lies to the people who built it.
What goes wrong. The second map is drawn by the same people who drew the first, with the same hopes, and it shows improvement because they want it to. The second reader's role here is to check three or four marks independently: to ask, directly, a party now marked V whether they would say so.
Week thirteen: the reading and the decision
Haute Lumière · The Embedded Market · Chapter Nine
What happens. A two-hour meeting of all four roles. The convenor presents the record. The second reader presents what she found. The sponsor reads both and makes the decision the whole quarter has been building to: what continues.
Three outcomes are possible, and each is legitimate.
Continue and widen. The channels opened are working; the Monday sentence has become a habit; the median response time has fallen; some silent parties have spoken. The scope is extended to the next set of lines from the original audit, and the rhythm continues with fewer formal gates.
Continue and narrow. Some channels are working and others are not. The work continues on the ones that are, and the sponsor takes a hard look, with the second reader, at why the others failed. Often it is a listener mismatch, a relationship that needed a different person; sometimes it is a relationship that was never going to carry voice and is better held at arm's length.
Stop, and say so. If the work has not taken, because the sponsor could not hold the rhythm, or because a reorganisation or a crisis overtook it, the right ending is a plain statement to everyone who took part: what was heard, what was done, and why the formal work is ending. This is the least comfortable outcome and the most important one to get right. An initiative that fades without an ending teaches people that voice is fashionable. One that ends honestly teaches them that it is taken seriously even when it stops.
What goes wrong. The meeting becomes a celebration. A good first quarter feels like an achievement, and it is one. But the house practice is to read for what the numbers did not look at before reading for what they show. Who was never in scope? Which exits in the quarter were not linked to any voice? Is the fall in response time real, or did the convenor start counting differently in week six? Organizational Soul Work has a sentence that belongs on the wall of this meeting: "The truth you cannot speak becomes the architecture you inhabit."
The formal sequence ends in week thirteen. The rhythm does not. From week fourteen, the sponsor keeps the Weekly Exit and Voice Review from the thirteenth chapter, fifteen minutes every Friday: exits this week and whether the reason is known; voices this week and what was done; the linked pairs, where an exit followed a voice that was not acted on; one call to someone who left; one thing given before it was asked for.
That review is the permanent form of the work. The quarter built the channels. The review keeps them breathing. Santa Anna's Luminous Moral Development makes the case for a small practice kept over a large one launched, in a sentence about personal courage that fits an organisation exactly: "You build it by lifting manageable weights repeatedly."
A household can run the same sequence in a much smaller form, and the chapter on the kitchen table sets out why it is worth doing.
Weeks one and two. One adult completes the Embeddedness Audit for the household's dozen most important relationships: employers, landlord or lender, the doctor, the school, the few providers of things that break, the people nearby who help. A second adult, or an older child, does the same separately. Compare the sheets.
Week three. Each adult completes a short Voice Cost Ledger for the household itself: the last five times a family member, including a child, raised a worry, and what happened.
Week four. Draw the membrane map on one sheet at the kitchen table, together.
Weeks five to ten. One conversation a week, fifteen minutes, at a regular time, in which each member is asked what is not working and what they are worried about that no one has asked. Write it down. By the next week, say what was done.
Weeks eleven to thirteen. Redraw the map, look at the ledger again, and decide together what continues.
The roles shrink. The sponsor is whichever adult holds the household's decisions; the convenor is whoever is better at keeping a list; the listener, for a child, is often the adult the child already talks to; the second reader can be a friend or relative who is asked, once, to look at the sheets and say what they notice.
The thirteen weeks have a shape that is worth seeing in one piece. The first four weeks listen and change nothing, so that the listening is believed. The middle six open channels and close loops, publicly and repeatedly, so that the price of voice falls. The last three read honestly, so that the work is judged by what happened rather than by what was hoped.
That shape is not invented here. It is the shape of every working membrane described in this book: a period of sensing, a period of exchange, and a period of reading what came through. The Toyota supplier network, the andon cord at Fremont, the inquiry at Stafford, told in chapters 5 to 8: each succeeded or failed according to whether signals raised at the edge reached someone who could act and came back with an answer. A quarter is one turn of that cycle. The organisation learns its rhythm by going round it.
Ecstatic Economics states the requirement in one line about organisations that work: "Vitality-centered organizations cultivate feedback richness — multiple channels, rapid loops, and the psychological safety to deliver and receive honest information without punishment." Multiple channels, rapid loops, safety: the three things a quarter can build, one week at a time. Ammanuel Santa Anna's full account is in Ecstatic Economics, and the companion book in this series on gainsharing, gainsharing, shows what becomes possible in an organisation once the people closest to the work are routinely heard. For setting the quarter's few objectives without turning them into targets, see the Alpha Paper Alpha OKR.
One sheet, kept by the convenor, read at every gate.
Reading the plan. At week thirteen, look for four things: the Monday sentence never missed, or missed honestly; median response time lower than at week seven; at least some S marks turned to V; and the number of concerns raised higher than at the start. If all four hold, the quarter worked. If concerns fell while response times rose, the organisation is moving toward silence, and the next quarter begins again at week one.
Your page — run the The Thirteen-Week Plan here, dated.
Chapter Ten
Counting What Speaks
Haute Lumière · The Embedded Market · Chapter Ten
At the end of a first quarter of listening, a convenor brings the sponsor a sheet of figures. Concerns raised: up by more than half on the quarter before. Grievances logged through the formal channel: up as well. Two suppliers have flagged quality problems that nobody had heard about. The sponsor looks at the sheet, and her first reaction is the one almost every leader has: things have got worse.
They have not. Almost certainly, the problems on the sheet were there before. What changed is where they arrived. Last quarter they travelled as silence, then as a resignation, then as a late delivery. This quarter they arrived as words, in a place where someone could act on them. The sheet is the best news the organisation has had in a year, and it looks like bad news.
That is the central difficulty of measuring voice, and it is why this chapter exists. Every instrument for counting what people say has a built-in trap: the measures of health and the measures of trouble are the same numbers, read in different directions. A leader who does not know which direction to read will dismantle a working system at the moment it starts working.
This chapter sets out what to count, how to read it honestly, and what the numbers cannot tell. It draws on four pieces of research already met in this book, and on one sentence from Ammanuel Santa Anna's Ecstatic Economics that is worth holding throughout: "Designing such measurement systems is among the most important — and most difficult — practical challenges the movement faces."
There are three reasons, and they compound.
The first is that silence produces no data. An exit leaves a trace: a cancelled contract, a resignation letter, a lost customer in the ledger. Voice leaves a trace when it is used. Its absence leaves nothing. An organisation that has stopped hearing its people does not see a falling line on a chart. It sees no line at all, and no line looks the same as calm. Morrison and Milliken's account of organisational silence is precisely about this: a shared belief that speaking up is futile or dangerous produces a climate in which people stay quiet together, and the quiet is easily read as agreement.37 Organizational Soul Work puts the same point in eight words: "Organizational silence is not the absence of speech." It is speech that has gone somewhere else, into the corridor, the exit interview, the next employer.
The second is that the numbers move in the wrong direction at first. When a channel opens, the backlog of things nobody said arrives through it. Counts rise. The rise is the measure working. A leader who reads it as deterioration, and a board that reads it the same way, will be tempted to close the channel, or to change how things are counted until the numbers look better.
The third is that the people being counted can see the count. Any measure that matters to a leader becomes, sooner or later, a target for the people it measures. The economist Charles Goodhart observed in the 1970s that a statistical regularity tends to break down once it is used for control, and the anthropologist Marilyn Strathern later gave the idea its widely quoted general form: a measure that becomes a target stops being a good measure.38 A team told that its number of raised concerns is being watched will raise more concerns, or fewer, depending on what it believes the leader wants. Either way, the number stops telling the leader what the team thinks.
None of these makes measurement impossible. Each changes how it has to be done.
Four studies carry most of the weight in this book's account of voice, and each is worth reading for what it counted as well as what it found.
Freeman, 1980. Richard Freeman used several large American labour-market data sets to compare union members, who had a formal voice channel through grievance procedures, with comparable non-members. The measure was exit: quits, permanent separations and job tenure. Union members exited significantly less, and Freeman argued that the grievance system played a major role.39 The lesson for measurement is that he did not count voice directly. He counted exit, and inferred voice from the gap. That is often the most honest route available: voice is hard to see, exit is easy, and a fall in avoidable exit where a voice channel has been opened is evidence the channel is doing something.
Edmondson, 1999. Amy Edmondson studied fifty-one work teams in a manufacturing company. She measured a shared belief, team psychological safety, with survey items; she measured learning behaviour, the asking of questions, seeking of feedback, discussion of errors; and she measured team performance. Psychological safety was associated with learning behaviour, and learning behaviour mediated the relationship between safety and performance.40 The lesson for measurement is that the thing being measured lived at the level of the team, not the individual. A survey score averaged across an organisation would have hidden it. Voice happens in a particular room with particular people. Count it there.
Detert and Burris, 2007. In a restaurant chain, with 3,149 employees and 223 managers, James Detert and Ethan Burris measured two kinds of leadership behaviour and the improvement-oriented voice of the people who reported to each manager. Managerial openness was more consistently related to voice than transformational leadership, the relationship ran through perceived psychological safety, and leadership behaviour had its strongest effect on the voice of the best-performing employees.41 The lesson for measurement is the last finding. If voice falls first among your strongest people, an average across everyone will show the fall late, and small. Count who speaks, not only how much is said.
Morrison and Milliken, 2000. This is a theoretical paper, not a measurement study, and that is its use here. It describes the conditions, a belief among managers that employees are self-interested, that disagreement is bad, that management knows best, under which silence becomes collective.37 The lesson for measurement is that silence has causes upstream of any individual's choice, in what leaders believe and in how decisions are structured. A count of voice that does not also look at those structures will find the symptom and miss the source.
Each of these studies is limited to its setting: American labour markets of one era, one manufacturing company, one restaurant chain, one theoretical framework. Together they do not prove that any particular measurement will work in your organisation. They show which things are worth counting, and at what level.
Six things to count
Haute Lumière · The Embedded Market · Chapter Ten
What follows is a small set of measures, chosen because each can be kept by a convenor with a notebook or a simple sheet, and each answers a question a leader needs answered. They are not a scorecard to be optimised. They are a set of readings to be taken together.
1. Voices raised, by source. The number of concerns, warnings, objections and ideas raised through the channels in scope, recorded each week, broken down by where they came from: which team, which supplier, which group. The total matters less than the distribution. A count that rises overall while one team stays at zero is telling you about that team.
2. Time to response. For each voice, the number of days from its being raised to the person who raised it being told what happened, whether the answer was acted, not acting with a reason, or investigating by a date. Record the median, not the mean, because one long-running item can pull a mean far from the typical experience. This is the measure most directly under the sponsor's control, and the one Detert and Burris's finding points to most clearly: what a leader visibly does with what she is told.
3. Closed loops. The share of voices raised in a period for which the raiser was told the outcome. A voice that is acted on but never reported back is, from the raiser's side, indistinguishable from a voice that was ignored. The loop is not closed until they know.
4. Linked pairs. From the Weekly Exit and Voice Review in the thirteenth chapter: the number of exits in a period that followed an earlier voice from the same person or party that was not acted on. These are the most expensive lines in the record, because they show voice arriving in time and being lost. A falling count of linked pairs is the clearest sign that the rhythm between exit and voice is working.
5. Who leaves first. For every exit in scope, a simple mark: was this a party the organisation would least like to lose? Hirschman's sharpest prediction, met in the first chapter, is that the most quality-sensitive members exit earliest. Detert and Burris's finding about best performers is the same pattern inside a firm. If the marked exits rise as a share of all exits, the organisation is losing its most informative critics, whatever the totals say.
6. Silence, as a count. For every party in scope, the number of weeks since they last raised anything. A party with a long run of silence is not necessarily content. Mark it, and ask. This is the measure the Membrane Map's S symbol exists for, kept as a number so that a lengthening silence can be seen before it ends in an exit.
Ammanuel Santa Anna's book asks for exactly this kind of breadth: "Ecstatic economics requires multi-dimensional measurement that includes financial sustainability and relational quality and ecological impact and somatic vitality." The six measures here are a small, practical piece of the relational dimension. They sit beside the financial figures a leader already keeps, not in place of them.
A count means little without a baseline. Before a quarter of listening begins, the convenor should reconstruct, as far as the records allow, the six measures for the previous two quarters. Most organisations can reconstruct exits and who left from personnel and supplier records. Many can reconstruct some formal grievances and complaints. Few can reconstruct informal voice, because nobody wrote it down, and that gap is itself a finding: a starting count of unknown for informal voice is honest, and better than a guess.
Three cautions about base rates.
Seasons. Exits cluster: after bonus dates, at the end of school years, at contract renewal. Compare a quarter with the same quarter last year, not only with the quarter before.
Small numbers. In a team of twelve, one resignation is eight per cent of the team. A change from one exit to two is not a trend. With small numbers, read the individual cases, not the rates. The point of the count in a small team is to make sure nothing is forgotten, not to compute a percentage.
Changes in counting. When a new channel opens, things that used to be raised informally and never recorded begin to be recorded. The count rises because the recording improved. Note the date any channel or recording method changes, and read every count across that date with care.
The most important rule for using these measures is that none of them should become a target for the people being measured. They are readings for the sponsor and the second reader, not goals for teams.
In practice that means three things.
Do not reward the count of voices. A team rewarded for raising more concerns will raise more concerns, some of them manufactured. A team penalised for raising many will raise fewer, and the real ones will go back to the corridor. The count of voices is read, never rewarded.
Do reward the response. The measures that can safely become commitments are the ones under the leader's own control: time to response and closed loops. A sponsor who commits publicly to answering every concern within three working days has made a promise about her own conduct. If she games it, by answering quickly and emptily, the people raising concerns will see that at once, which is a better check than any auditor.
Watch the measure that nobody is targeting. If the response measures improve while linked pairs and the exits of valued people do not, the response measures are being met in form and missed in substance. Linked pairs and who leaves first are hard to game, because they are recorded after the fact and depend on what other people decide.
Organizational Soul Work names the trap behind this rule. When a truth is named and immediately met with a fix, the book says, the organisation does not fully receive it. A measurement regime can do the same thing at scale: it converts every voice into an item to be closed, and the closing becomes the point. The count goes up, the loop is closed, and nobody has listened. The book's sentence is a fair warning for any convenor: "When a truth is named and immediately followed by a solution, the system doesn't fully receive the truth."
Reading the six together
Haute Lumière · The Embedded Market · Chapter Ten
The measures make sense only in combination. Four patterns are common, and each has a reading.
Voices up, response time down, linked pairs down, valued exits steady or down. The membrane is opening. Problems are arriving earlier and being answered. This is what a good first or second quarter looks like, and it is the pattern most likely to be misread as bad news, because the first number is up.
Voices down, response time up, linked pairs up. The organisation is moving toward silence. People raised things, waited, and stopped. The exits that follow will carry what the voices would have. This pattern calls for the Voice Cost Ledger, done honestly, at once.
Voices up, response time up, closed loops down. The channel has opened faster than the organisation can answer. People are speaking and not hearing back. This is fragile: if it lasts more than a few weeks, it will turn into the second pattern. The remedy is not to discourage voice but to answer more simply, even if the answer is only investigating, by this date.
Every measure steady, and quiet. The most difficult reading. It may mean genuine health. It may mean the channels are not reaching anyone. The test is the sixth measure, silence: if many parties in scope have said nothing for a long time, the steadiness is probably the quiet of a closed membrane. Ask them.
Measures of voice share the limits of every measure of a living system, and a convenor should be able to state them plainly to a sceptical board.
They cannot tell you whether what was said was true. A count of concerns raised records that someone spoke, not that they were right. Judging the content is the sponsor's work, not the convenor's.
They cannot tell you about people who were never in scope. The six measures cover the relationships chosen at the start of the quarter. The exits and silences outside that boundary are invisible to them. That is why the audit is repeated, and why the second reader's first question is always who did we not look at?
They cannot tell you what caused what. A fall in exits after a voice channel opens is consistent with the channel working, as Freeman's work suggests in one setting. It is also consistent with a better labour market, a new manager, or a dozen other things. Over several quarters and several teams, a consistent pattern becomes more persuasive. In one quarter it is a reason to keep going, not a proof.
And they cannot measure the thing the whole book has been describing, which is a quality of relationship. Ecstatic Economics makes the general case against a single number in its account of gross domestic product: "The metric that drives global economic policy literally cannot distinguish between activity that generates genuine prosperity and activity that destroys it." Voice counts are much humbler instruments, but they share the limit. They count events. Whether the events add up to an organisation that hears itself is a judgement, made by people, reading the counts beside everything else they know.
At every gate in the field manual, the second reader asks the same four questions of the record. They are worth stating as the close of this chapter, because they are the discipline that keeps measurement honest.
Who will actually use this number, and for what? A count read by the sponsor to decide where to listen is safe. The same count read by a board to rank managers is a target, and it will be gamed.
What did the count not look at? Every measure has a denominator: the parties in scope, the channels recorded, the weeks covered. State it next to the number, every time.
Is the improvement real, or did the counting change? Check the dates on which channels opened and recording methods changed. Read across them carefully.
Is this a gap or a boundary? A silent party may be a relationship where voice has died. It may also be a relationship that is rightly arm's-length, where the price says almost everything that needs saying, as chapter 13 argued. Mark which it is before treating the silence as a failure.
Ammanuel Santa Anna's Conscious Governance Models records something that boards which take measurement seriously often discover: "Many governance groups report that their most important insights emerge not from discussion but from the silence between discussions." The numbers in this chapter are a way of making that silence legible, of seeing where it falls and how long it lasts. They do not replace the listening. They tell you where to go and do it. The fuller argument for measuring what money alone cannot see is in Ecstatic Economics; the series book on patient capital, patient capital, sets out why owners who intend to stay are the ones most willing to wait for measures like these to mature. For a balanced reading that keeps financial and relational measures on one page, see the Alpha Paper Alpha balanced-scorecard-1.
One sheet, kept weekly by the convenor, read monthly by the sponsor and quarterly by the second reader.
Reading the sheet. After two quarters, the measures should show a rising or stable count of voices with a falling median response time and a falling number of linked pairs. Valued exits should be no higher as a share of all exits. A long silence column that is shrinking means the membrane is reaching further. If every number improves at once in the first month, check step 4 before believing it.
Your page — run the The Voice Measures Sheet here, dated.
Chapter Eleven
The Hard Conversations
Haute Lumière · The Embedded Market · Chapter Eleven
There is a moment in every attempt to open a voice channel when the work stops being a method and becomes a conversation. A leader has run the embeddedness audit and found that the board hears about the firm's suppliers only as a cost line. A buyer has read chapter 5 and wants to tell a supplier of eleven years that the next quarterly meeting will not be about price. A finance director has to explain to a sceptical chief executive why a quarterly review is going to spend ten minutes on what nobody has said. A parent wants a teenager to know that a worry brought home will not be met with a lecture.
Each of these conversations is short. Each is harder than it looks, because the other person is listening less to the words than to what they predict. A supplier who has been squeezed for a decade hears the word partnership as the opening of a negotiation. Helper and Henderson's account of General Motors in chapter 5 is, among other things, the story of a firm whose purchasing director announced "an enlightened era of mutual trust and confidence" while one of its own purchasing agents said, in the same trade article, that efficient purchasing still meant "a lot of vendors fighting over a job." The words were sincere. The people who heard them had a history that said otherwise, and they believed the history.
This chapter is about five such conversations: with a board, with staff and their representatives, with suppliers and partners, with a family, and with a sceptic. Each section sets out the situation, what tends to go wrong, and words a reader can adapt and say aloud. The words are scripts in the plainest sense: a place to start, to be put into your own voice. None of them works as a performance. Each works only if what follows it is consistent with it.
Three findings, set out earlier in the book, govern every conversation in this chapter, and they are worth gathering in one place.
The first is that people who could leave are more willing to speak. Caryl Rusbult and her colleagues, testing an expanded version of Hirschman's model in 1988, found that better outside alternatives encouraged exit and also voice.43 Speaking costs less when leaving would not be a catastrophe. A conversation that implies the other party has no choice will get compliance, and sometimes neglect, rather than voice.
The second is that what a leader visibly does matters more than what they say. James Detert and Ethan Burris, studying 3,149 employees and 223 managers in a restaurant chain, found that managerial openness was more consistently related to employees offering improvement ideas than the more celebrated behaviours of inspiring leadership, that the relationship ran through employees' sense that it was safe to speak, and that it was strongest among the best performers.44 Every script below therefore ends with a commitment the speaker can be seen to keep.
The third is that the channel matters, not only the climate. Richard Freeman's 1980 study found that workers with a formal grievance procedure quit less often than comparable workers without one.45 A conversation that opens a channel without saying where it goes, how often, and what happens to what is said in it, has opened very little.
A fourth finding, from Brian Uzzi's study of twenty-three apparel firms in New York, belongs here as a caution.46 The benefits of close relationships rose to a threshold, and beyond it reversed: firms that were too closely bound to too few partners became closed to information and exposed when a partner failed. Not every hard conversation is about opening a relationship. Some are about widening one that has become too comfortable to tell the truth.
The situation. A board meets four or six times a year. It receives papers prepared by management, hears presentations, asks questions and approves decisions. It hears the firm's customers, staff and suppliers mostly as figures: revenue, turnover, cost. In Hirschman's terms, it hears almost entirely through exit.
Santa Anna's Conscious Governance Models describes a second difficulty, internal to the board itself. It describes the new member who senses that the real decisions are taken elsewhere, and says that such a member "experiences a quiet erosion of trust." And it names what boards most need and least often have: "This capacity for principled dissent is absolutely essential in governance, where groupthink is perhaps the single most dangerous dynamic." The same book observes that by the time a quarterly board has deliberated and communicated its decisions, "the landscape has already changed."
What tends to go wrong. A leader brings voice to a board as an initiative, with a slide and a programme name. The board hears a cost and a management fashion. Or the leader brings a single dramatic complaint, and the board hears an anecdote, which it is right to distrust.
What works better is to bring the board a pattern from its own organisation, in the form of a question it is entitled to ask, and to ask for a standing place on the agenda rather than a decision.
Words to adapt.
"Every paper this board receives tells us what has already happened: who left, what it cost, how the numbers moved. I want to add one short item to each meeting that tells us what we are being told before it shows up in the numbers. Ten minutes. Three things a customer, a supplier or someone on staff has raised since we last met, what we did about each, and one thing nobody has raised that we think they should have. I am not asking the board to approve a programme. I am asking it to hear the organisation in a second way, and to judge after a year whether it has been worth the ten minutes."
And, to invite dissent inside the room:
"Before we vote, I would like one person to make the strongest case against this, and I would like it to be someone who does not already disagree. If the case is weak, we will have lost two minutes. If it is strong, we will have found it here rather than later."
What follows the words. The ten-minute item appears at the next meeting, every time, including the meeting where there is bad news in it. A leader who brings only the comfortable voices to a board has taught it to distrust the item. The first time the item carries something embarrassing, and the leader presents it without defending against it, is the meeting at which the board starts to believe it.
Two: staff and their representatives
Haute Lumière · The Embedded Market · Chapter Eleven
The situation. A leader wants to hear more from the people who do the work, and there is a history. There may be a union, a works council, a staff forum, or nothing formal at all. There may be a memory of a survey whose results were never shared, a restructuring announced after a consultation that changed nothing, a person who spoke up and was moved aside.
Luminous Moral Development puts the question that every one of these histories answers: "How many people stay silent in the face of wrongdoing because the power dynamics in the room make speaking up feel dangerous?" In most organisations the honest answer is not known, because the people who stay silent do not report it.
What tends to go wrong. The leader treats the existing representatives as an obstacle to be gone around, and invites staff to speak directly. Representatives hear an attempt to undermine them; staff hear an invitation that their representatives have not endorsed, and wait to see what happens to the first person who accepts it. Helper and Henderson describe a version of this at General Motors in 1984: the company proposed teams and joint problem-solving, a leaked memo suggested the purpose was to reduce headcount, and for years afterwards many union leaders believed lean production was a way to speed up the line.
What works better is to go to the existing channel first, to name the history before anyone else has to, and to offer something that costs the leader before asking for something that costs the staff.
Words to adapt, to representatives:
"I want people here to be able to tell us what is wrong earlier than they do now, and I know that the last time management said something like this, it did not end well. I am not asking you to vouch for me. I am asking you to help design how it works, and to hold us to it. Here is what we are committing to first: every concern raised through any channel gets a reply within five working days, even if the reply is that we have not decided yet, and you will see the log of every concern and every reply each month. If we miss that, you will know before anyone else."
To a team, after the channel is agreed:
"If you see something that is going wrong, I would rather hear it from you this week than read about it in a resignation letter next year. You will not be asked to fix it as the price of raising it. You will hear back within five days about what happened. If I get that wrong, say so, and that counts as voice too."
What follows the words. The first concerns that arrive will be tests: small, safe, sometimes trivial. Treat them with exactly the seriousness of large ones. The log is shown. The replies arrive on time. Santa Anna's Organizational Soul Work gives the discipline for the moment a hard truth finally comes: resist the reflex to explain or fix. "When a truth is named and immediately followed by a solution, the system doesn't fully receive the truth." Receive it first. The fix can come tomorrow.
The situation. A buyer wants to separate the conversation about problems from the conversation about price, as the supplier voice protocol in chapter 5 sets out. The supplier has been through annual tenders, reverse auctions and price-down letters, and has learned that information shared with a customer tends to come back as a demand.
What tends to go wrong. The buyer says partnership and trust in the first minute. The supplier hears a negotiation opening. Or the buyer runs the first problem conversation well, and then a colleague in purchasing uses something said in it to argue for a lower price in the next round. One breach closes the channel for years.
What works better is to promise less than the supplier expects, in writing, and to keep it in a way the supplier can check.
Words to adapt.
"I want to try something with you that we have not done before, and I would rather under-promise. Four times a year, separate from any commercial meeting, I would like forty minutes with your engineers and ours about what is not working, what you can see coming, and what you would change in our specifications if you could. Nothing said in those meetings will be used in a price negotiation. I have put that in writing, and the head of purchasing has signed it. If you ever think it has been broken, tell me, and I will tell you what I found."
And the question that does the most work, at the end of every such meeting:
"What have you stopped telling us, and why?"
What follows the words. The written undertaking is kept in front of everyone in purchasing, not only the buyer who made it. The supplier's early warnings are counted and acknowledged once a year. Uzzi's threshold applies here: the buyer keeps more than one source for the parts that matter, and talks to both. A relationship that has become the only one is a relationship that has stopped being able to say hard things, because both sides have too much to lose.
The same protocol, slightly adapted, serves partners who are not suppliers: a distributor, a funder, a co-venturer, a joint team in another firm. The structure is the same. Keep the problem conversation apart from the commercial one, promise less than expected, show what happened to what was said.
Four: the family
Haute Lumière · The Embedded Market · Chapter Eleven
The situation. Every household has a culture of voice, whether it has thought about it or not. Somebody notices that money is tight before anyone says so. A child senses a worry and decides not to mention their own. A partner learns about a debt from a statement rather than a conversation.
Ecstatic Economics asks the question that opens this conversation, in its chapter on money and the body: "When you discuss money with a partner or family member, what happens to both of your bodies?" For many people the honest answer is that something tightens before a word is said. That is the price of voice in that house, and it was set long before the conversation began.
What tends to go wrong. A parent or partner announces openness and then meets the first difficult disclosure with alarm, a lecture or a solution. The disclosure is not repeated. Or money is discussed only in crisis, so that the subject itself becomes a signal of danger.
What works better is to make the conversation regular, small and dull, so that it does not have to wait for a crisis, and to receive before responding.
Words to adapt, between partners:
"I would like us to spend twenty minutes on the first Sunday of each month going through where the money went and what is coming up. Not to judge, and not only when something has gone wrong. If either of us is worried about something, that is where it goes, and the other one listens all the way through before saying anything."
To a child:
"If something is bothering you, about school, about friends, about anything, I want to hear it. I will not be cross that you told me. I might need a minute to think before I say anything, and that is all the minute means."
What follows the words. The first Sunday happens, and so does the second. The parent who has promised not to be cross is not cross, the first time, even if the news warrants it. Chapter 12 takes the household conversation further, into the providers and institutions a family depends on.
The situation. Somewhere in most organisations there is a person, often senior and often right about many things, who regards all of this as cost and sentiment. The chief executive who wants numbers. The finance director who sees forty minutes of engineers' time per supplier per quarter. The colleague who says that people complain too much already.
What tends to go wrong. The advocate argues values against numbers, and loses, because the sceptic is responsible for the numbers. Or the advocate overclaims, promising returns the evidence cannot support, and the sceptic is right to discount everything else they say.
What works better is to agree with the part of the objection that is true, to bring the narrow evidence honestly, and to propose a test the sceptic can judge.
Chapter 13 sets out the objection from efficiency at full length, and grants most of it: for a great many transactions, a price says almost everything that needs saying, and an arm's-length market is the right membrane. The argument for voice applies to a narrower set of relationships, the ones where what needs to be known has not happened yet.
Words to adapt.
"You are right that most of what we buy should be bought on price, and I am not proposing to change that. I am proposing that for the twelve relationships where a failure would hurt us most, we find out about problems before they reach the numbers. The research I can point to is narrow but consistent. Suppliers selling to both Toyota and an American carmaker cut their defects for Toyota by half over six years, and by about a quarter for the American customer, according to the study Helper and Henderson cite. Workers with a grievance channel quit less often. That is not a promise about us. So here is the test: we run this for two quarters with those twelve, we count how many problems were raised early and what they would have cost late, and you decide whether to continue. If the count is zero, we stop."
What follows the words. The count is kept honestly, as chapter 10 describes, including the quarters where it is disappointing. A sceptic who sees an advocate report a weak result without spin will trust the next strong one. And a sceptic who is right should be thanked for it, in public. That is voice, too, and it is the voice most organisations find hardest to receive.
Conscious Governance Models names what lies underneath all five conversations. Voice, it says, cannot be only procedural; it must be cultural, "a genuine collective commitment that truth-telling is valued even when it is uncomfortable." The scripts are procedure. The culture is what the speaker does the week after.
No script repairs a history by itself. Helper and Henderson's account of General Motors is sobering on this point: the firm's supplier scores improved through the 1980s and then, by 1993, suppliers again felt GM could not be trusted. Credibility built over years can be spent in a quarter. The conversations in this chapter open a door. Whether anyone walks through it depends on what happens the first few times they do.
Nor do the scripts substitute for structure. A board item, a reply deadline, a written undertaking to a supplier, a monthly household meeting: each is a small piece of institution. Chapter 9 sets out how to build them in sequence, and chapter 10 how to tell whether they are working.
And the research behind the chapter is narrow. It comes from particular places and periods: a restaurant chain, American labour markets in the 1970s, apparel firms in New York, the car industry. It describes a consistent pattern, not a law. The scripts are ways of testing the pattern where you are.
One card, index-card size, prepared before any of the five conversations in this chapter. Allow twenty minutes to write it. Keep it, and look at it again a month later.
Reading the card. A card whose promise was kept, and whose back records a second conversation the other person started, describes a channel that has opened. A card whose promise was missed is the most useful card in the set: it shows where the next conversation has to begin, with the miss, said first and plainly.
Your page — run the The Conversation Card here, dated.
Chapter Twelve
The Kitchen Table
Haute Lumière · The Embedded Market · Chapter Twelve
It is a Tuesday evening, and there is a letter on the kitchen counter that nobody has opened. Everyone in the house knows roughly what it is. It is from the landlord, or the bank, or the energy company, and it is the second one. The first was opened, read, and put back in its envelope. Nobody said anything about it at dinner.
Every chapter of this book so far has been about organisations: a railway, a carmaker, a hospital, a firm and its suppliers. This one is about a household, because a household is where most people first meet the market, and where they first learn whether voice is safe. Karl Polanyi wrote about civilisations, and Albert Hirschman about firms and states, but the three things Polanyi said a market most easily forgets it is holding, people's working lives, the places they live, and the money they run on, arrive in every household as a wage, a rent and a balance. The letter on the counter is one of them.
The claim of this chapter is that a household has a membrane too, made of a dozen or so relationships that carry its income, its home and its health, and that the same rhythm of exit and voice keeps it alive. It also has a second membrane, inside the first: the conversations between the people who live there. A household that cannot speak about money at its own table will struggle to speak about it to anyone else.
The examples in this chapter are composites, assembled from ordinary situations and labelled as such. None describes a particular family. None of what follows is financial advice. It is a way of seeing which relationships in a household carry information and which have been reduced to a price.
Polanyi's argument, traced at the start of this book, was that labour, land and money are not commodities in the ordinary sense, because none of them was produced for sale. Labour is human life. Land is nature and place. Money is a token created by banks and states. Treating them as if they were ordinary goods, he argued, puts the people who depend on them at risk.
At the scale of a household those three abstractions become very concrete.
Labour is the job. For most households the largest single relationship is with an employer, or with the clients of someone self-employed. It carries the income on which everything else depends, and, very often, a large share of the household's daily time and stress.
Land is the home. A rent or a mortgage, a landlord or a lender, a neighbourhood, a school within reach. The place a household lives is, in Polanyi's sense, not something it can easily treat as a commodity, because moving is expensive, slow and hard on children.
Money is the bank and the bills. The current account, the credit card, any loan, and the providers of the things that must be paid for every month: energy, water, a phone, insurance.
For each of these, the question this book has asked of firms applies unchanged. Is this a relationship in which the other party can tell us something, and in which we can tell them something, before it is too late? Or is it a relationship in which the only signal is that one of us leaves, or stops paying, or sends a letter that nobody opens?
Consider a composite. A woman in her forties has worked for the same mid-sized company for eleven years. She is good at her work. For the past year her workload has risen steadily as colleagues who left were not replaced. She has not said anything, because the last person who raised workload in a team meeting was described afterwards as not a team player, and because she needs the job.
She is in Hirschman's position exactly. Exit is possible but costly: a search, a risk, perhaps a pay cut, at an age when changing employer is harder. Voice is possible but has been priced high by one remembered meeting. So she does neither. She stays, and she begins, without deciding to, to do a little less, to care a little less, to come in a little later. The researchers who extended Hirschman's model called this fourth response neglect, and it is the most common response to a job that has stopped listening.
Chapter 3 set out what a leader can do to lower the price of voice. This chapter asks what the employee can do, on her own side of the membrane, and the answer is three things.
First, separate the concern from the complaint. A concern framed as a request for information is cheaper to raise than a grievance. I want to make sure I am spending my time on the right things; can we look at the list together? carries the same information as I am overloaded, at a lower price.
Second, raise it in the channel where it costs least. For many people that is a one-to-one conversation rather than a meeting, and a written follow-up rather than a corridor remark. A short note after the conversation, saying what was agreed, turns a voice into a record.
Third, keep exit available, quietly. Rusbult and her colleagues found that people with better alternatives were more willing to use voice as well as exit, because speaking cost them less.48 Knowing what else is possible, keeping a professional network warm, having some savings set aside, does not mean planning to leave. It means being able to speak.
Luminous Moral Development asks the question that sits underneath every one of these choices: "How many people stay silent in the face of wrongdoing because the power dynamics in the room make speaking up feel dangerous?" A workload is not usually wrongdoing. But the same dynamic governs the smaller silences, and they add up.
The home: a landlord who knows your name
Haute Lumière · The Embedded Market · Chapter Twelve
A second composite. A couple with a young child rent a flat from a private landlord who owns three or four properties. In the first two years the landlord came round when something broke, mentioned when he was thinking of selling, and once let them pay a week late without fuss. Then he handed the properties to a letting agency, and the relationship became a portal. Repairs are logged online and assigned a reference number. The rent is collected automatically. Nobody at the agency has ever been inside the flat.
Nothing about this is wrong. It is very common, and for many tenants it is more efficient. But in the language of this book the relationship has been disembedded. Information that used to travel inside it, the boiler is making a noise, I may need to sell next year, we are having a hard month, now travels, if at all, as a reference number or a notice.
A household cannot reverse the letting agency. It can do three small things that keep the membrane alive.
Keep one human channel. Find out the name of the person at the agency responsible for the property, and use it. A repair request addressed to a person is answered differently from a ticket.
Report early and in writing. A problem raised when it is small, with a dated note, is information. The same problem raised when it has become a leak is a dispute.
Speak before the letter. If the household is going to have trouble with the rent, the cheapest moment to say so is before the payment is missed, not after. That is not always possible, and nobody should pretend it is easy. But the evidence of every chapter in this book points the same way: voice that arrives before exit is worth far more than voice that arrives after it.
Go back to the letter on the counter. It is the clearest single image of a household membrane that has stopped carrying signals inward. The information is physically present in the house. Nobody is reading it.
Ammanuel Santa Anna's Ecstatic Economics explains why in terms of the body rather than of arithmetic. Money anxiety, the book argues, is held in the nervous system, and a person in a state of shutdown about money is not being lazy or irresponsible. The book is blunt about what does not help: "Shaming someone in dorsal vagal about their financial behavior is not only cruel". The sentence goes on to say that shaming is also counterproductive, and it is. A letter that nobody opens is not a failure of character. It is a household in which the price of voice about money, even voice to oneself, has become too high.
The practical response is the same one this book has offered every organisation: lower the price of the signal. For a household that might mean opening post together, at a fixed time each week, so that no one person carries it alone. It might mean writing the three numbers that matter on a single sheet, income, fixed costs, and what is left, and looking at the sheet rather than at the bank app, which is designed for transactions rather than understanding. It might mean, for the household in difficulty, calling the lender before the missed payment, and asking what options exist.
The High Frequency Workbook gives a principle that applies here as much as to any budget: "Every financial decision is a values decision." Deciding not to open the letter is also a decision, and what it values is a few more days of not knowing. That is an understandable value. It is rarely the one a household would choose if it chose consciously.
The same book makes a larger claim, which this chapter tests rather than assumes: that "the patterns you see in your financial life are the same patterns you see everywhere else." If that is true, a household in which nobody opens the post is likely to be a household in which other uncomfortable information also waits on the counter. That is worth noticing without blame. It is also worth noticing that the reverse holds: a household that learns to open the post together is practising a habit that will serve it in every other conversation.
Every relationship above runs through a second, smaller membrane: the conversation between the adults in the house, and between the adults and the children. A household whose members cannot talk to each other about money will not talk well to a landlord, an employer or a bank, because whatever they say outside has not first been said inside.
Ecstatic Economics asks two questions about this that are worth putting to any household exactly as written. The first: "When you discuss money with a partner or family member, what happens to both of your bodies?" The second: "What economic somatic patterns did you inherit from your family of origin?"
These are not idle questions. Most adults learned what money talk feels like long before they had any money, at their own parents' table, from the tone in which bills were discussed, or the silence in which they were not. A person who grew up in a house where money was spoken about only in anger may find that their shoulders rise when a partner says can we look at the accounts. A person who grew up in a house where money was never mentioned may not notice that they change the subject every time.
Hirschman's three words apply here with surprising precision.
Exit, in a household, is withdrawal: the partner who stops sharing what they spend, the teenager who stops mentioning what they need, the adult who quietly opens a separate account. Sometimes exit is healthy. Often it is the household's version of the railway's best customers driving off: the information leaves with the person.
Voice is the conversation: the request, the worry, the disagreement said aloud.
Loyalty is what holds the household together while the conversation is difficult. It is built, as the sibling book on gift and reciprocity argues, from what has been given before it was asked for: patience, attention, the benefit of the doubt.
For children, the price of voice is set early and remembered for a long time. A child who mentions that a school trip costs money and is met with a sigh, or a lecture, learns that bringing money into the open is expensive. A child who is thanked for mentioning it, and told plainly what the household can and cannot do, learns that it is safe. Neither child will remember the trip. Both will remember the price.
The neighbour and the doctor
Haute Lumière · The Embedded Market · Chapter Twelve
Two relationships in a household's membrane deserve separate mention, because they are the ones most often allowed to thin without anyone deciding it.
The first is the neighbour. For most of history, a household's nearest neighbours were part of its economy: the people who lent a ladder, watched a child, noticed when the curtains stayed shut. Many households now know their neighbours only by sight. Nothing is lost on an ordinary day. On the day something goes wrong, a fall, a burst pipe, a lost job, the neighbour is often the only person close enough to notice before anyone else does. A neighbour who knows your name is a voice channel that costs almost nothing to keep and is very hard to rebuild in a hurry.
The second is the doctor. In many places a family that once saw the same general practitioner for years now sees whoever is free. That is often the right trade for a quick appointment. It also means that nobody in the system holds the household's history in their head, and so nobody is in a position to say this is different from last time. Where a choice exists, keeping continuity with one clinician for the people in the household with the most complicated health is the medical equivalent of the garage that knew the cars, described in chapter 5.
Neither of these is a financial relationship in the narrow sense. Both carry information about the household's health, safety and capacity to earn, which is to say about two of Polanyi's three fictitious commodities. Both are, in this book's terms, part of the membrane. And both are kept alive in the same way as any other: by small deposits made before they are needed.
The household is the scale at which this book's ideas are easiest to test and hardest to measure. There is research on financial stress and health, on couples' communication about money, and on the effects of financial literacy programmes, but this chapter has not drawn on it, because the book's argument does not depend on any one of those findings and because most of it does not test Hirschman's model directly. The composites here are illustrations, not evidence. And the advice to keep exit available, through savings or a wider network, assumes resources many households do not have. Voice is cheapest for the people who could leave. That is a finding about the world, not a recommendation, and it is one reason the book has argued throughout that organisations, landlords and lenders have a responsibility to lower the price of voice from their side of the membrane, rather than leaving it to the people with least room to move.
A last finer point for the inner membrane. Money conversations in a household go best when they are regular and dull rather than rare and dramatic. A fixed half hour once a month, at the same time, with the same sheet of paper, lowers the price of voice more than any single honest conversation can, because nobody has to choose the moment and nobody has to wonder what the summons means. The meeting becomes furniture. And furniture, in a house, is what people stop being afraid of.
Back to the kitchen. The letter is still on the counter. Imagine the household that opens it on Sunday morning, together, at the table, with tea. The news inside is no better than it would have been on Tuesday. But it is now information rather than dread. Someone writes down what it says. Someone else says, a little reluctantly, that they had a letter like it last month and did not mention it. There is a pause, and then a plan, and the plan includes a phone call on Monday to a person whose name they find on the second page.
Nothing about the household's position has changed except one thing. The signal has crossed the membrane. Polanyi wrote that markets must be held inside the life they serve. At the kitchen table, that holding is very small and entirely practical: a letter opened, a worry said aloud, a call made before the payment is missed. It is where most people first learn whether voice is safe, and where they can learn, at any age, to make it so.
---
This chapter draws on Ecstatic Economics by Ammanuel Santa Anna, Ecstatic Economics, and on The High Frequency Workbook. For the household version of reciprocity and the deposits that make difficult conversations possible, see the sibling book on gift, reciprocity and trust, gift and reciprocity. For the Alpha Paper on building wealth that lasts across a family's generations, see Alpha generational-wealth-1. More at hautelumiere.life.
For one person living alone, a couple, or a whole family. Allow ninety minutes the first time, at a table, with paper, ideally at a calm hour rather than after a difficult letter. Repeat every three months. It is not a budget, though it sits well beside one, and it gives no financial advice. It maps which of your household's relationships can carry a warning in time.
Reading the map. Count three things each quarter: lines marked portal only, lines marked silent, and unopened letters. The first two should fall over a year for the relationships that carry your income, your home and your health. The third should be small and should never be the same item two quarters running. A household whose most important relationships have names in them, and whose members can say a worry aloud at the table, has a membrane that listens.
Your page — run the The Household Membrane here, dated.
Chapter Thirteen
Holding the Market Well
Haute Lumière · The Embedded Market · Chapter Thirteen
In 1982 Albert Hirschman, twelve years after Exit, Voice, and Loyalty, published an essay that set three old opinions about the market side by side.52 The first, which he traced to the eighteenth century, held that commerce civilises: people who trade with each other learn to be polite, reliable and peaceable, because their interests depend on it. The second, which ran through Marx and many conservatives alike, held that the market destroys: it dissolves the traditions, loyalties and moral restraints on which it depends, and so eats away at its own foundations. The third held that the market is feeble: that in many places it had never fully freed itself from older structures of power, and that its problems came from too little market rather than too much.
Hirschman's point was not to choose. It was to notice that each opinion was true somewhere, and that people tend to hold one of them as if it were the whole truth. Marion Fourcade and Kieran Healy, reviewing the sociology of markets and morality in 2007, built their survey on his framework and found all three views still alive in the research, often in the same field.53
This book has spent its first chapters on two books that belong, roughly, to Hirschman's second opinion. Karl Polanyi's The Great Transformation argued that a market lifted out of society destroys the substance it lives on. Exit, Voice, and Loyalty described how an organisation that can hear only departures declines without understanding why. Both are warnings. But neither Polanyi nor Hirschman was an enemy of markets, and the closing claim of this book is that their warnings are, in practice, instructions for holding a market well. A market held inside working relationships, with voice that travels and exit that stays available, is the civilising market of the first opinion. The destroying market of the second is the same institution, held badly.
The question for a leader or a household is therefore not whether to have markets, but how to hold them. This chapter joins that question to three other books in Ecstatic Economics, sets out what to do on Monday, and ends with one tool and one image.
The nearest neighbour to this book is the work of Elinor Ostrom, whose study of how communities govern shared resources is the subject of a companion volume in this series, The Pasture Was Never Empty. Ostrom's design principles for long-enduring commons include several that are, in Hirschman's language, voice channels with rules: arrangements that let the people affected by the rules take part in changing them, monitors who are accountable to the users, graduated sanctions, and cheap, local ways of resolving conflict.
Read beside this book, a working commons is a case of a resource that is neither fully marketised nor fully controlled by the state. It is embedded, in Polanyi's sense, in the relationships of the people who use it. And it stays embedded because voice works: people who see a problem can raise it, and the rules can change. Ammanuel Santa Anna's Ecstatic Economics describes the opposite case, enclosure, as the moment when "the relational web of mutual obligation that governed their use is severed." Ostrom's commons is that web kept alive by design.
The practical lesson for a leader is that voice does not have to be informal to be real. A shared resource inside an organisation, a budget pool, a lab's equipment, a team's time, can be governed by a few clear rules about who decides, how concerns are raised and how disputes are settled. Those rules are the membrane described in chapter 4, written down.
The second neighbour is the economics of gift and reciprocity, the subject of gift and reciprocity. Polanyi drew directly on the anthropology of the gift, and so does Santa Anna's book, which describes, in the gift economies Marcel Mauss and later scholars studied, a society in which status came from giving rather than holding. "To hoard was not shrewd," the book says of those economies. "It was shameful."
The connection to this book is through loyalty. Hirschman's loyalty is the attachment that holds a member in place long enough for voice to work. In a gift relationship, that attachment is built from what has been given and received over time, which is why it can bear the weight of a complaint. A customer who has been treated generously can tell you something hard without the relationship breaking. A customer who has only ever been charged correctly has less reason to stay and speak.
The book is careful not to romanticise this, and this book has followed it. Gift relations can carry obligation and hierarchy as well as warmth. But the principle stands: loyalty with roots is built by giving something before it is asked for. In practical terms, this is the supplier who is told about a change in plans before the contract requires it, the employee whose good work is noticed before review season, the neighbour who is helped before they ask. Each of these deposits makes the next difficult conversation possible.
Patient capital: time for voice to work
Haute Lumière · The Embedded Market · Chapter Thirteen
The third neighbour is patient capital, the subject of patient capital. Voice takes time. A complaint has to be heard, understood, acted on and followed up, and the improvement it leads to may not show in the next quarter's figures. An organisation under pressure to show results every ninety days is under constant pressure to choose exit over voice: to replace the supplier rather than work with them, to let the unhappy employee go rather than find out why, to cut the underperforming product rather than ask its customers what is wrong.
That is why the question of who owns a firm, and how long they intend to hold it, is part of this book's subject. Owners and lenders who expect to be present for years have reason to invest in voice, because they will be there when it pays. Owners who expect to sell soon have reason to prefer the cleaner signals of exit, because those show up on the timescale they care about. Chapter 5's contrast between General Motors and Toyota is in part a story about which kind of patience each firm brought to its suppliers.
Santa Anna's book draws on the Haudenosaunee principle of considering decisions across seven generations as one expression of the same idea: the people who will live with a decision have a claim on it. A leader does not need seven generations to apply it. Three years is enough to change which signal an organisation prefers.
The four have one thing in common, and it is what this book has been circling. Each is a way of keeping a market inside relationships that can carry information. The commons does it with rules. The gift does it with obligation freely taken on. Patient capital does it with time. Embeddedness and voice name the result: a market that can still hear what it is doing.
The book puts the underlying principle in its account of what it calls us-two logic, drawn from the work of Tyson Yunkaporta, in which every economic act happens within a web of relationships. "To make an economic decision without considering these relationships is not rational," the book says. That sentence would have satisfied Polanyi. Hirschman would have added only the practical corollary: and you will not know what the relationships are telling you unless you have built a way for them to tell you.
The Luminous 100 gives the shape of the whole in its reading of the Viable System Model: "holons all the way up and all the way down, each simultaneously autonomous and embedded." A household inside a neighbourhood, a team inside a firm, a firm inside a market, a market inside a society, a society inside the living world. Each one makes its own decisions. Each one is held inside something larger that it needs to hear.
That nesting is also why the practical steps below work at any size. The same three questions, where am I hearing only exit, what does it cost to speak, and which conversations are kept apart from price, apply to a family deciding which providers to trust, a team lead with a dozen people, and a board responsible for a supply chain across several countries. The scale changes the effort. It does not change the question.
There is a serious objection to everything, and it deserves an answer before the practical steps. Markets work, the objection runs, precisely because they do not require relationships. A price carries an enormous amount of information in a single number, and lets strangers cooperate without knowing or trusting each other. Every hour spent on voice, on long relationships and supplier conversations and grievance channels, is an hour not spent producing. Exit is cheap, fast and impersonal, and that is its virtue. Why replace it with something slow and costly?
The answer this book gives is not that the objection is wrong. It is right about a great many transactions. A household buying petrol, a firm buying standard stationery, an investor buying a widely traded share: in each case, the price says almost everything that needs saying, and an arm's-length market is the best membrane available. Polanyi himself never argued otherwise for genuine commodities.
The objection fails in a narrower and more important set of cases: where what is being exchanged is complex, changes over time, or involves the things Polanyi called fictitious, people's working lives, land and the places people depend on, and the credit that keeps an organisation running. In those cases, the price cannot carry what needs to be known, because what needs to be known has not happened yet. The supplier's engineer knows something about next year's part that no price this year can reflect. The employee knows something about next quarter's customer that no salary can encode. In those relationships, exit tells you late and voice tells you in time.
The skill this book asks of a leader is to know which relationships are which, and to hold each one accordingly: lightly where a price says enough, closely where it cannot.
What to do on Monday
Haute Lumière · The Embedded Market · Chapter Thirteen
For a leader, three things, in order.
First, find out where you are hearing only exit. Use the embeddedness audit from chapter 2, or simply list your ten most important relationships and ask, for each, when they last told you something you did not know. Wherever the answer is never or only when they left, you have found a place where the market around you has been disembedded.
Second, lower the price of one voice. Chapter 3's voice cost ledger shows where speaking up is expensive in your organisation. Choose one cost, slow response, no follow-up, private-only channels, and change one habit for a month. The research on voice suggests that what you visibly do with what you are told matters more than what you say about wanting to hear it.
Third, separate one voice channel from price. If you buy from others, use the supplier voice protocol from chapter 5 with your three most important suppliers. If you manage people, hold one conversation each month that is explicitly not about performance ratings or pay, and ask what they are worried about that you have not asked.
For a household, the same three steps, smaller. Notice where the family learns about problems only from a bill or a departure. Lower the price of speaking up at home, especially for children, by thanking a concern before answering it. And keep one or two long relationships with the people and businesses you depend on, a doctor, a mechanic, a bank, a neighbour, deep enough that they would call before something goes wrong.
The household version deserves a little more space, because it is where most people first meet the market and where the lessons of this book are easiest to test. A household's membrane is made of perhaps a dozen relationships that matter: an employer or two, a landlord or lender, a doctor, a school, a few providers of things that break, the neighbours. For most families, some of these have already become purely priced, met through an app or a call centre, and some remain personal. Neither is wrong. The question is whether the personal ones are the right ones: whether the relationships that carry the family's health, home and income are the ones where someone would tell them something in time.
The High Frequency Workbook describes money itself as "a river of energy that flows through the world in patterns shaped by both individual choice and collective structure." A household cannot change the collective structure on Monday. It can choose which of its relationships to hold closely, and it can make it easy for its own members, children included, to say when something is wrong. That is embeddedness at the scale of a kitchen table, and it is the scale at which most people learn whether voice is safe.
None of these steps is large. Santa Anna's Luminous Moral Development makes the case for starting small in a sentence about personal courage that applies just as well to institutions: "you don't build strength by lifting the heaviest weight once. You build it by lifting manageable weights repeatedly." An organisation learns to hear itself the same way, one returned call at a time.
A book that argues for honesty about limits should end with its own. It has relied on two classic books, one historical argument and one theoretical framework, and on a body of research that supports their central mechanisms in particular settings: grievance channels and quit rates in American labour markets, managerial openness and employee voice in one restaurant chain, supplier relationships in the car industry. None of that research shows that more voice is always better, that embedded relationships are always wise, or that any particular practice will produce a particular result in a particular organisation. It shows a consistent pattern, and the tools are ways of finding out whether the pattern holds where you are.
The living-systems reading in chapter 4 is a lens for noticing, not a measurement. Its value is in what it lets a leader see. Whether what they see is accurate is a question for the audit, the ledger and the conversations, not for the metaphor.
Here is the image this part of the book ends on. A supplier's engineer is standing at a workbench with a part that does not quite fit. A carmaker's engineer has driven over to look at it. Neither of them is negotiating. One of them is pointing at the place where the seal wears, and the other has a hand on the bench beside it, leaning in. In a few minutes the drawing will change. Nobody outside that room will ever know.
That is what an embedded market looks like when it is working. Not a price on a screen, though there will be one. Not a contract, though there is one of those too. Two people inside a relationship that can carry a small, inconvenient truth from the place where it was seen to the place where it can be acted on. Polanyi wrote a book to say that markets cannot live without that. Hirschman wrote one to explain what happens when it fails. The bench is where it is kept.
Fifteen minutes, once a week, for a leader or a household. It keeps the rhythm of exit and voice visible so that neither phase stops unnoticed. Keep the answers in one notebook or one document, dated, so that the weeks can be compared.
Reading the notebook. After a month, count three things: exits marked unknown, voices marked nothing yet, and linked pairs. All three should fall over a quarter. If voices rise while exits fall, the rhythm is working: people are speaking before they leave, and you are hearing them. If exits rise while voices fall, the organisation or household is moving toward silence, and the next step is the audit from chapter 2, done in full.
Your page — run the The Weekly Exit and Voice Review here, dated.
Chapter Fourteen
The Listening Edge
Haute Lumière · The Embedded Market · Chapter Fourteen
Thirteen chapters have circled one thing from several sides. It is worth saying plainly what it is.
Every living whole, a cell, a person, a household, a firm, a market, a society, has an edge where it meets what it depends on. The edge does two jobs at once. It holds the whole together, so that what is inside stays organised. And it lets things through, materials, money, people, information, so that the whole knows what is happening around it and can respond. Karl Polanyi's word for a market whose edge is working was embedded. Albert Hirschman's three words, exit, voice and loyalty, describe how that edge senses: by losing members, by hearing them, and by holding them long enough to hear.
The book's claim has been that these are not two ideas but one, seen from two distances, and that the place to read the health of a market, a firm or a household is the moment when somebody has something to say. Chapter 2 traced the idea to its sources. Chapter 3 let the critics narrow it. Chapter 4 read it as a membrane, a breathing rhythm and a root system. The cases followed it into a supply chain, a factory, a fire and a hospital ward. The manual, the measures, the conversations and the household chapter turned it into practice.
This last chapter reads the whole once more, as a living system, and ends on one image.
In chapter 4 the cell on the slide stood for the market's embedding: a membrane a few molecules thick, selectively permeable, carrying signals, constantly rebuilt from inside. The cases give the image weight.
At General Motors in the years Helper and Henderson describe, the membrane between the carmaker and its suppliers was built for one kind of signal. Contracts of a year, six or eight suppliers per part, purchase decisions driven overwhelmingly by price. A senior purchasing manager could say, in 1993, that the firm did not need to understand its suppliers' technologies because it let the market tell it. A membrane like that lets prices through and very little else. The supplier's engineer who saw that a bracket would not hold had no reason to say so, and the warning arrived later as a defect.
At Fremont, the same plant that had been one of GM's least productive, with absenteeism that often exceeded twenty per cent and a backlog of more than a thousand grievances, was reopened with largely the same workforce and a cord above each work station. The membrane between the line and the people who could fix it had been rebuilt. The signal that had travelled as grievance and sabotage began to travel as a pulled cord.
At Kariya, when the only plant making a valve every Toyota car needed burned, the edge between dozens of firms turned out to be a membrane rather than a wall. Knowledge, people and machines moved across it in days, because the firms had been moving knowledge across it for years.
At Stafford, the edge between a hospital and its patients carried complaints, letters and visits from relatives, and the inquiry found that the trust had failed to listen to them. The membrane let the words in and did not carry them to anyone who acted.
Each case is the same structure. What differed was not whether the organisation had an edge, but whether the edge listened.
Chapter 4 proposed that exit and voice are not alternatives but two phases of one rhythm. Exit is the out-breath, how a living system releases what no longer fits. Voice is the in-breath, how it takes in what it needs to know. A system that only breathes out loses what it needed to learn. A system that only breathes in, that holds people who should have gone, becomes Hirschman's lazy monopoly and, in the language of chapter 3, gets neglect, the held breath of people who stayed and stopped trying.
The measures in chapter 10 are, read this way, a way of listening to the breath. Exits marked unknown. Voices marked nothing yet. Linked pairs, where somebody spoke and later left, which are the places where an in-breath arrived and was not taken. A rhythm in good health shows voices rising while unexplained exits fall: people speaking before they leave, and being heard.
Santa Anna's Ecstatic Economics describes the same rhythm in the language of trust, in a sentence about organisational life: "trust circulates, and where trust circulates, value flows." Circulation is the right word. Neither trust nor voice can be stored. Both exist only while they are moving.
The roots, again
Haute Lumière · The Embedded Market · Chapter Fourteen
Loyalty was the root system: the part of the organism below the surface that holds it through a bad season and lets it draw on reserves. The Fremont workers who pulled a cord in their first months, before they had any proof that it would not count against them, were drawing on something: perhaps on the agreement to rehire them, perhaps on the plain wish to keep the plant open. The suppliers who sent people to rebuild a competitor's product line in the week after the fire were drawing on years of relationship that no contract required them to honour.
Roots can also hold a plant in poor soil. Chapter 3 described how loyalty can be used against the loyal, and Uzzi's apparel firms showed what happens when a firm's relationships become too few and too close: the depth that carried trust also carried blindness. Deep roots need a wide spread. The households and firms that held up best in these chapters kept a few relationships close and several at arm's length, and talked honestly in both.
The Luminous 100's reading of the Viable System Model has stood behind this book from the second chapter: "holons all the way up and all the way down, each simultaneously autonomous and embedded." The cases showed the nesting at every scale. A worker inside a team inside a plant inside a joint venture inside two companies. A supplier inside an association inside a network. A ward inside a hospital inside a trust inside a national health service. A child inside a household inside a neighbourhood.
At every level the same question applied, and it is the only question this book has really asked. When somebody inside this whole has something to say, is there a place to say it, and does anybody have to listen?
Ecstatic Economics gives the deeper reason the question matters, in its account of the economies Polanyi studied: "Every economic act occurs within a web of relationships." A web is held at many points. When one strand carries only price, the web still holds; it simply knows less about itself. When enough strands carry only price, it holds until it does not, and it learns why only afterwards.
What the reading asks of a reader
Haute Lumière · The Embedded Market · Chapter Fourteen
The living-systems reading is a lens. It does not measure, and the book has tried to keep it from pretending to. What it does is change what a person notices. A silence becomes a signal. A departure becomes a question rather than a statistic. A complaint becomes the most valuable document in the week's post. A supplier's early warning becomes a saving that never appears in the accounts.
Organizational Soul Work gives the posture that the reading asks for, and it is harder than any tool in this book: "listen first, deeply and patiently, before you do anything at all." Every method in the preceding chapters, the audit, the ledger, the map, the protocol, the weekly review, the conversation card, is a way of making that posture into a habit an organisation can keep when the person who believed in it has moved on.
The same book has a sentence for the moment when an organisation first hears what it has not been hearing, and it is the right sentence to carry out of this one. When a long-suppressed truth is finally said aloud, it says, the only fitting response is to let it land. Not to explain it, not to fix it yet. To receive it.
This book is one of a set. The questions it leaves open are taken up in its companions in Ecstatic Economics, all kept at hautelumiere.life.
How a shared resource stays governed by the people who use it, through rules that are themselves voice channels, is the subject of the volume on the commons. How loyalty with roots is built by what is given before it is asked for is the subject of the volume on gift and reciprocity. How the time horizon of an organisation's owners decides whether it prefers exit or voice is the subject of the volume on patient capital. How voice can be built into the ownership of the firm itself is the subject of the volume on employee ownership. And how the gains that come from shared improvement have been returned, for nearly a century, to the people who made them, is the subject of the volume on gainsharing, which is where the argument of this book leads for any organisation that wants to share in what its listening finds.
Behind all of them is the book they grew from, Ammanuel Santa Anna's Ecstatic Economics, part of the Luminous Developmental Canon: an account of money, markets and institutions as living systems that live in the body as well as in the ledger.
For leaders who want to read their own organisation this way and act on what they find, Ammanuel Santa Anna works with a small number of them, one to one. hautelumiere.life
Here is the image the book ends on.
It is late afternoon in a supplier's workshop. A part lies on a bench under a lamp: a small bracket, a millimetre off, or a seal that wears faster than the drawing said it would. Two engineers stand on either side of the bench. One works for the firm that makes the part. The other works for the firm that buys it, and has driven over to look. Neither of them is negotiating. One is pointing to the place where the wear begins. The other is leaning in, one hand flat on the bench beside the part, listening.
In a few minutes the drawing will change. The part that would have failed in a customer's car will never be made. The warranty claim that would have followed will never be filed, and the purchasing department will never go looking for a new supplier. None of this will appear in either firm's accounts, because the cost it prevented was never incurred. Nobody outside the workshop will ever know it happened.
That is what an embedded market looks like when it is working. There is a price, and a contract, and an exit available to both of them if it comes to that. But what is happening at the bench is none of those things. It is two people inside a relationship that can carry a small, inconvenient truth from the place where it was seen to the place where it can be acted on.
Polanyi wrote a book to say that markets cannot live without that. Hirschman wrote one to explain what happens when it fails. The bench is where it is kept: a hand on the wood, a lamp, and somebody listening.
Chapter Fifteen
The Workbook
Haute Lumière · The Embedded Market · Chapter Fifteen
An afternoon's work for a leader, an hour's for a household. Use a single sheet of paper, landscape, with five columns.
Reading the sheet. Lines with a recent voice, a named channel and no exits are working embeddings; leave them alone and thank the people in them. Lines marked exit only among your most important relationships are where you are learning about trouble last. Repeat the audit in ninety days, and compare the second column. The measure of progress is not fewer complaints. It is more lines where someone told you something in time.
Your page — The Embeddedness Audit.
A tool for a leader of a team of any size, or for a household. It measures what it costs, in practice, for someone to tell you something is wrong. Allow one hour to set it up and ten minutes a month to keep it.
Reading the ledger. Average cost per concern should fall over three months. The number of concerns may rise; that is the measure working, not failing. If the missing voices start to appear, you have lowered the price of voice where it was highest. If cost stays high and concerns fall, you are moving toward silence, and the next exits will tell you what the concerns would have.
Your page — The Voice Cost Ledger.
A one-page map of where your organisation or household meets the world, and how it senses what is happening there. Allow ninety minutes the first time. It can be drawn on paper or a whiteboard.
Reading the map. A healthy membrane shows mostly Vs, a few Es on relationships that are rightly arm's-length, and few or no Ss. Silences are the most important marks on the page, because they are the places where voice once worked and stopped. Redraw the map every six months. A map that has more Vs and the same or fewer departures is a membrane that is listening.
Your page — The Membrane Map.
For any leader responsible for buying from other organisations, from a team lead with three key vendors to a director with a supply base of hundreds. It sets up a voice channel that is kept apart from price. Adapt the word supplier to partner, contractor or provider as fits.
Reading the results. In the first year, expect the number of problems raised to rise. That is the protocol working, not the supply base getting worse. Over two to three years, the measure that matters is whether problems are being found earlier, in design and planning rather than in production and with customers. For a household, the same protocol shrinks to one question asked of the three or four providers you rely on most, once a year: what should we know that we have not asked you about?
Your page — The Supplier Voice Protocol.
A tool for any leader of a team, department or organisation, and in a smaller form for a household. It finds where your people can stop a process when they see a fault, and what happens to them when they do. Allow two hours for the first inventory and thirty minutes each quarter after.
Reading the inventory. Over a year, the number of pulls should rise and then settle, and the harm the processes cause should fall. A process where pulls stay at zero is not necessarily safe; it may be silent. A process where pulls rise and the people who pull them are thanked is a process with a voice. Repeat the inventory each quarter, and compare step four. The measure of progress is not how few faults you have. It is how early you hear about them, and from whom.
Your page — The Stop-Right Inventory.
A tool for any leader who depends on other organisations, from a manager with five key vendors to a director responsible for a whole supply base. In its household form, it maps the people a family could call on in a bad week. Allow two hours the first time; repeat once a year.
Reading the map. A healthy map has rim lines around every square and bad-day answers that name other spokes. A map that is all spokes, however many there are, belongs to an organisation that will face its first fire alone. Redraw it each year and count two things: squares without rim lines, and bad-day answers that name someone else on the page. The first should fall. The second should rise.
Your page — The Rim Map.
For a leader responsible for any group of people who depend on the organisation and cannot easily leave it: patients, residents, tenants, small dependent suppliers, staff with few alternatives, members locked in by contract. Allow two hours the first time and one hour each quarter after.
Reading the review. Over a year, the measure is not fewer complaints. It is shorter paths in step 2, more recurring themes acted on in step 3, more staff voices in step 4, and at least four quarters in which the organisation spoke first. A captive group that hears from you before it has to complain has a membrane that is beginning to listen.
Your page — The Captive Voice Review.
One sheet, kept by the convenor, read at every gate.
Reading the plan. At week thirteen, look for four things: the Monday sentence never missed, or missed honestly; median response time lower than at week seven; at least some S marks turned to V; and the number of concerns raised higher than at the start. If all four hold, the quarter worked. If concerns fell while response times rose, the organisation is moving toward silence, and the next quarter begins again at week one.
Your page — The Thirteen-Week Plan.
One sheet, kept weekly by the convenor, read monthly by the sponsor and quarterly by the second reader.
Reading the sheet. After two quarters, the measures should show a rising or stable count of voices with a falling median response time and a falling number of linked pairs. Valued exits should be no higher as a share of all exits. A long silence column that is shrinking means the membrane is reaching further. If every number improves at once in the first month, check step 4 before believing it.
Your page — The Voice Measures Sheet.
One card, index-card size, prepared before any of the five conversations in this chapter. Allow twenty minutes to write it. Keep it, and look at it again a month later.
Reading the card. A card whose promise was kept, and whose back records a second conversation the other person started, describes a channel that has opened. A card whose promise was missed is the most useful card in the set: it shows where the next conversation has to begin, with the miss, said first and plainly.
Your page — The Conversation Card.
For one person living alone, a couple, or a whole family. Allow ninety minutes the first time, at a table, with paper, ideally at a calm hour rather than after a difficult letter. Repeat every three months. It is not a budget, though it sits well beside one, and it gives no financial advice. It maps which of your household's relationships can carry a warning in time.
Reading the map. Count three things each quarter: lines marked portal only, lines marked silent, and unopened letters. The first two should fall over a year for the relationships that carry your income, your home and your health. The third should be small and should never be the same item two quarters running. A household whose most important relationships have names in them, and whose members can say a worry aloud at the table, has a membrane that listens.
Your page — The Household Membrane.
Fifteen minutes, once a week, for a leader or a household. It keeps the rhythm of exit and voice visible so that neither phase stops unnoticed. Keep the answers in one notebook or one document, dated, so that the weeks can be compared.
Reading the notebook. After a month, count three things: exits marked unknown, voices marked nothing yet, and linked pairs. All three should fall over a quarter. If voices rise while exits fall, the rhythm is working: people are speaking before they leave, and you are hearing them. If exits rise while voices fall, the organisation or household is moving toward silence, and the next step is the audit from chapter 2, done in full.
Your page — The Weekly Exit and Voice Review.
Thirteen weeks, three months, one quarter. It follows the field manual in chapter 9 and puts every tool in this workbook in its place. A leader runs the left-hand line; a household runs the right-hand line, which is shorter. Allow about two hours a week for the leader's line and about one hour a week for the household's, more in the first fortnight.
Days 1 to 7 · See the edge. Leader: run the Embeddedness Audit in full. Choose the one line where an opened channel would matter most, and make the one call within the week. Name a sponsor, a convenor and a second reader, as chapter 9 describes. Household: run the Household Membrane, steps 1 to 3. Hold the first monthly money conversation, using the words in chapter 11.
Days 8 to 14 · Find the price of speaking. Leader: build the Voice Cost Ledger from the last ten concerns. Mark the missing voices. Household: write down the last five times someone at home raised a worry, and what happened next. Note who has not raised one.
Days 15 to 21 · Draw the map. Leader: draw the Membrane Map with two or three colleagues who see different edges of the organisation. Circle the held breaths. Household: finish the Household Membrane. Choose one provider to keep closely and one to let go well.
Days 22 to 28 · Write the cards. Leader: write a Conversation Card for each of the three hardest conversations the map points to, usually a board, a staff group or representative, and one supplier or partner. Household: write one card for the conversation you have been postponing.
Days 29 to 42 · Open the channels. Leader: hold the three conversations. Start the Supplier Voice Protocol with the five most important suppliers or partners. Begin the Weekly Exit and Voice Review, and do not miss a week. Household: hold the postponed conversation. Ask the three or four providers you rely on most the one question: what should we know that we have not asked you about?
Days 43 to 56 · Keep the promises. Leader: reply to every concern within the deadline you set. Show the log to whoever you promised it to. Start the Voice Measures Sheet. Household: keep the monthly money conversation. Thank every concern raised at home before answering it.
Days 57 to 70 · Read what you hear. Leader: run the case-chapter tools where they fit. The Captive Voice Review for any group that cannot easily leave you: patients, tenants, pupils, long-contract customers. The tools from chapters 6 and 7 for the line and for the network. Household: run the Weekly Exit and Voice Review at the kitchen table for four weeks.
Days 71 to 84 · Count honestly. Leader: bring the Voice Measures Sheet to the second reader. Ask the four questions in chapter 10. Expect voices raised to have risen; read that as the measure working. Household: compare this month's money conversation with the first. Note what was said that would not have been said in the first.
Days 85 to 90 · Decide. Leader: redo the Embeddedness Audit and compare the second column with day one. Make the week-thirteen decision in chapter 9: continue, widen, or stop and say why. Write it down and tell the people who spoke. Household: redraw the Household Membrane. Keep what worked. Set the next quarter's first Sunday.
Reading the ninety days. The measure of the quarter is not fewer complaints. It is more lines on the audit where somebody told you something in time, fewer exits marked unknown, and at least one conversation that the other person started.
The argument behind every tool here is set out at full length in Ammanuel Santa Anna's Ecstatic Economics , kept with the rest of the series at hautelumiere.life.
Every market tells you how it is doing in two ways: somebody leaves, or somebody speaks. The second can be heard a season before the first is counted.
Luminous Prosperity Inc. takes this work into organisations as a gainshare. The flow is opened from the inside, with the people who already hold what the firm knows. The gain is counted from records the firm already keeps, against a measure both sides sign before anyone knows the answer, and the house is paid as a share of what is found: between thirty and ten per cent of it.
This book teaches the hearing. What your own market is saying, and what it would be worth to hear it a season sooner, is read from your records and your people. That begins with a conversation.