Chapter Two: The Infrastructure We Built and Then Dismantled

There is a particular kind of forgetting that happens so gradually that nobody notices it is underway. Not the sharp forgetting of loss, the cleared desk, the locked door, the eulogy, but the slow forgetting of accumulation, where each individual disappearance seems too small to mourn until one day you look around and find that the entire landscape has changed, and nobody organised a protest because nobody could pinpoint the moment it started.

This is how we dismantled the social infrastructure of the twentieth century.

We didn't do it in a day. We didn't do it by design. We did it through ten thousand individually rational decisions, to close a pub that wasn't turning a profit, to defund a community centre that wasn't in the budget, to merge two churches whose congregations had dwindled to the elderly and the committed, to let a union hall go dark because the union itself was a shadow of what it once was. Each decision made sense in isolation. The cumulative effect was the quiet erasure of the architecture that had held communities together for a hundred and fifty years.

To understand where we are, and why the arrival of AI at scale matters so much more than a routine technological disruption, you have to understand what we had. And then you have to understand, in some detail, what we lost.

The Factory as Society

Begin in the middle of the nineteenth century, when the industrial revolution did something that had never quite been done before: it organised social life around work.

This is easy to state and difficult to fully appreciate, because we are so accustomed to the primacy of work in modern life that we have forgotten it was a historical novelty. Before industrialisation, most people lived and worked in the same place, the smallholding, the workshop, the market town. Community preceded employment. You belonged to a village before you belonged to a trade. Your social identity was rooted in geography and kinship long before it was rooted in occupation.

The industrial revolution reversed this. As the factories drew workers from the countryside into the mill towns of the north of England, the textile districts of the Midlands, the mining valleys of South Wales and County Durham, and later the steel towns of Sheffield and Motherwell, it created a new kind of community, one organised entirely around production. The factory was the gravitational centre around which everything else arranged itself.

And the factory owners knew it. Whether from paternalism, self-interest, or some combination of both, the great industrial employers of the Victorian and Edwardian periods built something much more comprehensive than workplaces. They built social worlds.

Titus Salt built Saltaire: a model village outside Bradford, completed in 1876, containing 850 houses, a church, a school, a hospital, almshouses, baths and washhouses, a library, a mechanics' institute, a park, and a boating lake. The entire infrastructure of a self-contained community, funded and administered by a single employer. Salt's workers didn't just earn a wage at the mill, they lived in its houses, educated their children in its schools, and worshipped in its church.

Port Sunlight, built by Lever Brothers from 1888 onward on the Wirral Peninsula, provided 3,500 residents with similarly comprehensive provision: schools, a church, an art gallery, a cottage hospital, a theatre, a gymnasium, and some of the finest workers' housing in Britain. Bournville, built by the Cadbury family outside Birmingham from 1895, added parks, swimming pools, sports grounds, and eventually a further education college. These weren't marginal amenities. They were the operating system of an entire community.

The mining industry took this model and extended it across hundreds of communities. The Miners' Welfare Fund, established in 1920 with a levy on coal production, financed an astonishing range of social provision: libraries, institutes, sports grounds, swimming baths, recreation halls, convalescent homes. By the 1930s, the miners' welfare institutes of the South Wales valleys were among the most significant cultural institutions in working-class Britain, places where a miner finishing a shift could attend a lecture on philosophy, borrow a volume of poetry, rehearse with the colliery brass band, or simply sit in a warm room and read a newspaper among men who understood exactly what his working life entailed. These were not incidental benefits. They were part of the deal: the implicit social contract that the industrial economy had struck with the people who made it work.

The working men's club movement was perhaps the most remarkable expression of this order. Beginning in the 1860s as temperance-adjacent alternatives to the pub, the clubs quickly evolved into something much more ambitious: mutual aid societies, entertainment venues, political education centres, and above all community anchors. By the mid-twentieth century, the Club and Institute Union represented over four thousand affiliated clubs across Britain, with a combined membership approaching four million. These weren't venues people attended occasionally. They were institutions people belonged to, sometimes across generations, sometimes for a lifetime. A man might join his local club at twenty-one, serve on its committee at forty, drink at its bar at eighty, and be toasted at his send-off in its function room when he died.

The union hall completed the picture. From the 1880s to the 1970s, the trade union movement provided not merely industrial representation but social infrastructure of extraordinary breadth. The Transport and General Workers' Union, at its peak, was not just a collective bargaining organisation, it was a network of meeting rooms, legal advice centres, educational programmes, holiday schemes, and mutual support systems. The NUM, the AUEW, the TGWU, the NUR: these organisations gave their members not just leverage in the workplace but structure in the rest of their lives. When you were a union member, you had people to call when things went wrong. You had a network that extended beyond your immediate family. You had a community of interest that met regularly and knew your name.

The effect of all this, taken together, was to make work the master organising principle of social life. You got your income from your employer. You got your housing, often, from your employer. You got your sport from the company football club. You got your healthcare from the sick club. You got your learning from the mechanics' institute. You got your entertainment from the working men's club and the union-subsidised trip to Blackpool. You got your solidarity from the union. You got your identity, your sense of who you were in the world and where you stood, from the occupation you shared with the people around you.

This was not an ideal arrangement. The dependency it created was real, and paternalistic employers could use it as a form of control as easily as a form of care. The company towns of Pennsylvania, Pullman, Illinois, being the most famous example, showed how thoroughly a workforce could be managed through the totality of its social provision. The system produced solidarity and it produced deference. It created community and it created captivity. It was not utopia.

But it was structure. And structure, it turns out, is what human beings require to build and sustain social connection.

The Gradual Unmaking

By the early 1970s, this industrial social architecture was already beginning to fray. The deindustrialisation of the British economy, the decline of coal, steel, textiles, and heavy manufacturing that had sustained the old working-class communities, did not just remove jobs. It removed the entire social ecosystem that had been built around those jobs.

When the colliery closed, the miners' welfare institute lost its funding. When the factory shut, the company sports ground fell into disrepair. When the union was broken or simply dwindled, the union hall stood empty or was sold for conversion into flats. The physical infrastructure persisted for a while, the buildings were still there, the clubs were still open, but the economic foundations that had sustained them had been pulled away.

The working men's clubs, which had seemed like permanent features of the social landscape, began their long decline. Membership fell, then fell further. The demographics of the remaining membership aged upward. The younger generation, either mobile enough to leave the old communities or employed in service industries that didn't generate the same collective identity, didn't join. Four thousand clubs had become three thousand, then two thousand, then fewer.

This was happening, crucially, not only in the former industrial heartlands. Across the entire economy, the social infrastructure that had been constructed around collective working life was becoming obsolete. The shift to service industries, to white-collar work, to individual rather than collective employment relationships, to short-term contracts and portfolio careers, all of this dissolved the bonds that had connected work to community.

The pub, for all its differences from the union hall, had served a complementary function. It was the third place, not home, not work, but the regular gathering point where the community assembled, where strangers became familiar, where the rituals of shared life were enacted over a pint and a conversation. In 1979, there were approximately 70,000 pubs in the United Kingdom. By 2024, that figure had fallen below 39,000 for the first time in modern record-keeping, a loss of more than 14,000 establishments in the preceding fifteen years alone, and nearly half the total estate over the longer arc of decline. Each closure was reported as a business failure. The aggregate was the removal of one of the primary mechanisms through which community had reproduced itself.

The local authority cuts that followed the financial crisis of 2008, and the subsequent decade of austerity, dealt a further blow to what remained. Community centres that had survived deindustrialisation were shuttered. Libraries that had been the last free public gathering spaces in many towns had their hours cut and their staff reduced. Leisure centres were sold, privatised, or simply closed. The public social infrastructure, the part that had been funded by collective provision rather than employer provision, was dismantled in the name of fiscal necessity.

It was not, to be clear, a conspiracy. It was the aggregate effect of ten thousand local decisions made by people under financial pressure doing what seemed rational in the circumstances. The community centre in this particular borough wasn't full enough to justify the heating bill. The library in this ward could be run at lower cost if it reduced opening hours. The leisure centre in this district could be sold to a private operator who would keep it open, for a while, for those who could afford it.

Each decision was individually defensible. Together, they amounted to the systematic withdrawal of the public social infrastructure that had replaced the industrial social infrastructure as the latter declined.

Bowling Alone: The Evidence Assembled

In 2000, Robert Putnam, the Harvard political scientist, published the book that named what was happening.

Bowling Alone had been preceded, in 1995, by a journal article of the same title, an unusual piece of academic writing that attracted enough public attention to generate a formal rebuttal from the White House and an invitation to Camp David from President Clinton. The book expanded the argument with five years of additional data. Its thesis was simple and devastating: the social capital of American society, the networks of trust, reciprocity, and civic engagement that make communities function, had been in systematic decline for three decades.

Putnam's title came from a specific observation about a specific activity. Bowling, as a pastime, was more popular in the United States in 2000 than it had been in 1965. But bowling leagues, the organised social form of the activity, in which people bowled together on a regular basis, as members of a group, had collapsed. More bowling was happening. Less belonging was being built. The social glue that had made bowling into community had dissolved, leaving only the individual act.

He measured the decline across seven categories: political participation, civic participation, religious participation, workplace connections, informal social connections, mutual trust, and altruism. In every single category, without exception, the trend line ran in the same direction. Americans were signing fewer petitions. Joining fewer organisations that actually met in person. Attending fewer community meetings. Knowing their neighbours less. Meeting with friends less frequently. Trusting strangers less. Volunteering less. Even spending less time in informal socialisation with their own families.

"By virtually every conceivable measure," Putnam wrote, "social capital has eroded steadily and sometimes dramatically over the past two generations."

He was careful about causation. He considered and largely set aside the pressures of time and money, the increasing participation of women in the workforce, suburban sprawl, and generational change. His primary villain, the one he felt most confident attributing causal weight, was television. The privatisation of leisure time through the television set, he argued, had done more damage to civic participation than any other single factor, because it had given people a way to spend their evenings that was solitary, passive, and non-reciprocal, while feeling entertaining enough not to seem like deprivation.

He did not know, writing in 2000, that television was merely the first draft of a far more powerful technology. The smartphone, which Putnam had no occasion to consider, because it did not yet exist, would prove to be television's destructive heir: a device more intimate, more constant, more addictive, and more effective at replacing the activity of social participation with its simulation.

The twenty-five-year retrospective published in Fortune in December 2025 noted that the crisis Putnam had identified had become "starker than ever" in the intervening quarter-century. The organisations he had already been mourning, the Elks Club, the Rotary, the PTA, the bowling league, had continued their slide toward irrelevance. The social capital metrics he had tracked had continued to deteriorate. If Bowling Alone had been a warning, the warning had not been heeded.

The Concrete Losses

What does a social capital deficit actually look like? Not in the abstract, not in the regression analyses and the survey data and the trend lines, but in the granular, physical reality of communities across Britain and America.

It looks like 14,000 fewer pubs. It looks like the Red Lion in a market town that hasn't changed its sign or its opening hours in sixty years, suddenly dark at 8pm on a Wednesday because the landlord is trying to manage the energy bill. It looks like the social club that has been in the same building since 1923, finally voting to sell the site for flats because there aren't enough members left to sustain it. It looks like the community centre that was once the venue for amateur dramatics productions, parent and toddler groups, Alcoholics Anonymous meetings, and the over-sixties lunch club, now being used as a food bank distribution point, which is itself a measure of something having gone badly wrong.

The Church of England reported that weekly attendance had halved since 1980, falling from approximately 1.1 million to under 600,000 in 2023. These numbers represent something more than a theological trend. A church congregation, at its functioning best, is a mutual aid society that meets every week. When someone dies, the congregation organises. When someone has a baby, the congregation marks it. When someone falls ill, the congregation brings food. When someone moves to a new town, the congregation provides an immediate social network. The decline of organised religion is simultaneously, separately from its spiritual dimension, the loss of one of the most effective mechanisms for maintaining social connection that Western societies have ever produced.

Trade union membership tells a parallel story. In 1979, at its peak, British trade union membership stood at approximately 13 million, representing more than half the employed workforce. By 2024, that figure had fallen to under 6 million, and concentrated overwhelmingly in the public sector. The union hall in the manufacturing town was not merely a bargaining unit. It was a place of assembly, education, mutual support, and social solidarity. Its decline removed not just industrial protection but social infrastructure.

The civic clubs that Putnam had already identified as declining in 2000, Rotary, Lions, Round Table, Soroptimist, continued their retreat across Britain and America alike. These organisations, at their peak, had provided a regular, reliable, cross-class gathering point for professionals and businesspeople who might otherwise have had no occasion to meet outside their immediate circles. They ran charitable projects, they mentored young people, they connected local business communities, and, most importantly for our purposes, they gave their members somewhere to be, among familiar faces, on a predictable schedule. Their decline was unremarked upon in the mainstream press, because they were perceived as old-fashioned. Their social function was not replaced.

Each of these losses, the pub, the club, the church, the union, the civic organisation, had the same structural effect. Each was a regular, reliable reason to be among other people: people who were not your family and not your colleagues, but who were familiar, who knew your name, who would notice your absence. Social scientists call these weak ties: connections that are not the primary bonds of family or close friendship, but that play a disproportionate role in maintaining a sense of social integration and community. The sociologist Mark Granovetter, in work that predated Putnam by three decades, had shown that weak ties are often more important than strong ones for the actual functioning of social life. They are how you hear about a job. How you learn about a new doctor. How you navigate a crisis without having to mobilise your innermost circle. And they require, to be maintained, regular, low-stakes physical proximity. They require places.

We removed the places.

The Technological Substitution That Wasn't

It is worth pausing, before we get to the technologies that were supposed to save us, to appreciate the magnitude of what had already been lost. The pub, the working men's club, the union hall, the miners' welfare, the civic organisation, these were not merely places to socialise. They were, in a technical sense that sociology has tried hard to formalise, third places: spaces that were neither home nor work, and that performed social functions that neither home nor work could replicate.

Ray Oldenburg, the American sociologist who coined the term in 1989, described third places as "the core settings of informal public life", the places where people gather without agenda, without appointment, in the kind of relaxed, low-stakes setting that allows strangers to become familiar and the familiar to become close. The church on Sunday morning. The pub on Thursday evening. The club on Friday night. The café where the same people appear at the same time every day and eventually cannot imagine starting their morning any other way. Third places have two essential qualities that home and work cannot provide: they are open to all-comers from the surrounding community, and they require nothing more than your presence to confer membership. You don't apply. You don't have to perform. You simply show up, regularly enough, and you become part of the fabric.

Oldenburg was already troubled, writing in 1989, by the extent to which American life had reorganised itself around home and work, leaving little room for the third place to function. By 2024, his concerns looked like understatement.

Social media arrived, in the mid-2000s, with a promise so seductive that it took a full decade for its failure to become undeniable.

The promise was connection at scale. Facebook, launched in 2004, would let you stay in touch with everyone, school friends, cousins, former colleagues, neighbours, without the friction and effort that real-world maintenance required. Twitter would give you access to the public conversation. LinkedIn would make professional networking frictionless. Instagram would let you share your life with the people who cared about it. And all of them were free, which was to say, their cost was paid in attention and personal data, not in money, and therefore the transaction was nearly invisible.

What they delivered was something Sherry Turkle had anticipated and named before most people were paying attention.

Turkle, the MIT social scientist who had spent three decades studying the relationship between technology and human psychology, published Alone Together in 2011, when the smartphone revolution was in its early stages but already producing measurable effects. Her thesis, arrived at through years of ethnographic observation rather than survey data, was that networked technology did not deepen connection. It substituted for it. "We sacrifice conversation for mere connection," she wrote. The distinction was precise and important. Connection, a notification, a like, a text, a comment, creates the neural stimulation of social contact without its substance. It activates the social reward pathways of the brain while withholding what those pathways actually require: sustained, mutual, embodied, reciprocal exchange with another consciousness.

The ambiance of social media, the sense of always being surrounded by the chatter of others, always on the periphery of many conversations, is not the same as being present in a community. It is, in Turkle's formulation, being alone together: surrounded by the digital representation of social life while experiencing none of its actual benefits.

The scale of the substitution became visible in the data. The percentage of Americans who reported having no one with whom they could discuss a personal matter tripled between 1985 and 2004, the period precisely corresponding to the decline of the third places and the rise of the internet as a domestic medium. The percentage who reported having only one such confidant rose sharply. Americans were not becoming more connected. They were becoming more isolated, more quickly, while appearing to interact more intensively than ever before.

The progressive elimination of reasons to leave the house accelerated this dynamic in ways that went beyond social media. Streaming services made the cinema redundant for most viewing. Online shopping made the high street redundant for most purchasing. Delivery apps made the restaurant visit optional. Home fitness platforms made the gym a choice rather than a necessity. And then, arriving too recently for Putnam to have anticipated and too fast for most institutions to have processed, remote work made the office optional for perhaps thirty percent of the working population.

Consider what the office had been, beneath its productive surface. It had been a place where you were among other people every day. Where you learned names and faces and personal histories. Where you navigated the small negotiations of shared space. Where you developed, over time, the kind of low-stakes familiar relationships, the colleague you got coffee with, the person you complained to about the photocopier, the team you ate lunch with on Fridays, that constitute the weak tie fabric of a social life. The office was not, for most people, a source of deep friendship. But it was a source of daily human texture that, for many people living alone or in small households, constituted a significant portion of their total social contact.

Remote work removed that. The video call gave you the face and the voice. It did not give you the walk between the desk and the kitchen. The overheard conversation. The person who noticed you looked tired and asked if you were all right. The embodied, incidental, unremarkable being-among-people that had been the daily background of working life for generations.

The screen gave you everything except what you actually needed.

The Gen Z Paradox

The generation that grew up with smartphones, born, by most researchers' reckoning, from around 1997 onward, with full smartphone saturation arriving around 2012, represents the most instructive case study in what happens when digital connectivity is offered as a substitute for social infrastructure from the very beginning.

Jean Twenge, the social psychologist at San Diego State University who has tracked generational change in American adolescents over several decades, documented the inflection point with characteristic precision. In iGen, published in 2017, she showed that the generation that came of age with the smartphone was fundamentally different from the Millennials who had preceded them, not in the gradual way that each generation differs from the last, but in the sharp, sudden way that suggested something had changed in the environment rather than merely in the culture.

The chart was clear. From 2012 onward, the rates of adolescent depression, anxiety, and self-reported loneliness began to rise, sharply, consistently, and disproportionately among girls. The same period showed a steep decline in face-to-face socialising, in the number of teenagers who reported spending time with friends in person, in the proportion who went to parties, dated, or simply hung out. They were, paradoxically, spending more time "connected", on their phones, in their rooms, on platforms designed to show them the social lives of others, while spending less time in actual social situations.

The American Psychological Association's 2024 data showed that more than 60 percent of US adults reported elevated stress from social fragmentation, the sense of disconnection from community, from shared purpose, from the social fabric that gives life meaning and texture. But the figures for young adults were consistently higher than for any other demographic. Generation Z, the generation that had never known a world without the internet, that had been socialised on platforms rather than in the street or the club or the youth group, reported the highest loneliness scores of any age group, in surveys conducted across the English-speaking world.

This is the paradox that demands explanation, because it is so precisely opposite to what anyone would have predicted. The most connected generation in history, connected by every technological measure, with access to more tools for communication, more channels of social interaction, more means of reaching other people than any previous cohort, is the loneliest.

The explanation lies precisely in the substitution. Digital connectivity, as Turkle diagnosed, does not provide what social participation provides. It provides something that resembles social participation at a surface level, the engagement, the response, the stimulation, while withholding the depth, the mutuality, the physical presence, the accumulated shared experience that turns mere contact into belonging. For a generation that was offered the simulacrum before it had fully experienced the original, the substitution was total. They learned to manage their social needs through the screen before they had developed, in the way that previous generations had developed through the physical social infrastructure of youth clubs, school societies, local sports, and neighbourhood life, the habits and expectations of embodied community.

The result was a generation that is, in a precise clinical sense, socially malnourished: fed on the calories of digital connection without receiving the nutrition that human beings actually require. The rise in depression and anxiety among Gen Z is not incidental to their digital lives. It is, the evidence increasingly suggests, causally connected to them, not because social media is evil in any simple sense, but because it replaced something that could not be replaced with something that could not do the replacing.

The mechanisms are now reasonably well understood. Social comparison on image-based platforms produces measurable increases in anxiety and depressive symptomatology, particularly among adolescent girls. The intermittent reinforcement schedule of social media, the unpredictable arrival of likes, comments, and shares, produces engagement patterns that neurobiologically resemble addiction. The displacement of time spent in unstructured, face-to-face socialising reduces the opportunities to practise and develop the social competencies, reading emotional cues, managing conflict, sustaining reciprocal conversation, that embodied social life requires. And the constant availability of the screen as a retreat from social discomfort means that the low-grade anxiety of new social situations, the anxiety that, in previous generations, was managed through repeated exposure, is never worked through but simply avoided.

Cigna's longitudinal research found that 79 percent of Gen Z reported feeling lonely sometimes or often. That is not a marginal statistic. That is almost four in five members of an entire generation telling you that despite, or because of, their unprecedented access to connection, they do not feel they belong.

There is a particular cruelty in this finding that deserves naming. Previous generations experienced the loss of belonging as a subtraction, something that had been present was removed. Gen Z is experiencing it as an absence: they grew up in a world where the old infrastructure was already gone, where the digital substitute was the only model on offer, and where the gap between what they were getting and what they needed was never visible to them because they had no prior experience of the difference. You cannot mourn a library you have never seen. You cannot miss a union hall you never knew existed. What you can do, what the data shows Gen Z doing, with considerable precision, is feel the effects of the absence without having language for its cause.

The Ground State

Step back from the individual data points and look at the whole picture.

What we have built, over the last sixty years, through the combined effects of deindustrialisation, social change, technological substitution, and institutional retreat, is a society operating on a structural belonging deficit. The mechanisms that produced community, the factory and its associated social world, the union hall, the church, the civic club, the local pub, the community centre, have been removed, one by one, without replacement. The attempt to replace them digitally has produced the opposite of the intended effect. And the generation that inherited this depleted landscape has not found a way to rebuild it. They are, instead, experiencing the consequences of its absence in the form of the highest recorded loneliness scores in the history of modern survey research.

This is the ground state. This is not the crisis about to arrive. This is the crisis already underway.

To summarise the concrete evidence: in the United Kingdom, the number of pubs fell below 39,000 in 2024 for the first time in modern record-keeping, down from 70,000 in 1979. Trade union membership fell from 13 million at its peak to under 6 million. Church attendance halved since 1980. Community centres were shuttered across local authority estates. The working men's clubs that had once served four million members numbered a fraction of their former selves. In the United States, church membership fell below 50 percent for the first time in eight decades of Gallup tracking. Civic club memberships continued to decline across every category. The number of Americans reporting no close confidants tripled over two decades.

The WHO's Commission on Social Connection, publishing in June 2025, identified loneliness as a global health emergency killing 871,000 people per year. The US Surgeon General had already declared it a public health crisis comparable in mortality impact to smoking fifteen cigarettes a day. The evidence trail, from Putnam's original mapping in 2000, through Twenge's generational analysis, through Turkle's diagnosis of digital substitution, through the WHO's mortality data, describes a single, consistent phenomenon: the systematic removal of the social infrastructure that human beings require to function.

And now, against this backdrop, a society already running at social capital deficit, already experiencing the consequences of institutional collapse, already watching one generation demonstrate what happens when the substitution is total, comes a further disruption of the one remaining structure that, for a substantial portion of the working population, still provides what all the others once provided.

The office. The daily gathering. The obligatory being-among-people. The place where you were not alone.

Artificial intelligence, as the following chapters will argue, is not merely a productivity technology. It is the final pressure on the last reliable source of daily human structure that modern life still routinely provides. As remote work already weakened the social function of work, AI-driven automation threatens to remove the work itself for large segments of the workforce, the same segments that Putnam identified as already experiencing the steepest decline in every other form of social participation, and that Cox identified as the most socially isolated before any technological disruption arrives.

The infrastructure of belonging was built across a century and a half of industrial social life. It was dismantled across sixty years of economic transformation, institutional retreat, and technological substitution. What remains is thinner and more precarious than most people appreciate. And what is coming will stress it further.

Before we can understand what comes next, we need to understand this: the belonging crisis did not begin with AI. AI is arriving into a landscape that was already depleted. The question is not whether we will experience a crisis of connection. That crisis is already recorded in every dataset, visible in every survey, manifest in the loneliness scores of the most technologically connected generation in human history.

The question is whether, in the decade ahead, anyone builds the replacement infrastructure in time, and what that infrastructure might look like.

That is what the remainder of this book is about.