8 Change
What actually produces change?
- Describe the conditions under which social movements succeed.
- Explain a “policy window” and assess whether one is open.
- Connect a specific social-media or AI harm to each of the day’s six prior lessons.
- Sketch a realistic reform strategy and name their own leverage as a technologist.
Name one thing that is better now than it was fifty years ago. Now ask who made it better, and how. Gay marriage is legal in more places than it was, the ozone hole is closing, cars have seatbelts, and far fewer people die in factory fires than did in 1911. None of these things happened by themselves, and none of them happened because the people with power were persuaded. They happened because organized groups applied sustained pressure over time, and they rarely happen any other way.
8.1 How Change Happens
The historical record is consistent across very different contexts. Change happens when three things come together: organization, opportunity, and framing. The civil rights movement, environmental regulation, seat belts, tobacco control, marriage equality, and worker safety all followed this pattern— slow, non-linear, and won by pressure rather than persuasion alone. The popular story that Rosa Parks was a spontaneous individual is factually wrong: she was the secretary of the Montgomery chapter of the NAACP, had recently attended a training center for organizers, and was chosen as a plaintiff deliberately (Beckerman 2022). The bus boycott that followed was coordinated by hundreds of people whose names are not remembered.
John Kingdon gave the pattern a precise description (Kingdon 1995). Change becomes possible when three streams align: a problem defined as a crisis, a politics stream with a constituency and a mood, and a policy stream containing a ready-made solution. The window opens rarely and closes quickly, so the people who want change must have built the solution before the problem becomes undeniable.
Three mechanisms illustrate how this works in practice. ACT UP, founded in 1987, used direct action to accelerate drug approval, targeting the FDA and pharmaceutical companies until they changed their timelines— a demonstration that the process of regulation itself can be shifted by organized pressure. The abolitionist movement distributed slave narratives such as Olaudah Equiano’s 1789 narrative and Frederick Douglass’s 1845 memoir, making suffering imaginable to readers who had never seen it, and ran the 1790s sugar boycott as the first mass consumer boycott in British history (Hochschild 2006). And Kerala, in southern India, won land reform in 1969 not through occupation but through election, breaking the veto of the landlord class and unlocking decades of gains in literacy and life expectancy (Drèze and Sen 2013).
There is also a quieter side to the record. Erica Chenoweth found that campaigns engaging the active participation of roughly 3.5 percent of a population have repeatedly been enough to force change (Chenoweth and Stephan 2011). In a city of half a million people, that is seventeen thousand people doing something instead of just complaining. And the capacity for that something is real. After the 1917 Halifax explosion and Hurricane Katrina, ordinary people organized mutual aid with remarkable speed and generosity, even as authorities and media reported violence that later investigation showed never happened (Solnit 2009; Quarantelli 1998). The capacity is there; the question is whether it is organized.
8.2 Worker Power
Worker power is the clearest example of gains that were won, not conceded. The Combination Acts criminalized unions in Britain; the Tolpuddle Martyrs were transported to Australia for forming one; the 1926 General Strike was defeated; and Thatcher spent a decade breaking what remained. The eight-hour day did not arrive because employers became convinced workers deserved it; it arrived after the First World War, when the fear of revolution made the cost of continued resistance higher than the cost of concession. The Melbourne stonemasons won the first eight-hour day in 1856 not through moral persuasion but by applying collective leverage at a moment of their opponents’ weakness.
The demand was older than the Melbourne victory. Robert Owen had called for “eight hours labour, eight hours recreation, eight hours rest” in 1817, but Owen could not make it happen through goodwill and argument. It took workers who understood that argument only mattered when backed by the capacity to impose costs. In May 1886, a general strike for the eight-hour day shut down factories across the United States. On May 4, a bomb exploded in Haymarket Square during a protest in Chicago, killing both police officers and civilians. The bomb-thrower was never identified, but eight anarchists were tried anyway, and four were hanged. The Haymarket affair became a rallying point for the international labor movement; May Day was adopted as a workers’ holiday across most of the world, with the notable exception of the United States and Canada, which moved their Labor Day to September specifically to avoid the association. In 1889, the newly formed Second International declared May 1 an international day of solidarity for the eight-hour day. But spreading a demand and achieving it are different things. In most countries, forty years passed between the first serious campaigns and its legal establishment. What closed the gap was not sympathy but the end of the First World War, a political crisis severe enough that governments feared revolution. France legislated the eight-hour day in April 1919. Britain passed the Hours of Work Convention that same year. The first convention adopted by the International Labour Organization in 1919 was an eight-hour day and forty-eight hour week for industrial workers. Then and now, labor gains were not won because employers became convinced that workers deserved better. They were won because the cost of continued resistance exceeded the cost of concession.
The same lesson runs through the Triangle Shirtwaist fire. On March 25, 1911, a fire broke out on the eighth floor of a building in Greenwich Village. The managers had locked the stairwell doors to prevent unauthorized breaks and theft. 146 workers died, most of them young immigrant women, many Jewish and Italian. Some jumped from the windows. The owners were acquitted of manslaughter and collected the insurance. The conditions were standard practice: locked exits, no sprinklers, overcrowding, abusive supervisors. What was not standard was the combination of a catastrophic death toll and the organizational capacity that existed to exploit the political opening. The Women’s Trade Union League, the International Ladies’ Garment Workers’ Union, and the socialist press had spent years building that capacity. Within days, Rose Schneiderman gave a speech that reframed the debate: this was not a tragedy but a crime, and the perpetrators were not the invisible people who set fires but the visible people who locked the doors and blocked regulation. The result was a legislative investigation that produced fifty-six pieces of legislation over three years, covering fire safety, working hours, factory inspection, and child labor. The fire itself did not produce the laws; it was the fire plus organized political capacity.
The institutions that made these gains stick were deliberately built. German codetermination puts workers on corporate supervisory boards, half of them in large firms. Australia’s Harvester judgment of 1907 established a “basic wage” through a system of arbitration rather than bargaining alone. And COSATU, the South African trade-union federation, was a decisive force in ending apartheid. Australia’s Conciliation and Arbitration Act of 1904 built a labor system unlike either the British or German models. A federal industrial court, not collective bargaining, set wages and conditions through legally binding awards that applied across whole industries. The 1907 Harvester judgment established the “basic wage” as the minimum needed to support a family in frugal comfort. Union membership was high because belonging to a union was a condition of accessing the award system, not because workers had achieved industrial leverage through organizing. New Zealand took a similar path: compulsory arbitration from 1894, the eight-hour day for most workers in 1891, and old-age pensions by 1898, thirty-seven years before the United States got Social Security. These systems were dismantled during the 1990s under market liberalization, replaced by enterprise bargaining that produced wider variation in wages and conditions.
Employers have also learned how to prevent organizing. High individual compensation is the most direct approach: workers paid well relative to their labor-market alternatives have less to gain from collective bargaining. Union avoidance is a formal management discipline with a consulting industry behind it: labor relations firms advise employers on how to structure workplaces and compensation to prevent the conditions under which organizing becomes attractive. The legal boundary between employee and contractor is another tool: reclassifying workers as independent contractors removes them from the framework that protects organizing activity. The pattern persists today. When the Rana Plaza factory collapse killed 1,134 garment workers in 2013, it produced the Bangladesh Accord, a real gain, then weakened, then re-won— two steps forward, one step back.
8.3 The Technologist’s Role
Technologists hold unusual cards. Their skills are scarce, they have access to systems that others cannot see, and they carry inside knowledge of how the machinery actually works. They also face real constraints: contracts, career risk, and the ideology of neutrality that says an engineer’s job is to build, not to choose. Erin Cech calls the ideology the passion principle: “do what you love” is converted into a reason to accept extraction, and its Japanese endpoint is karoshi, death from overwork (Cech 2021). The passion principle falls hardest on workers from less privileged backgrounds, who have the most to lose from accepting its terms. Workers from more privileged backgrounds can absorb the costs of passion-coded careers: family wealth provides a cushion, networks provide alternative paths, and prior social capital makes it easier to exit without losing everything. For workers without those resources, the passion principle extracts years of discounted labor before returning them to the general labor market with depleted savings and credentials that do not transfer. Precarious workers like gig workers and seasonal workers are hit hardest. Organizing takes time they may not have, and requires a willingness to accept the risk of termination that workers without income security cannot absorb. Many are formally classified as independent contractors, which in most jurisdictions removes them from the legal framework that protects organizing. The legal boundary between employee and contractor has been contested in dozens of jurisdictions, with outcomes that vary and have been inconsistent even within jurisdictions. Over and over, rules are rewritten to ensure that the workers most exposed to poor conditions are least able to fight back (Jaffe 2021; Kelly 2022). The software-engineering literature is a symptom of the same blind spot: it covers architecture and management extensively but has almost nothing to say about workers’ rights or cooperatives, the way medicine once hid its own error rate. Medicine’s encounter with its own error rate is a case study in professional taboo. For most of the twentieth century, the dominant framework for medical error was individual blame: a bad outcome was the fault of a bad physician, and the appropriate response was to identify and punish the individual. This framework served the profession’s interest in self-regulation and resisted external oversight. The finding that between 44,000 and 98,000 Americans died annually from preventable medical errors was not new to researchers, but it was taboo because it endangered the profession as a whole. Aviation-style checklists, which treat error as a predictable product of complex systems, were resisted for decades by physicians who experienced them as an affront to professional judgment. The evidence that checklists reduced deaths was not in dispute. The resistance was to the implication that standardized protocols should constrain clinical practice. Software engineering has its equivalent silences. The books cover project management, architecture, team dynamics, and career development. They do not cover workers’ rights, noncompete agreements, ownership of employee inventions, alternative ownership structures, or the systematic harassment and exclusion of women. The topics placed out of bounds are those that, if examined, would reveal tensions in the profession’s self-understanding or power structure. The placement of a boundary is a form of power.
The leverage is not theoretical. In 2023, the Writers Guild of America struck for five months and won explicit contract language that AI cannot write or rewrite scripts, and that scripts cannot be used to train AI systems (Kelly 2022). No individual writer could have secured that alone. The Screen Actors Guild reached a parallel agreement that included restrictions on the digital replication of performers’ likenesses without ongoing consent. These victories established enforceable contractual limits on what employers could do with AI— limits that individual workers negotiating alone could never have secured. Regulation has also moved faster than the industry claims is possible. The EU AI Act requires transparency for high-risk systems, mandates human oversight for consequential automated decisions, and bans specific applications outright. Before the GDPR took effect in 2018, industry associations described it as unworkable and predicted that it would destroy European tech competitiveness. By 2024 it had generated approximately $4 billion in fines and had changed how companies worldwide handle personal data, including companies with no European operations that chose to comply rather than maintain two systems. Technologists can refuse, whistleblow, organize, build alternatives, or work on policy, and each of those is a form of pressure.
8.5 Alternatives That Scaled
The claim that alternatives cannot work at scale is false, and the counterexamples are large enough to have annual reports. In 1956, a Catholic priest named José María Arizmendiarrieta gathered five workers in a converted school in Mondragón, in the Basque region of Spain, and started making paraffin heaters. By 2024, the Mondragón Corporation employed roughly 80,000 people across manufacturing, retail, financial services, and education. Its workers own the enterprise. They elect the management, negotiate pay scales internally, and when Spain’s economy collapsed after 2008, Mondragón moved workers between cooperatives rather than letting them go (Morrison 1991).
In the 1860s, Friedrich Raiffeisen was running a rural municipality in what is now Germany and watching small farmers get crushed by moneylenders. He organized the farmers to lend to each other. Members pooled deposits, the cooperative extended credit at rates commercial lenders could not match, and the surplus stayed within the cooperative. Today the Raiffeisen network is one of Europe’s largest banking groups, and the World Council of Credit Unions estimates that roughly a billion people worldwide are members of credit cooperatives of one kind or another (Whyte and Whyte 1991).
Vienna in the 1920s faced a housing crisis that would look familiar to anyone paying rent in a large city today. The Social Democratic city government decided to treat housing as a public utility rather than a market good. Between 1923 and 1934, the city built roughly 60,000 apartments in large municipal complexes called the Gemeindebau. They were substantial brick buildings with courtyards, laundries, libraries, kindergartens, and clinics. The Karl-Marx-Hof, completed in 1930, is over a kilometer long. The program was funded by a progressive tax on luxury goods: people buying fur coats and racehorses paid for workers to have decent apartments. The city still owns about 220,000 apartments directly and subsidizes roughly another 200,000 through a nonprofit cooperative sector. About 60 percent of Vienna’s residents live in subsidized housing of one kind or another, and rents are set by law rather than by the market.
Amul is an Indian dairy cooperative founded in 1946 in Gujarat after farmers organized to break the monopoly of a single private contractor who controlled their access to the Bombay milk market. Today Amul is the world’s largest dairy cooperative by volume, collecting milk from roughly 3.6 million farmer-members. Canada’s Desjardins, founded in 1900 by a journalist named Alphonse Desjardins, is now one of the largest financial institutions in Canada, with roughly eight million members and assets over 400 billion Canadian dollars. Members own it, members elect the board, and surplus is returned to members. It survived the 2008 financial crisis in considerably better shape than many of the shareholder-owned banks.
8.6 Five Tools
The finale of this workshop offers an alternative taxonomy of the reform menu: five tools that have historically done the work. Collective organizing means building institutions that can act in coordination. Regulatory pressure means participating in the slow work of making rules. Alternative structures means building existence proofs that the current arrangement is not the only one: Mondragón has run as a worker cooperative for nearly seventy years, Vienna’s Gemeindebau still houses working-class families a century on, Amul coordinates 3.6 million farmer-members, and Desjardins serves roughly eight million members and survived 2008 in considerably better shape than the big banks. Strategic communication means engaging with the psychology of belief change, because facts do not change contested minds—stories told by trusted messengers do. And sustained participation means showing up for decades: the Federalist Society built its judicial pipeline over forty years, and the labor movement built the eight-hour day over the same span.
This is the hardest part of the message for people accustomed to fast feedback loops. No single action produces structural change; what produces it is organizational capacity built before the political opening arrives, and the willingness to use that capacity when the opening comes. Refusing to vote or join a union is a political act, and refusing to engage simply delegates the decision to whoever will. How society works is knowable, and knowing it is the prerequisite for changing it.
The harms are not mysterious and the responses are not untried. Every tool discussed in the workshop is available to people who organize. Technologists hold real cards; the question is whether they play them.
8.7 Exercises
- Policy-window map.
- Pick one concrete harm. For Kingdon’s three streams: assess whether a window is open and what would open it.
- Coalition design.
- Choose a reform (e.g., mandatory algorithmic audits). List likely supporters, likely opponents, and the unexpected allies (parents, small business, civil-liberty groups, religious communities) that could make it a winning coalition.
- Leverage inventory.
- As a technologist, list your specific sources of leverage (skills, access, capital, community) and one concrete action you could take in each of the next 1, 6, and 12 months.
- Harm-to-lesson mapping.
- Take one harm (e.g., AI-driven hiring discrimination) and write one sentence connecting it to each lesson. The point is to demonstrate that every real harm has structural, economic, psychological, stratificational, regulatory, and ideological dimensions.
- Reform-menu debate.
- Assign groups different positions (e.g., audits only, liability, breakups, public options, worker power, “it cannot be fixed”). Each group argues its case; then the room discusses which combination is realistic and coherent.
- Backward plan.
- Pick a target outcome five years out (e.g., “audits of high-impact models are mandatory and meaningful”). Work backward to name the milestones, coalitions, and political conditions required at each stage.
- Worker-power audit.
- Choose one workplace harm (surveillance, deskilling, on-call exhaustion, pay opacity). Which of the worker-power institutions (codetermination, sectoral bargaining, arbitration, a union) would address it, and who would have to move first?
- Five tools.
- Take a single reform (e.g., algorithmic audits) and map it onto the five tools—collective organizing, regulatory pressure, alternative structures, strategic communication, sustained participation. Which tool is missing from most tech-industry versions of this plan?