5 Inequality
Why are some people consistently better off, and is that “natural”?
- Contrast structural and individual explanations of social outcomes.
- Explain cumulative advantage and opportunity hoarding.
- Describe intersectionality and why single-axis analyses miss the worst harms.
When you explain why someone succeeded or failed, which verbs do you reach for? If you are like most people, you reach for verbs about the person— she worked hard, he didn’t apply himself— rather than verbs about the system she was working inside. This chapter is about how much that reflex hides, and what it costs.
5.1 Three Dimensions of Position
Max Weber argued that social position is not one thing but three, and that the three do not always travel together (Weber 1946). Class is economic position. Status is prestige and honor. Power is the capacity to realize one’s will. A staff engineer at a large company may have high class but limited power; a founder has power; a content moderator has neither. The mismatches matter, because people assume that money, respect, and influence are interchangeable, and they are not.
The differences between positions are not just individual; they are social and, at the level of whole countries, measurable. Richard Wilkinson and Kate Pickett compared rich societies and found that more unequal ones do worse on nearly every social indicator— health, violence, trust, education, even mental illness— regardless of how wealthy the society is overall (Wilkinson and Pickett 2011). Their explanation was status anxiety: in an unequal society, people spend their lives comparing themselves to those above them, and the anxiety that produces is itself destructive. Inequality is not only a distribution of resources; it is a distribution of stress.
5.2 Structure and the Individual
The fundamental attribution error is the tendency to explain other people’s outcomes by their traits and our own by circumstance. Applied to society, it becomes the default explanation for inequality: the successful are talented and hardworking, the unsuccessful are not. The evidence points the other way.
The first mechanism is cumulative advantage, sometimes called the Matthew effect: advantage begets advantage. A wealthy school, a first internship, seed funding, an early follower count— each early advantage compounds, so that small differences at the start become large differences over a lifetime. The second is opportunity hoarding (Tilly 1998): groups use credentials, networks, and gatekeeping to keep advantages for themselves while appearing to reward merit. The third is intersectionality (Crenshaw 1989): overlapping systems of disadvantage interact rather than simply add, so that a person facing racism and sexism experiences something worse than the sum of the two, and a system designed to detect one kind of bias misses the people caught in both.
The discrimination is concrete and, where tested, consistent. In audit studies, researchers send identical resumes to employers, varying only the name at the top, and names read as white consistently receive more callbacks (Rivera 2015). Lauren Rivera found that interviewers at elite firms described candidates as “polished” or “rough around the edges” in ways that tracked class background rather than competence. Amazon’s internal recruiting tool, abandoned in 2018, penalized resumes containing the word “women’s.” And women’s share of computing was higher in the 1980s than it is today, which directly contradicts the “pipeline” explanation that blames the shortage on a lack of interested women.
Gary Becker predicted in 1957 that competition would eliminate discrimination, because a firm that ignored talent would be outcompeted (Becker 1971). The prediction was falsified. Discrimination persists in competitive markets because favored-group workers benefit from it, which is what Lyndon Johnson meant when he said that if you can convince the lowest white man he is better than the best Black man, he won’t notice you are picking his pocket.
Economists distinguish taste-based discrimination, in which employers simply prefer not to hire members of certain groups, from statistical discrimination, in which employers use group membership as a proxy for characteristics they cannot observe. An employer who uses race or gender as a proxy for reliability is drawing inferences from population-level patterns that are themselves often the product of prior discrimination, which creates a vicious circle. The psychological research on implicit bias complicates the picture further (Banaji and Greenwald 2013). Studies using the Implicit Association Test show that most people, including those who explicitly reject prejudice, harbor automatic associations between social groups and evaluative attributes. These associations operate below the conscious level, and lead well-intentioned people to evaluate resumes differently depending on the apparent race or gender of the applicant.
5.3 The Ledger Beneath the Surface
The categories we treat as natural facts about people were mostly built by someone, for a reason. In 1950, South Africa’s apartheid government created the Race Classification Board, whose examiners used the “pencil test”— if a pencil inserted into someone’s hair stayed in place, the person might be classified as “Coloured” rather than “White.” Siblings were assigned to different categories with different legal rights and different permitted neighborhoods. The US “Hispanic” category is younger still: it does not appear in census data before 1970, when the Nixon administration created it to aggregate Spanish-speaking populations for federal programs (Omi and Winant 2015). The sociologists Michael Omi and Howard Winant called this process racial formation. Irish immigrants in the 1840s appeared in popular cartoons as racially distinct from Anglo-Saxons, while Italian immigrants in the early twentieth century were subject to legal discrimination partly justified on racial grounds. Each of these groups eventually became “white” through political processes that included them while excluding Black Americans. Understanding how these categories are built, and by whom, is necessary for understanding how algorithmic systems operate in societies organized around race. A hiring model trained on historical promotion decisions learns to prefer candidates who resemble the people who were previously promoted, in organizations that excluded people by race and class. A recidivism prediction tool trained on arrest data learns patterns from policing decisions that were themselves racially disparate.
These categories leave a measurable trace. Neighborhoods that were redlined in the twentieth century still show lower average credit scores today, not because their residents are less creditworthy, but because they were systematically excluded from the wealth-building mechanisms that credit history reflects. Credit bureaus sit at the center of this arrangement. They collect financial data about hundreds of millions of people without consent, profit whether the data is accurate or not, and in the 2017 Equifax breach exposed the personal information of roughly 147 million people while facing no criminal charges. Accuracy and fairness are not the same thing: a score that accurately summarizes a history of exclusion still perpetuates it. The three major credit bureaus collect financial data about hundreds of millions of people without their active participation or consent. They do not collect data from the people whose files they maintain. They collect it from creditors—banks, credit card issuers, debt collectors— who report account status and payment history. The person whose data is being collected is not notified when a new entry appears, has no opportunity to contest it before it is recorded, and may not know it exists until they apply for credit and are denied. The bureau’s relationship is with the furnisher and the purchaser of the data, not with the subject. Credit data determines whether people can rent housing or get a job. The companies profit whether the data is accurate or not. A person whose file contains an error bears the full cost of that error while the bureau bears none. The Fair Credit Reporting Act gives consumers some rights to dispute errors, but the dispute process is designed and operated by the bureaus themselves, and has repeatedly been found to be inadequate.
Equifax’s response to the breach was to offer credit monitoring services—sold by Equifax. The settlement reached in 2019 provided most affected consumers with a few dollars apiece, after the claims fund was overwhelmed by the number of applicants. No one at Equifax faced criminal charges.
China’s Social Credit System bars individuals with low scores from purchasing plane and train tickets, enrolling their children in private schools, and accessing certain financial services. This is the endpoint of the credit bureau model, where a score derived from past behavior determines present civic participation.
Algorithmic hiring systems, insurance scoring models, and predictive policing tools share two structural features that determine their impact: they collect data without meaningful consent, and the subjects cannot see or challenge the decisions made about them. Credit bureaus are the most mature example of this model, and the clearest preview of where the others are heading.
5.4 Poverty Is a Tax on Cognition
Sendhil Mullainathan and Eldar Shafir found that scarcity depletes cognitive capacity the way sleep deprivation does (Mullainathan and Shafir 2013). Managing an unpredictable income, tracking which bills are overdue, deciding whether the bulk item is worth the storage cost— these consume the cognitive bandwidth that a person then does not have for navigating bureaucracies or resisting temptation. Peggy McIntosh called the flip side an “invisible knapsack”: a set of advantages so routine that those who carry them do not experience them as advantages at all (McIntosh 1989). You do not notice that your name gets callbacks, that your neighborhood has a library, or that the official treats you as a legitimate claimant rather than a probable fraud.
This is why “just try harder” is not a serious explanation. When Britain’s Universal Credit system told claimants to “just go to the library” to apply online, the suggestion ignored that the library required a membership card, that the card required proof of address, and that the application timed out and lost your work if you did not finish it in a single session. The word “just” was doing an enormous amount of lifting. Each “just” solution transfers the cost of a structural failure onto the person least able to bear it (Eubanks 2018). “Just” does that in a lot of conversations. When someone with relative power is told about a problem they have not personally faced, a common response is to suggest an individual solution: “Why don’t you just move to a better neighborhood, just report the harassment to HR, just open a bank account, just apply for a scholarship.” The suggestion is not usually made in bad faith. It is made because the person offering it has done that thing and found the process manageable. What they cannot see is what made it manageable for them and what makes it unmanageable for someone else.
The most obvious thing “just” conceals is cascading prerequisites. Many systems assume that the person using them already has a set of prior resources. Online-only access is built on a cascade of prior requirements, each of which depends on the one before it. A second thing “just” hides is the cost of the transaction. Taking a day off work to stand in line costs money that people without savings cannot spare. Challenging a decision by a government agency or a platform company requires knowing how to challenge it, having the literacy and the time to fill in forms, and being willing to risk the relationship with the institution you depend on. A third thing is cognitive load. The person telling someone to “just” do something is typically not paying this tax. The pattern is particularly vicious on tech platforms. Companies routinely advise users to “just report” harassment, “just use a VPN” to avoid surveillance, or “just switch platforms” when a service degrades. Each suggestion assumes that the user has time, technical literacy, and social capital that many do not have, and none of them fix the underlying system.
5.5 Meritocracy, Examined
The word “meritocracy” was coined by Michael Young in a 1958 satire (Young 1958). His dystopia imagined a society that perfected the measurement of talent and sorted people ruthlessly by it, producing a ruling class convinced of its own deservingness. The word was then adopted as a term of praise, and Young spent the rest of his life pointing out the confusion.
The claim does not survive contact with the data. Organizations that explicitly adopt merit-based pay principles show larger gender pay gaps than those that do not (Castilla and Benard 2010). The mechanism appears to be moral licensing: people who believe the system is already fair feel less need to monitor their own judgments for bias. Believing in meritocracy also predicts harsher judgment of people who fail— if outcomes are deserved, then failure is evidence of personal deficiency— which conveniently shifts responsibility from structures to individuals.
How elite selection actually works is different from how institutions describe it. Studies of admissions to elite universities, hiring at professional services firms, and promotion within corporations consistently find that formal credentials matter less than social legibility: the ability to display the cultural capital—tastes, manners, cultural references— that signals membership in the right networks. Lauren Rivera’s research on hiring at elite firms found that interviewers described candidates as “polished” or “rough around the edges” in ways that correlated with class background rather than competence. The stated meritocratic criteria were real, but they operated within a prior filter that most candidates never saw.
The meritocratic ideology takes specific forms in tech. The puzzle interview, popularized by Microsoft in the 1990s, selects for people who have prepared for puzzle interviews, which correlates with having the time, networks, and educational background that make such preparation possible. “Culture fit” as a hiring criterion is even less bounded: it is assessed subjectively and consistently reproduces the demographic composition of existing teams. Neither criterion is without signal about likely job performance, but neither is calibrated against actual job performance data, because most organizations never close that loop.
India’s IIT and IIM entrance examinations were designed as objective filters, but a coaching industry whose best programs charge fees equivalent to several years of average household income ensures that upper-caste and wealthy families dominate admission lists despite formally caste-neutral scoring.
The most instructive case is Andrew Carnegie. He accumulated his fortune partly by suppressing wages and, in 1892, hiring Pinkerton detectives to break a strike at his Homestead plant that left ten workers dead. He spent his final decades funding libraries, universities, and concert halls, on his terms: cities had to provide the land and commit to maintaining the buildings in perpetuity. The more consequential precedent was the Flexner Report of 1910, commissioned by the Carnegie Foundation; it is discussed in the next section. Both are opportunity hoarding wearing a merit badge. Mobility between generations is far lower and stickier than the “American Dream” story implies, so “meritocracy” is best treated not as an assumption but as an empirical claim to test.
5.6 Professional Exclusion
The history of how discrimination gets built into professional structures is worth understanding before applying it to computing. The Flexner Report of 1910 recommended consolidating medical education around a small number of university-based schools modeled on Johns Hopkins. What is less often noted is what this consolidation closed: the eclectic and homeopathic schools that had trained a significant proportion of women physicians, and the historically Black medical colleges that were the primary pathway into medicine for Black practitioners. The reorganization of medicine around a single credentialing model was also a reorganization of medicine around a single demographic profile. The American Medical Association’s subsequent decades of opposition to women’s participation in medical societies and to Black physicians’ hospital admitting privileges extended and consolidated what the Flexner Report had begun.
The legal profession followed the same pattern. Bar associations across Britain, the United States, and continental Europe formally prohibited women from practice until courts or legislatures intervened. In the UK, women were not admitted to the legal profession until 1919. In the US, the Supreme Court upheld Illinois’s exclusion of women from the bar in 1872, with a concurrence that stated women’s domestic roles were divinely ordained. The formal barriers were eventually removed. The informal barriers persisted: partnerships that did not hire women, chambers that did not take women as pupils, and professional associations that did not see women as natural members. These required decades more of pressure to dislodge, and have not been fully dislodged.
The undercounting of women’s scientific contributions is documented across many countries and disciplines. The pattern includes attribution of women’s work to male collaborators or supervisors, systematic exclusion from authorship norms that credited the laboratory head rather than the person who performed the experiments, and deliberate suppression documented in specific cases such as Rosalind Franklin’s role in the determination of DNA’s structure. Studies of citation patterns, grant success rates, and peer review outcomes continue to find systematic disadvantages for women researchers in many fields, controlling for measurable differences in output.
Mary Beard’s analysis of women’s exclusion from public speech draws the pattern to its full historical length (Beard 2017). From the moment in the Odyssey where Telemachus tells Penelope to go back inside, through Roman rhetoric’s treatment of women’s public speech as inherently disgraceful, through centuries of legal and customary prohibition on women speaking in court and in public assemblies, the mechanisms of exclusion are remarkably consistent. They include treating women’s speech as categorically out of place in authoritative contexts, dismissing or ridiculing women who do speak in those contexts rather than engaging with what they say, and redefining the terms of authority in response to women’s gains so that what women achieve is always reclassified as less important than what they have not yet achieved.
In contemporary computing, the primary mechanism operates under the label of culture fit. Hiring decisions made on that basis are rarely experienced as discriminatory by those who make them. The judgment is framed as whether a candidate seems like one of us, whether they would enjoy working with the existing team, whether they share the team’s values and habits. But when the existing team is demographically homogeneous, hiring for cultural fit reproduces that homogeneity. The criterion functions as a demographic filter while providing its users with a clean conscience.
The automation of hiring decisions does not solve this problem. It replicates it at scale. Algorithmic hiring tools trained on historical hiring data learn to identify candidates who resemble those previously hired and promoted. If the historical data reflects discriminatory patterns, the algorithm encodes those patterns into a system that is harder to scrutinize and challenge than a human decision-maker. Amazon’s internal recruiting tool, abandoned in 2018, penalized resumes that included the word “women’s” and downgraded graduates of all-women’s colleges, because it had learned from a decade of prior hiring decisions that were heavily male.
The standard response to demographic imbalance in tech is to point to the pipeline: not enough women or underrepresented minorities in computer science programs, and the problem will correct itself when universities produce more qualified candidates. This explanation is contradicted by the evidence. Women’s participation in computing was substantially higher in the 1980s than it is today. The decline occurred while women’s representation in law, medicine, and other professions was increasing. The pipeline analogy locates the problem in the wrong place. Demographic problems that originate in hiring practices, retention conditions, and professional culture cannot be solved by changing who enters the pipeline. They require changing the conditions that determine who stays in, who is promoted, and who leaves.
5.7 Who Gets to Decide
The Carnegie story has a contemporary counterpart. In the 2000s, the Gates Foundation, working alongside the Walton Family Foundation and the Eli and Edythe Broad Foundation, channeled hundreds of millions of dollars into reshaping American K-12 education. Their agenda included charter schools, standardized testing, merit pay for teachers, and the Common Core State Standards— reforms chosen by funders who were not elected and could not be removed by the parents, students, or teachers affected. In 2017, Gates Foundation CEO Sue Desmond-Hellmann published an open letter acknowledging that the strategy had not produced the results intended. She described it as a learning experience. Most of the teachers whose work and lives had been thrown into turmoil used stronger language. The same structure extends to global health: for many years the Gates Foundation was the second-largest funder of the World Health Organization, behind only the United States government. This is philanthrocapitalism: the conversion of taxable wealth into permanently controlled endowments that shape public policy without democratic accountability. Donor-advised funds extend it further: the donor takes an immediate tax deduction while retaining the ability to direct distributions for decades, and such funds held over $230 billion in assets by 2022 (Giridharadas 2018). Whatever the intentions, that is influence exercised without the legitimacy that comes from being elected or removable.
5.8 The Gendered Economy
On October 24, 1975, ninety percent of Icelandic women refused to work— at paid jobs, in the home, or anywhere else. Schools closed, factories shut down, flights were cancelled, and the economy of Iceland effectively stopped. The women had made visible something the national accounts refuse to see: the work that has no price.
Marilyn Waring spent years documenting the decisions embedded in GDP (Waring 1988). A woman who cooks dinner for her family contributes nothing to GDP, but if she hires a cook and goes out to work herself, GDP rises twice— once for her wages and once for the cook’s. The system counts the second arrangement as more productive, a choice made by people who had mostly never performed the work in question. In Japan, the term sengyo shufu, meaning “full-time housewife”, became widespread only in the postwar period, when rapid industrialization in the 1950s and 1960s created male industrial jobs and pushed women into domestic roles. The image of the sengyo shufu as a traditional Japanese institution is largely the invention of corporations and government policy from a period of perhaps thirty years. Before industrialization, Japanese women had worked extensively in agriculture, small business, and textile production. The tradition was assembled recently under identifiable political and economic pressure.
The sociologist Arlie Hochschild found that women in dual-income couples were coming home and working a second shift of domestic labor (Hochschild and Machung 1989). Over the course of a year, the women she studied worked roughly a full additional month compared to their partners, and then explained the inequality away with “family myths”— stories about why the division made sense in their particular case. The rationalization is the key point: exploitation is most durable when those experiencing it explain it to themselves as freedom. The historian Silvia Federici argued that the witch trials of the sixteenth and seventeenth centuries, during which tens of thousands of women were murdered, were primarily a way to discipline female labor (Federici 2004). The women most often accused were those who lived outside household structures: widows, healers, and midwives. The trials coincided with the enclosure of common land and the creation of a landless wage-labor force. Federici showed that enclosing women’s reproductive and domestic labor inside the household, unpaid and culturally devalued, was an economic project as much as a cultural one; it produced a class of workers whose labor was essential and whose compensation was called love.
The platform economy has reproduced the structure with new interfaces. Apps coordinating cleaning, childcare, and elder care present themselves as markets connecting independent contractors with clients, but what they actually do is take informal, feminized labor that was already undervalued and formalize it while keeping the wages and protections of the informal economy. The worker bears the risk, the platform takes the margin, and the devaluation is the same one the national accounts have performed for decades. The global economy depends on nursing, childcare, eldercare, and domestic service, which are systematically undervalued because they are feminized. The workers are underpaid not because care work is unskilled but because the labor market for care work is structured differently from the labor market for work that men predominantly perform. In India, the national health system relies heavily on informal women’s health workers, the ASHAs, who are classified as volunteers rather than employees and receive incentive payments rather than wages. This is not an oversight; it is a legal and administrative decision with direct economic consequences for nearly a million women. The ASHAs have organized and struck repeatedly for employee status and wage protections. The government has consistently refused.
Invisible labor is endemic to tech companies. Research has shown again and again that women perform more of the relational maintenance that holds teams together, such as mentoring junior colleagues and managing interpersonal conflict, without receiving credit for it in performance reviews or promotion decisions. A study of a large software company found that women spent significantly more time on what the researchers called non-promotable tasks that benefited the organization but were invisible to the criteria used to evaluate careers. When it was pointed out that women were doing more of this work, management encouraged everyone to do less of it, which predictably meant that women continued to do it and men continued not to.
Five years after the 1975 strike, Iceland elected its first female president, who went on to serve four terms. The organizing that preceded the strike built the political infrastructure that made the election possible. Fifty years on, the gap between men’s and women’s wages in Iceland has not closed, and women still perform more unpaid domestic labor (Folbre 2001).
Inequality is not a sorting of individuals by merit; it is the accumulated effect of rules, networks, and compounding advantage. Meritocracy is a claim to test, and often a story that legitimizes inheritance.
5.9 Exercises
- Explain the gap.
- Take a real inequality (who gets into a selective program, who receives venture funding). Produce two explanations (one individual, one structural) then discuss which better fits the evidence and why the individual one is the cultural default.
- Three-dimension map.
- For four tech roles (e.g., startup founder, staff engineer, content moderator, gig worker), rate class/status/power as high or low. Discuss the mismatches (e.g., high status but low power; high income but low security).
- Cumulative advantage.
- Trace how one early advantage like a wealthy school, a first internship, seed funding, or an early follower count compounds across a career. Identify the points where policy or design could interrupt the compounding.
- Opportunity-hoarding audit.
- Find a credential, interview process, or referral network that functions to hoard opportunity. Distinguish its stated purpose from its actual effect.
- Intersectional case.
- Analyze how a policy or product harms people differently across overlapping identities (a facial-recognition system, a hiring algorithm, an automated welfare screen). Why does a single-axis analysis miss the worst harms?
- Merit audit.
- Take a supposedly merit-based process (a hiring funnel, an awards program, a promotion ladder). Check whether it produces the outcomes meritocracy predicts, and identify where credentials and networks do the actual sorting.
- Bandwidth check.
- List five decisions you make easily when rested that become hard when you are tired or under financial pressure. What would a product look like if it assumed its users were often in that state?