The Complacent Class

Americans used to be the most mobile people on earth. Today they are less mobile than Europeans. Interstate migration has fallen roughly in half since the 1980s. The country whose mythology is built on movement has stopped moving.

Tyler Cowen has been tracking this for a decade. The picture he draws is uncomfortable across ideological lines, because the people who stopped moving are the same people who have written most of the books about American freedom.

This article is about what happened when the winners of the meritocratic order quit competing and started defending. Article 5 traced the moral side of that order — what the rhetoric of rising has done to the moral self-portrait of the credentialed class. This article traces the structural side: what the credentialed class has built around itself to keep what it was told it earned. The diagnoses are independent. They are also mutually reinforcing. Sandel describes the hubris. Cowen describes the architecture the hubris has built to protect itself. The architecture is not new. It has been visible for forty years to anyone willing to read the migration data, the startup data, the residential segregation data, and the assortative mating data side by side. The reason it has not been described as such is that the people most able to describe it are the people the description fits.

I. The Numbers

Cowen’s 2017 book The Complacent Class opens with a simple empirical claim: since roughly 1980, the United States has lost its historical restlessness. Six different measures of dynamism — geographic mobility, job-to-job mobility, startup formation, residential income segregation, cross-class marriage, and the rate at which Americans actually do anything new — all bend the same way at the same time. They all bend toward less.

This is an unusual kind of claim, because the loss is not a crisis. There is no headline event. There is no scandal. There is just a slow, multi-decade decline in the willingness of the country’s population to do the kinds of things the country’s mythology says it does. Americans, the mythology says, are the people who pull up stakes and move west, who quit the steady job to start the company, who marry the person from the wrong side of the tracks, who take the risk because America is the place where taking risks pays off. The mythology is roughly true for the period from the Civil War through the 1970s. After 1980 it is roughly false, and it is false in ways the mythology has not absorbed.

Take these one at a time. Each is unremarkable on its own. Together they describe a shift the political language has not caught up with.

Geographic mobility. Through the 1970s, roughly three to three and a half percent of Americans moved across state lines in any given year. By the 2010s, the rate had fallen below 1.5 percent. The decline is documented in Census Bureau Current Population Survey data and has been tracked carefully by economists at the Federal Reserve and the Brookings Institution. The drop is real, it is large, and it is not explained by the aging of the population. Even controlling for age, Americans move less than they used to. They move less than Canadians, they move less than Australians, and in many recent years they have moved less than the citizens of several European countries whose populations the American mythology used to describe as “rooted” in contrast to American restlessness. The contrast has reversed without anyone noticing.

Startup formation. The share of American employment in firms less than five years old has fallen from roughly eighteen percent in the late 1970s to roughly ten percent in the 2010s. The annual rate at which new firms are founded has fallen from about thirteen percent of all firms in 1978 to about eight percent in the 2010s. The economists Ryan Decker and John Haltiwanger have tracked this carefully and have ruled out most of the obvious explanations — it is not just about the dot-com bust, it is not just about Walmart, it is not just about the Great Recession. The decline is broad, persistent, and predates each of the events that have been blamed for it. Something happened to the rate at which Americans start new things, and it happened across the entire economy beginning around 1980.

Residential income segregation. The Pew Research Center documented in 2012 that residential segregation by income has risen sharply since 1970. The share of American children growing up in mixed-income neighborhoods fell from roughly two-thirds in 1970 to roughly forty percent in the 2010s. Rich neighborhoods got richer. Poor neighborhoods got poorer. The middle, which used to absorb most American children, hollowed out. Sean Reardon at Stanford has run the same analysis on school districts and found the same pattern: the share of children attending schools where most of their classmates are from a different income background has fallen sharply, and the fall is larger at the top of the income distribution than at the bottom. The rich are sorting themselves more aggressively than the poor, because the rich have more options.

Assortative mating. Educated people increasingly marry other educated people. In 1960, about half of college-educated Americans married someone with a college degree. By the 2010s, the share had risen to over seventy percent and was still climbing. The change is mostly driven by women: as women’s educational attainment caught up with men’s, the pool of college-educated potential partners for college-educated women grew, and the marriage market sorted accordingly. Christine Schwartz at Wisconsin and Robert Mare at UCLA have done much of the underlying work. The consequence: cognitive and economic advantage now compounds inside households across generations in a way it did not in 1960. Two professionals marry each other, raise children in a household with two professional incomes and two professional vocabularies, and produce children who attend the schools their parents’ income permits and inherit the social networks their parents’ positions have built. The same compounding works in the other direction. The household with two non-college-educated incomes does not have access to the same compounding, and the gap between the two trajectories now widens generation by generation in a way the mythology of mobility has not absorbed.

Job-to-job mobility. The rate at which American workers switch jobs has fallen by roughly twenty-five percent since 2000. This is not just about the Great Recession, although the Great Recession did not help. The decline is structural and predates the recession. Workers who would have switched jobs in 1985 stay put in 2025. The reasons are debated — non-compete clauses, the rise of two-income households whose moves require coordination, employer-tied health insurance, the difficulty of cross-state professional licensing, the way matching technology has made the job market more efficient and therefore stickier — but the underlying fact is not in dispute. American workers have become less likely to leave the employer they have for a different one.

Each of these measures, on its own, is the kind of thing that gets a paper in a regional economics journal. Together they are the picture of a country that has stopped doing the things its mythology says it does. The complacent class is not a moral category. It is the empirical claim that the American population has, since 1980, gradually become less willing to move, to switch, to start, to mix, and to risk. The class that has done this most aggressively is not the bottom of the income distribution — the bottom never had the mobility the mythology claimed for it — but the top thirty to forty percent of earners. The professional, college-educated, urban-and-suburban Americans who run the country’s institutions and write the country’s books are also the Americans who have most clearly stopped moving.

The complacent class is the empirical observation that the people who won the meritocratic game have, by every measurable behavior of dynamism, stopped competing and started defending. The defense is rational at the household level. It is destructive at the system level. The system is held together by the dynamism of the people the defense excludes.

II. What Is Being Defended

To say the class is complacent is to invite the question: complacent about what? What is being held in place? The data above describes the symptoms. It does not describe the thing being protected. The thing being protected is the position itself — the set of advantages the credentialed class has accumulated over a generation and is unwilling to risk losing in the next round of the sorting.

The position has multiple components. The first is a school district. The second is a credential. The third is a network. The fourth is a marriage. The fifth is a neighborhood that other people’s children cannot easily enter. The sixth is a portfolio of small institutional advantages that compound across the life cycle — the doctor who returns the call, the lawyer who answers the email, the contractor who shows up on time because the household belongs to a network the contractor wants to maintain access to. None of these are illegal. Most of them are not even improper. They are the ordinary outputs of an ordinary upper-middle-class life in a country that allows market sorting to reach into more and more domains. Each of them, on its own, is a benefit any household would want. Together they constitute the position the complacent class is now organized around defending.

The defense takes specific forms. The most visible is residential. Single-family residential zoning, exclusionary minimum lot sizes, restrictive parking requirements, height limits, setbacks, design review boards, historic preservation overlays — these are the architecture of residential class sorting in modern America. None of them mention income. None of them mention race. All of them function as filters because they raise the marginal cost of entry into a neighborhood above the marginal income of households below a particular threshold. The threshold is calibrated, in most expensive American cities, to roughly the median household income of the existing residents. Below the threshold, you cannot buy in. Above the threshold, you can. The school district that the neighborhood feeds is, by extension, a school district whose students are sorted by household income before any educational decision has been made.

This is the link between Articles 2 and 6 of this series. Article 2 traced the federal-institutional decisions, from 1872 through 1974, that built the property-tax-and-district-line architecture of American school finance. Article 6 is about how the people who benefit from that architecture today defend it. The defense is local, decentralized, and conducted in the language of property rights, neighborhood character, school quality, and the “preservation of community.” None of these phrases mention class. All of them function as class-sorting mechanisms because the things being preserved are the things that property values and exclusionary zoning have produced. The defense is so well integrated into ordinary upper-middle-class civic participation that the people doing the defending experience themselves as good citizens engaged in legitimate local politics. They are not lying to themselves. They are also not seeing the system the politics serves.

The second form is educational. Article 4 traced the elite reproduction engine from Harvard’s 1636 founding through the 1922 invention of holistic admissions to the Arcidiacono numbers from the SFFA case: at Harvard, legacy applicants admitted at 33.6 percent versus 5.9 percent for non-legacies; recruited athletes at 86 percent; Dean’s Interest List at 42 percent; ALDC applicants making up 5 percent of the applicant pool but 30 percent of the admitted class. Those numbers describe the gates. This article is about what happens behind the gates — the K-12 architecture of the complacent class’s feeder system, the prep schools, the college consultants, the summer enrichment programs, the test prep capital, the extracurricular optimization that begins in middle school and accelerates through high school and ends in the application portfolio that the gates were built to recognize.

The third form is reproductive. Assortative mating compounds the position into the next generation. The household that pairs two professional incomes also pairs two professional vocabularies, two professional networks, two sets of cultural capital, and two sets of expectations about what the children will become. The children inherit all of it. They inherit it without being told, in most cases, that they are inheriting anything — because the meritocratic frame from Article 5 has trained the parents to describe their own position as earned, and the description gets passed on along with the position. The children grow up believing the household’s advantages are the result of their parents’ effort and that the same effort will produce the same advantages for them. The belief is partly true and largely false, in proportions the meritocratic frame is not equipped to disentangle.

The fourth form is the matching technology that has emerged in the last twenty years to make the sorting more efficient. This is Cowen’s most original contribution and it deserves its own section.

III. Matching Technology

The phrase is Cowen’s. He uses it to describe a family of tools and platforms that, beginning in the early 2000s, have made it dramatically easier for individuals and households to find the niche that matches their existing preferences. The tools are familiar. The Common App lets students apply to twenty colleges from a single submission, which means colleges now receive far more applications per available seat and can sort the applicants more aggressively along whatever dimensions they choose. Zillow lets families view real-estate listings filtered by school district, which means families can sort themselves into the exact district that matches their income bracket without ever visiting a neighborhood they would not want to live in. OkCupid, Tinder, Bumble, Hinge, and their successors let educated singles filter potential partners by educational attainment, profession, religion, height, political affiliation, drinking habits, and a dozen other criteria, with the result that the dating pool effectively segregates itself before any face-to-face contact occurs. Spotify and the streaming services use algorithmic recommendation to deliver music and content closely matched to the user’s prior listening, which means listeners are increasingly never exposed to anything jarring, unexpected, or outside their existing taste profile.

Each of these tools is, individually, a delivery of choice to the user. Each is marketed as an expansion of options. Each is, in practice, a delivery of efficient sorting and, by mathematical consequence, a contraction of the kinds of contact between unlike people that produced cross-class learning, friendship, and movement in the prior era.

The paradox Cowen identifies is the central counterintuitive finding of his book: matching technology, by making sorting easier, has made mobility harder. When everyone can sort perfectly into the niche that suits them, nobody needs to cross a boundary. The friction that used to force cross-class contact — living in a mixed neighborhood because that was the only available housing, marrying locally because long-distance matching was impossible, hearing music you didn’t pick because the radio station played whatever it played, attending the high school whose catchment area you happened to be in — has been engineered away. The result is a society that is simultaneously more diverse in aggregate and more segregated in practice. There are more total kinds of people. They are more efficiently sorted into the niches where they encounter only people like themselves.

This is not a critique of any one of the tools individually. Zillow is a useful site. The Common App is a useful service. Dating apps have measurable benefits for the people who use them. The point is what happens when all of the matching technologies operate together at scale: the cumulative effect is the engineering of a society in which the upper-middle-class household never has to do the cross-class work the prior society’s ordinary frictions made unavoidable. The household’s exposure to people unlike itself becomes voluntary. Voluntary exposure, at scale, is no exposure. The household’s children grow up never having gone to school with the children of the people who clean the buildings the household’s parents work in. They will encounter those children, if at all, in college, where the encounter is structured by the architecture of the institution, monitored by deans, and confined to a four-year bubble after which the sorting resumes.

One of the things matching technology has done is destroy the kind of ordinary cross-class friendship that used to result from accidental proximity. Anthony Jack’s ethnography of low-income students at elite colleges, which Article 4 of this series cited, contains a quietly devastating finding: the affluent students at the elite college do not become friends with the low-income students. They are physically close. They eat in the same dining halls. They live on the same floor. The friendships do not form, because the cultural distance between the two groups is larger than the physical distance, and matching technology in the prior eighteen years of each student’s life has made sure that the affluent student arrives without the practiced ability to bridge the distance. The bridges are no longer built, because the eighteen-year curriculum of bridge-building — the mixed neighborhood, the mixed school, the mixed extracurricular — has been replaced by the optimized sorting that delivered each child to exactly the environment that matched their parents’ income.

This is also where Habermas’s framework returns. The lifeworld of the complacent-class household has been colonized by the optimization logic of the matching technologies. Family deliberation about what childhood is for has been replaced by extracurricular planning. Family deliberation about what college is for has been replaced by admissions strategy. Family deliberation about what kind of community to live in has been replaced by school-district shopping on Zillow. The colonization is so complete that the original questions can no longer be asked inside the household, because the household’s decision-making vocabulary has been replaced with the decision-making vocabulary of the platforms. A parent who wanted to ask “what is the right school for my child” in any sense richer than “which school’s test scores are highest within commuting distance of the homes we can afford” has to actively resist the platform’s framing, because the platform answers the second question by default and the first question is not even a query the platform recognizes. The parent who wants to ask the first question is asking the platform to do something the platform was not built for, and most parents, exhausted, accept the framing the platform delivers.

This is the colonization Habermas predicted in the early 1980s when he could not have imagined Zillow. He wrote about money and power displacing the communicative processes of family and community. The matching technologies are the contemporary form of the displacement. They are not money exactly, and they are not power exactly. They are interface, design, and the structuring of the choice architecture, and they perform the displacement function more effectively than any prior tool because they appear to be giving the household what the household asked for. The household asked for choice. The platform delivered efficient sorting, which is what choice looks like at scale when everyone uses the same platform. The choice and the sorting are the same operation seen from two angles, and the angle the household sees is the choice.

IV. The Cleanest Joke in the Data

Here is the only joke this article will tell. It is in the data. The author is not making it up.

The American college towns are, by income, among the most segregated places in the country. Bloomington, Indiana. Ann Arbor, Michigan. Chapel Hill, North Carolina. Berkeley, California. Cambridge, Massachusetts. Ithaca, New York. Madison, Wisconsin. Each of these is the political identity capital of progressive professional America. Each of these is, on the actual residential income data, more segregated by class than the average American metropolitan area.

Sean Reardon’s residential segregation work, which has been tracking American income segregation at the metropolitan level for two decades, is the place to find the numbers. The pattern is consistent. The metro areas with the highest college-educated populations and the most progressive voting records are also the metro areas where the rich and the poor live in the most thoroughly separated places. The university campuses sit in the middle of cities whose neighborhoods have been sorted with such precision that a low-income household and a professional household will live a fifteen-minute drive apart and never share a street, a school, a grocery store, a bus line, a dentist, a public park, or a coffee shop. The political identity of the residents says one thing. The zoning maps say another.

This is a finding the residents of the college towns find difficult to absorb. The residents have spent decades thinking of themselves as the good ones — the ones who voted for the school referendum, who put up the yard sign, who attended the racial-justice rally, who serve on the diversity committee at work. All of this is true. The yard signs are real. The rallies happened. The diversity committees met. The fact that they are reading this paragraph from inside a single-family home in a residential zone whose minimum lot size and design review process have functioned for forty years to keep their neighborhood at exactly the household-income profile their progressive politics rhetorically opposes is the part that the political identity has not registered. The two facts coexist in the same biography because the political identity is being maintained by a different mechanism than the residential pattern, and the two mechanisms have never been put in the same room.

The college town is the cleanest case because the political identity and the structural reality are most directly opposed there. The same pattern exists in San Francisco, in Seattle, in Portland, in Brooklyn, in Washington D.C., and in every other professional-class metropolitan area whose voting record is solidly Democratic and whose zoning record is solidly exclusionary. The pattern is not an aberration. It is the architecture of the complacent class wherever the complacent class has settled. The college towns are the most legible version because their population is the most homogeneous in its political self-understanding and the most embarrassing in its structural self-portrait.

It would be easy to make this observation cruel. The article will not. The residents of the college towns are not bad people. The structural pattern is not their personal achievement. They inherited a city whose zoning was set decades ago, they bought into it because it was where the jobs and the schools were, and they have spent their adult lives doing the local civic work the city asks of them. None of that is hypocrisy in the personal sense. The hypocrisy, to the extent there is one, is structural — the political identity that runs in the household’s self-description has not yet collided with the residential architecture that runs in the household’s actual life. When the collision happens, it is uncomfortable, because the household has to choose which fact to revise: the political identity or the residential pattern. Most households revise neither. They hold both, in different rooms of the same brain, and the contradiction does not produce action because the rooms have separate doors.

This is the joke in the data, and the joke is also the diagnosis. The joke lands because the contradiction is real and visible to anyone willing to look at the residential income segregation index next to the precinct-level voting returns. It lands because the people the joke is on are the people the article is being written for, which means the laughter has to happen in the first person plural or it is not honest. The author lives in a similar pattern. So, statistically, do the readers. The joke is not at someone else’s expense. It is at ours.

V. The Sorting Begins Earlier

The numbers from Article 4 are the most-quoted line in this series, and they bear repeating here because the mechanism of the complacent class is the mechanism by which those numbers reproduce themselves across generations.

0.8 percent of American college graduates attend Ivy-Plus institutions. Ivy-Plus graduates hold 41.7 percent of U.S. presidencies and 71.4 percent of Supreme Court seats. Children with parents in the top one percent are 77 times more likely to attend an Ivy-Plus college than children with parents in the bottom twenty percent. These numbers are the gateway. The complacent class is the apparatus that produces the children who walk through the gateway.

The apparatus begins early. In New York City, the elite K-12 ecosystem is the most legible example because the institutions are concentrated, the prices are public, and the application sequences are documented. Dalton, Trinity, Horace Mann, Collegiate, Brearley, Spence, Chapin, Riverdale Country — the K-12 schools whose alumni populate the Ivy League at rates that exceed any public high school in the country — charge tuition north of sixty thousand dollars per child per year. The price is increasing faster than inflation and has been for two decades. There are waitlists. The waitlists begin in pre-kindergarten, which means parents apply to nursery schools that feed into the K-12 schools that feed into the Ivy League. The application sequence starts when the child is two years old. By the time the child is four, the household has made decisions whose cumulative cost over the child’s K-12 education will exceed the median lifetime earnings of an American household, before the child has applied to a single college.

This is not nostalgia for the public schools. The same pattern is back-propagating into the elite public-school system through magnet programs, gifted-and-talented tracks, and the elite charter movement, which has emerged in the last fifteen years as the public-school version of the prep school feeder. Stuyvesant, Bronx Science, and Brooklyn Technical High School in New York; Walter Payton in Chicago; Thomas Jefferson High School for Science and Technology in northern Virginia — each of these is a public institution that has, by the design of its admissions criteria, produced an entering class whose demographics resemble the prep schools more than the public schools they sit beside. The mechanism is the test, the application essay, the interview, the supplementary materials, the parental sophistication required to navigate the application process, and the test prep capital required to clear the test bar. None of these is officially exclusionary. All of them are class filters that operate at a stage early enough that the affluent households can convert their resources into outcomes for their children before the public-school system can absorb the children of the non-affluent.

The elite K-12 schools, public and private, do not pretend that what they sell is instruction alone. The marketing materials describe community, network, character, leadership, opportunity, and access. The descriptions are honest. The community is real. The network is real. The character development is real. None of these is the same as instruction. All of them are the position the complacent class is paying for, and the price is what it is because the position has scarcity and the scarcity is what the price is buying.

This is where the article’s most important question becomes audible, and it is the question Russ Roberts has been asking on EconTalk for twenty years in dozens of different contexts. The question is: what is the buyer actually buying?

When a Manhattan household pays sixty-five thousand dollars per year for Dalton, what is the household paying for? The instruction is excellent. The article will not pretend otherwise. Dalton’s teachers are accomplished, the curriculum is rigorous, the small classes allow attention the public schools cannot match. The instruction is part of what is being purchased. But the instruction is not what makes Dalton cost sixty-five thousand dollars per year, because instruction comparable in quality is available at many of the better-funded public school districts in the same metropolitan area, and at a fraction of the price. The differential is paying for something else. It is paying for the network the school’s alumni population produces. It is paying for the access to the Ivy League admissions pipeline that the school’s relationships with elite university admissions offices have built. It is paying for the cultural capital the children acquire by spending twelve years inside an environment whose ambient assumptions are the assumptions of the credentialed class. It is paying for the sorted peer group that the parents have screened for them. It is paying for the position.

The Russ Roberts move is to ask this question slowly and let the answer accumulate without rhetorical heat. The answer is not bad faith. The parents who write the check are not lying about why they write it. They genuinely value the instruction. They also genuinely value the network and the access and the position, and they are not always able to disentangle which part of the value drives the willingness to pay. The honest disaggregation would say: a quarter of the price is for the instruction, three-quarters of the price is for the position. The price is the price because the position is what the position is, and the position has scarcity because the institution is selective, and the institution is selective because keeping the position scarce is what gives the position its value.

This is the carpenter/house distinction from Article 7 starting to take shape, and the article is going to leave the full development of the distinction to the synthesis piece. For now, what matters is that the K-12 architecture of the complacent class is doing the same kind of work the elite undergraduate institutions are doing, and the work is structurally inseparable from the work of the matching technologies and the residential zoning and the assortative mating. They are the same machine. The machine is the architecture of the complacent class’s reproduction across generations, and the position it is reproducing is the position the meritocratic frame from Article 5 has been training the parents to describe as something they earned.

VI. Why the Sorting Is Sticky

Nicholas Christakis’s network science is the best framework for explaining why the position, once held, is so hard to dislodge across generations. His work, building on the studies of social network propagation that began with the Framingham Heart Study and have since been replicated in dozens of contexts, shows that behaviors, opportunities, information, and trust spread along the edges of a social network and stop at the points where the edges stop. A child inside a thick, persistent, high-trust network of credentialed adults absorbs information about how to apply to college, how to write a professional email, how to ask for an introduction, what to call adults, what to wear to an interview, how to follow up after a meeting, when to push back and when to defer, how to negotiate a salary, what kinds of jobs to want and what kinds to dismiss, and a thousand other small acts of professional conduct that the credentialed class takes for granted and that the non-credentialed class learns, if at all, in the first three painful years of attempting to enter the professional world without the apprenticeship.

The information is not secret. It is not hidden behind a paywall. It is also not propagated to people who are not on the network, because the propagation mechanism is conversation, observation, and modeling, and the conversation, observation, and modeling happen inside households, schools, neighborhoods, and friend groups whose composition the matching technologies and the residential sorting have already filtered. A child outside the network does not know what the child inside the network knows, because the information transmission happens in contexts the outside child has no access to.

This is the “hidden curriculum” Anthony Jack documented in his ethnography of low-income students at elite colleges. The students who arrived from prep schools knew how to talk to professors, how to ask for letters of recommendation, how to leverage office hours, how to navigate the alumni network, how to convert a summer internship into a full-time offer. The students who arrived from under-resourced public schools, even those whose academic preparation was equivalent, did not know any of this. They had to learn it on the job, in real time, while also keeping up with coursework, and they paid an enormous penalty for the asymmetry. Jack’s book is called The Privileged Poor because the students who had attended elite boarding schools on scholarship before arriving at elite colleges performed dramatically better than the students who had arrived directly from public high schools, even though both groups were equally low-income. The difference was the years of exposure to the hidden curriculum. The exposure is the network. The network is the product.

This is also why the Stony Brook finding from Article 4 is the cleanest indictment of the elite system. Stony Brook’s bottom-quintile-to-top-quintile mobility rate is 8.4 percent. The Ivy-Plus average is 2.2 percent. Stony Brook produces almost four times more upward mobility than Harvard, not because Stony Brook’s instruction is better, but because Stony Brook actually admits low-income students in numbers Harvard does not. Once admitted, the low-income students at Stony Brook do almost as well as the low-income students at Harvard, because the marginal effect of the elite credential on a student who has already met the academic bar is smaller than the marginal effect of being admitted at all. Harvard’s sorting hurts the country because the sorting is what produces the credential’s scarcity, and the scarcity is what makes the credential valuable, and the value is what reproduces the complacent class. Harvard could double its enrollment tomorrow without sacrificing instructional quality. It does not, because doubling the enrollment would dilute the credential’s signal value, and the signal value is what the institution is structured to preserve.

None of this is hidden. The Chetty data is in Science. The Arcidiacono numbers are in the SFFA case file. The Anthony Jack ethnography is in print. The Stony Brook comparison is in the Opportunity Insights mobility report cards. The information has been public for a decade. The complacent class has not changed its behavior in response to the information, because the information requires a kind of self-recognition the meritocratic frame does not permit. The frame has trained the class to read the data as a description of someone else — the bad elite, the legacy admits, the bribe takers, the Varsity Blues parents — and to exempt itself. The exemption is the trap. The class, as a whole, behaves the way the data describes it. The exemption is what keeps the description from converting into action.

VII. The Stakes Are Higher Now

Cowen’s earlier book, Average Is Over, makes an argument that intensifies the stakes of everything described above. The argument: the American labor market is bifurcating. A shrinking middle of routine cognitive and routine manual jobs is being hollowed out by automation and globalization. What remains is a growing high-skill sector — people who work productively with intelligent machines — and a growing low-skill service sector — people who perform tasks machines cannot. The middle, which used to be the landing place for the average student, no longer exists in the proportions it used to. Being average is no longer enough.

The educational implications are direct. The meritocratic sorting machine described in Articles 4 and 5 maps directly onto this bifurcation. The winners of the sorting are funneled into the high-skill sector, where they will work alongside the intelligent machines and capture an outsized share of the productivity gains. The losers of the sorting are funneled into the low-skill service sector, where they will do the work the machines cannot do at wages the credentialed class will pay for the convenience. The middle — the stable, decent-wage, non-credentialed work that built the postwar working-class community — is disappearing. There is no longer a fallback position for the student who merely meets the average bar.

This compounds Sandel’s point about meritocratic humiliation in a way that is essential to understanding why the diagnosis from Article 5 has gotten worse rather than better in the last decade. In the postwar period, losing the educational sorting did not mean losing access to dignity. It meant the steel mill instead of the college. The steel mill paid a wage that supported a family, came with a pension and benefits, and was respected enough in the surrounding culture to anchor a community. The losing was real but the loss was bearable. Losing the educational sorting today means something different. It means being funneled into a service sector whose wages are stagnant, whose benefits are thin, whose work is increasingly precarious, and whose surrounding culture has stopped acknowledging the work as a contribution to anything. The losing is the same. The loss is much worse.

The complacent class understands this in its bones, even when it does not articulate it. Part of what is driving the household-level investment in the K-12 architecture, the matching technologies, the prep schools, and the consultants is the implicit recognition that the cost of falling out of the credentialed track is now higher than it has ever been. The households are not being irrational. They are reading the labor market correctly and responding to it with the only tools they have: invest more aggressively in their own children’s position, defend the existing residential and educational architecture, optimize every advantage the children can carry into the next round of the sorting. The investment is rational at the household level. It is also, at the system level, the mechanism by which the bifurcation gets worse, because every household that successfully invests in its own children’s position is a household that has helped reduce the share of seats available to the children of households that cannot make the same investment. The defense and the destruction are the same operation.

This is the recursion the complacent class is caught in. The bifurcation is real. Investing in your children is rational. The investment, when scaled up across the entire complacent class, is also the thing producing the bifurcation. The class cannot escape the trap by trying harder, because trying harder is what the trap is built out of. The escape requires the class to recognize that the system it is investing in is the system that is hurting the country, and the recognition requires a kind of self-perception the meritocratic frame is structured to prevent.

VIII. Why Complacency Is Self-Defeating

Cowen’s title contains the argument. The complacent class thinks it is preserving its advantages. By doing so it is strangling the dynamism that produced those advantages in the first place.

The argument is straightforward and uncomfortable. The American economy that produced the postwar middle class was built by people who took risks: started companies, moved across the country, married outside their hometown, took jobs in industries that did not yet exist, raised children in neighborhoods that mixed across class. The risk-taking was the source of the dynamism. The dynamism was the source of the abundance. The abundance was the source of the credentialed positions that the complacent class now holds and is now defending. The defense, however rational at the household level, has the cumulative effect of reducing the risk-taking that produced the abundance in the first place.

The startup formation data is the cleanest case. The decline in the share of new firms across the American economy is concentrated in exactly the demographic that the complacent class belongs to: the educated, the urban, the upper-middle-class, the people who have the resources to start companies but increasingly choose not to. There are explanations — the rise of giant incumbents, the consolidation of capital, the regulatory burden, the difficulty of competing against firms with massive scale economies — and the explanations are not wrong. They are also not complete. Part of the story is that the people most able to take entrepreneurial risk have, over the last forty years, become more risk-averse. They have professional jobs they do not want to leave. They have mortgages on houses in school districts they do not want to disrupt. They have children whose extracurricular schedules do not survive a parent’s career change. They have spouses whose careers are tied to specific cities. The household, having been organized around defending the position, no longer has the spare capacity to take the kinds of risks that built the position in the first place.

The interstate migration data tells the same story. Americans used to move when economic opportunity moved. The classic cases — the Black migration from the South to the industrial North, the Dust Bowl migration to California, the postwar movement from the Northeast to the Sun Belt — all involved millions of people pulling up stakes and following the work. The contemporary Americans most able to follow the work are the credentialed urban professionals whose skills are portable. They are also the least likely to actually follow the work, because the household’s investment in the existing school district, neighborhood, and professional network is too large to risk. The household stays put, and the regional economic divergence between thriving metro areas and stagnating ones widens, and the people who would have moved a generation ago to bridge the divergence stay home.

The intermarriage data tells the same story. The decline in cross-class marriage is concentrated in the upper-middle-class, where the dating and marriage markets have become so efficiently sorted by educational attainment that the pool of cross-class partners has effectively closed. Cross-class marriage is not a romantic ideal. It is an empirical mechanism by which information, expectations, vocabulary, and network access have historically flowed across class lines. When it stops happening, the information stops flowing. The complacent-class household no longer has anyone in its extended family who knows what the lives of the non-credentialed look like from inside, because no one in the extended family is non-credentialed. The whole compound family has been sorted into the same income bracket and the same educational profile, and the absence of dissent inside the household becomes the absence of any internal corrective on the household’s self-understanding.

Each of these mechanisms is rational for the household and destructive for the system. The household that does not move keeps its school district. The household that does not start a company keeps its salary. The household that marries another credentialed household compounds its position into the next generation. None of these decisions is wrong from inside the household’s frame. All of them, summed across the complacent class, are the structural reason American dynamism has declined for forty years.

Cowen frames the long-term consequence as a shift from a linear model of American history to a cyclical one. The linear model says that each generation has more opportunity than the last, that the country gets richer and freer and more capable over time, that progress is the default and stagnation is the exception. The cyclical model says that periods of stagnation are followed by periods of crisis, that the crisis breaks open the calcified structures and forces a new round of dynamism, and that the new round eventually settles into its own complacency, which produces its own crisis. The cyclical model is older than the American mythology. It is the model most of human history has run on. The complacent-class era, in the cyclical reading, is one phase of a longer cycle, and the next phase is going to be uncomfortable for the people the current phase has favored.

This is uncomfortable to write and uncomfortable to read because it implies that the resolution of the complacent-class problem will not come from the complacent class’s voluntary reform. Voluntary reform is what the complacent class has been promising for forty years. The promises have not produced reform because the household-level rationality of the defense is too strong for any individual household to break. The reform, if it comes, will come from outside the household’s frame — from a political eruption, an economic shock, a generational shift, or a combination of all three. The eruption will be ugly, because eruptions are. The complacent class will not enjoy it, because the eruption is what the class has spent forty years insulating itself against. The insulation is part of why the eruption, when it arrives, will arrive with so little warning to the people who built the insulation.

IX. The Author Is in the Frame

The author of this series has used the architecture this article describes. The matching technologies the article critiques are matching technologies the writer has used. The residential decisions the article traces — school district reasoning, zoning-driven housing choices, the whole bundle of optimization defaults about where to live and what to weigh — are decisions the writer has weighed. The K-12 architecture the article names is the architecture the writer has navigated. There is no point in pretending otherwise. The article is written from inside the system, not above it.

This is not a confession. It is a positioning. The Cowen thesis is sharper because the writer has used the apparatus the article critiques. The article would be less honest, not more, if it were written from a notional outside view that pretended the writer had no experience of the architecture being described. There is no such view. The architecture does not filter by identity before operating on a household. It operates on anyone who uses it, and it is difficult to live in America and make ordinary household decisions about where to live and how to educate children without touching it. The recognition that the article is built on is the recognition of having touched it.

This positioning matters for a specific reason. The voice of this article is the voice of someone who recognized the pattern by noticing his own behavior inside it first, not through abstract analysis of someone else’s life. The writer noticed that ordinary household decisions — their structure, their defaults, their optimization logic — matched the description Cowen had been publishing for a decade. The match was not a moral indictment; it was a structural observation. The architecture had the writer’s behavior in its description. Once the writer saw that, the pattern could not be unseen. The writing of this article is partly the working out of what to do with the recognition.

The honest answer to what to do with the recognition is: not much, individually. The structural pattern is structural. It does not yield to individual choices. Small revisions against the default — a residential decision that trades commute time for cross-household exposure, a refusal of the most aggressive of the optimization advice, summer programs that mix across income lines, friendships chosen against the grain of the matching technologies — each is real and each is a small revision. None of them break the pattern. The pattern is built out of millions of household-level decisions whose aggregate is the system. A single household’s revision changes that household’s trajectory. It does not change the system.

What the recognition does change is what the writer is willing to say in print. It changes the willingness to describe the system as a system rather than as the personal failing of the wrong households. It changes the willingness to use the word “we” in the sense of “people who have used this architecture” rather than “people who are innocent of it.” It changes the willingness to count the writer’s own behavior in the data. None of this is a substitute for structural change. It is the precondition for the structural change being thinkable, because the change will require a critical mass of people who have used the architecture to stop performing innocence about having used it.

The writer is offering this article to the readers in that spirit. The article is not asking the reader to feel guilty. Guilt is a useless category for structural problems. Guilt produces personal repentance and personal repentance does not move zoning boards or school admissions policies or labor market regulation. The article is asking the reader to do what the writer had to do: notice the pattern, notice their own participation in it, and stop performing the exemption that the meritocratic frame from Article 5 has been training them to perform. The exemption is the part that has to go first. Everything else follows.

X. What to Look At

The article will not end with a policy program. The reasons are the same as Article 5’s reasons. The diagnosis is structural and cumulative. It does not convert into legislation cleanly. The closest thing to a useful intervention the article can offer is a set of small acts of attention that the reader can practice, beginning today, that gradually shift the household’s frame from defense to recognition. None of these is a fix. All of them are the precondition for any fix that does not begin with the same complacency under a different name.

The first is to look at the residential income segregation index for the metropolitan area you live in. The data is public. Sean Reardon at Stanford and the Brookings Institution have both published interactive versions. Find your metro on the map. Look at where it ranks. Notice whether the ranking matches the political identity the residents of the metro use to describe themselves. The mismatch, where it exists, is the diagnosis applied to your specific city, and the diagnosis is more useful than any abstract version because it has your name on it.

The second is to count the number of cross-household friendships in your life that span meaningful differences in income, credential level, or neighborhood. Not acquaintances. Not service-provider relationships. Friends. People you have over for dinner. People whose households operate differently enough from your own that spending time together requires a small act of crossing. The count is, for most households that have used the matching architecture this article describes, very low. The count is not a moral failure. It is a structural feature of how the matching technologies and the residential zoning have organized the available friend pool. Notice the count. Notice what would have to change for the count to grow. The noticing is the first part of any change.

The third is to read the zoning code of the municipality you live in. Most readers have never done this. The zoning code is the architecture from Article 2 in its current local form. It is the document that determines who can build what, where, at what cost, and therefore who can move into the neighborhood and who cannot. The code is usually fewer than two hundred pages, written in technical but accessible English, and freely available on the municipal website. Read it. Notice what it permits and what it prohibits. Notice which of the prohibitions function as class filters even though none of them mention class. Once the code is in your head as a document rather than as an abstraction, the structural picture from this article becomes much harder to push away.

The fourth is to read the admissions policies of the public school district your children attend or would attend. Notice the gifted-and-talented criteria. Notice the magnet program admissions process. Notice the residency requirements. Notice the documentation the district requires from parents and what kinds of households can produce that documentation easily. The mechanisms by which the public schools are sorted are usually discoverable in twenty minutes of reading on the district’s website. The mechanisms are not hidden. They are routinely opaque to the parents whose children benefit from them, because the parents have never had to look.

The fifth is to ask, the next time the household uses one of the matching technologies, what the matching is doing. When you filter the Zillow listings by school district, ask yourself which households the filter is excluding from the listings you are seeing. When you swipe through the dating app, ask yourself which profiles the algorithm is not showing you. When the Common App suggests colleges to apply to, ask yourself how the algorithm built the suggestion list and what kinds of colleges the list contains and excludes. The matching technologies are not doing anything wrong. They are doing exactly what they were built to do. The asking is for the user, not the platform. The user is the one who has to notice that the choice on offer is also a sorting on offer, and that the sorting is the part the platform is not naming.

The sixth is to read the Chetty data on intergenerational mobility for the county or commuting zone you live in. Opportunity Insights publishes the data interactively. The map is searchable. Find your place. Notice the upward mobility rate for children born to the bottom income quintile in your specific commuting zone. Notice how it compares to the national average and to the commuting zones around it. The number is a description of the system as it operates where you live, and the number is more useful than any aggregate national number because it is your number.

None of these acts of attention is a program. None of them changes the system. All of them change the user’s relationship to the system, and the change in relationship is the precondition for any system-level change that does not begin with the same complacency that built the system.

The next and final article in this series brings all four temporal arcs together: the architectural design, the Reconstruction betrayal, the postwar abundance window, the elite reproduction engine, the moral diagnosis, and the structural diagnosis. The synthesis is built around Sandel’s deepest distinction, which the series has been preparing the reader to recognize for six articles: between honest commerce that pays for the good itself, and corrupted commerce that pays for position in a sorting mechanism whose value depends on its purchase being invisible. The carpenter builds the house. The Latin tutor teaches the language. The admissions consultant games the gate. These are not three versions of the same activity. The article that follows will explain why the difference matters.