The Reproduction Engine

This is the president of Harvard in 1922, writing to alumni who had complained about Jewish enrollment. His logic is self-executing: the presence of Jews causes anti-Semitism; therefore, to prevent anti-Semitism, we must prevent Jews. Read the sentence twice. The first reading delivers the racial prejudice the modern reader expects. The second reading delivers the deeper move, which is that Lowell has assigned causal agency for an effect (anti-Semitism) to its victims (Jews). The technical name for this is blaming the victim. The technical home of the move, in this letter, is the office of the president of the most prestigious university in the United States.

Lowell did not stop at the letter. He proposed an open numerical quota: a hard cap on the percentage of the entering class that could be Jewish. The Harvard Corporation balked. Not on principle. The Corporation balked on optics. A printed numerical quota was too nakedly discriminatory to survive in public, particularly after the Wilson-era reforms had made overt religious discrimination politically embarrassing in the kind of institutions that congratulated themselves on their Enlightenment values. So Lowell and his administration invented something more durable. They invented an admissions apparatus that would produce the same exclusionary outcome through a vocabulary nobody could call discriminatory in court.

The apparatus had a name. They called it the “New Plan,” introduced in 1926. We now call it “holistic admissions.” Applicants were required to submit photographs, so admissions officers could see what the applicant looked like and identify her likely religion and race. Applicants were required to list “race and color,” “religious preference,” and the “maiden name of mother.” They were assessed on “character,” “leadership,” and “manliness” — categories that no admissions officer was required to define and no court could review. Admissions officers visited secondary schools to interview applicants in person, an innovation that conveniently allowed the office to inspect the applicant’s home neighborhood, accent, and dress. Teacher recommendations became mandatory, which meant the admissions process could pre-filter for students whose teachers shared the office’s implicit understanding of what an “ideal” Harvard man looked like. A “geographic distribution” requirement was added, explicitly to dilute the urban Northeast applicant pool with boys from Iowa and Montana, where Jews were rarer.

Inside the admissions office, applicants from New York were filed under one of three codes. J1 meant conclusively Jewish. J2 meant probably Jewish. J3 meant possibly Jewish. The classification was based on surname and photograph. This is not folklore. It is in the admissions files. It was disclosed in federal litigation. It is a documentary fact.

The result was instant. Harvard’s Jewish enrollment fell from 28 percent in 1925 to approximately 15 percent by 1933, where it was held for decades. Yale and Princeton adopted parallel measures. Princeton, which had fewer Jewish applicants to begin with because of its social reputation, codified the “character” interview into its own admissions apparatus. The architecture spread across the Ivy League and the elite liberal arts colleges that emulated it. Within a single decade, every selective institution in the United States had a vocabulary it could use to exclude unwanted applicants without naming the exclusion.

The vocabulary survived the Civil Rights era. It survived the Conant-era SAT meritocratic opening. It survived the 1960s push to open the elite institutions to working-class and minority applicants. It survived the 1990s diversity push and the 2003 Grutter v. Bollinger decision. It survived the 2023 Students for Fair Admissions v. Harvard ruling that struck down race-conscious admissions on the theory that it was discriminating against Asian Americans the way Lowell’s machine had once discriminated against Jews. The vocabulary is doing the same political work in 2026 that it did in 1926. It works for the same reason it worked then: because it is opaque, because “character” cannot be falsified on a form, and because an admissions officer’s gut feeling about a 17-year-old is beyond judicial review.

This article is about the entire arc of that machine. It walks four centuries of elite reproduction at the prestigious American universities, from the Latin examinations of 1636 through the Lowell apparatus of 1926, through the brief postwar opening, through test prep capital, through the independent educational consultant industry, through the Arcidiacono numbers from the SFFA trial, through the development cases Daniel Golden documented, through Varsity Blues, and out to the Chetty network science that quantifies what the credential is actually worth and what the credential’s value depends on. The thesis is the one Jerome Karabel arrived at in The Chosen after years in the Harvard archives. The prestigious American universities have never been in the business of distributing instruction. They have been in the business of sorting and reproducing an elite. The instruction is the cover story. The network is the product. The machine works as designed. The design is the scandal.

I. Four Centuries of Latin and Letters

Harvard was chartered in 1636, sixteen years after the Mayflower, before the Massachusetts Bay Colony could reliably feed itself. The founding document gave the reason: the founders dreaded “to leave an illiterate Ministry to the Churches, when our present Ministers shall be in the Dust.” The institution was a clerical reproduction mechanism. Its job was to train the next generation of ministers for the Puritan churches that the colony required for its social order.

The curriculum was classical. Latin, Greek, Hebrew, logic, divinity. Students were admitted if they could “read the original of the Old and New Testament into the Latin tongue, and resolve them logically” — a filter that had nothing to do with raw aptitude and everything to do with whether the applicant had spent his childhood being tutored in Latin, which is to say whether the applicant was a son of a household that could afford to keep him out of the fields. The filter was the curriculum. The curriculum was the filter. There was no separate admissions office, because the curriculum did the sorting before the application.

Harvard’s own historical record notes that Puritan families in England sent their sons across the Atlantic to be educated there, marketing the institution to the colonial elite from its opening year. The colonial college sequence that followed — William & Mary in 1693, Yale in 1701, Princeton in 1746, Columbia in 1754, Penn in 1755, Brown in 1764, Dartmouth in 1769 — repeated the pattern with denominational variations. These were not universities in any modern sense. They were small clerical seminaries with a sideline in finishing the sons of merchants, planters, and ministers. There was no “admissions office.” There was no “application” in the contemporary sense. There was a letter of introduction from a clergyman or a prominent alumnus, a fee paid by the family, and a classical examination that functioned as a cultural passport. A farmhand could not pass it because a farmhand had never been taught to construe Cicero. The exclusion was not racial in the modern sense. It was class-based, and the class structure was itself maintained by the prior question of who got tutored in Latin as a child.

This is the prehistory that matters. For more than 250 years before the word “meritocracy” was coined in 1958 by the British sociologist Michael Young (who, importantly, intended the word as a pejorative warning), the prestigious American colleges were straightforwardly reproducing a hereditary Protestant gentry. The mechanism was the curriculum itself. The sorting happened upstream, in who got taught Latin. Nobody at Harvard in 1700 was confused about what the institution was for. The pretense that Harvard was sorting for talent rather than for inheritance is a much later invention — an invention that became necessary only when the population eligible to attempt the Latin examination grew large enough that the elite needed a more sophisticated filtering mechanism.

The growth happened in the late nineteenth century. The American population swelled with immigration from Eastern and Southern Europe. Public secondary education expanded. Jewish families in particular, with their long cultural premium on literacy and study, began producing sons who could pass the classical entrance examinations the colleges had assumed only their traditional clientele could pass. The filter that had worked for two and a half centuries was about to fail. The institution had a choice. It could let the new sons in, redistributing access to the credential it was now offering. Or it could invent a more opaque filter that the new sons could not pass. It chose the second.

II. The 1922 Exclusion Crisis

By 1900, Jews were 7 percent of Harvard’s undergraduates. By 1922, they were 21.5 percent. The increase was the product of the “top seventh rule,” under which Harvard admitted students based on academic standing in their secondary schools. The rule was working too well. Brilliant sons of Jewish immigrants from the Lower East Side and Brooklyn were displacing the sons of Beacon Hill in measurable numbers. The Brahmin alumni began to write letters to the president.

President A. Lawrence Lowell, Boston Brahmin himself, scion of the Lowell family that had given American letters Robert and James Russell, decided to fix the problem. He left a paper trail. The historian Jerome Karabel went into the Harvard archives in the early 2000s and emerged with a documentary record so complete that the institution’s own subsequent denials of intent collapse on first contact. The Chosen: The Hidden History of Admission and Exclusion at Harvard, Yale, and Princeton is the result, and it remains the indispensable text on the subject.

The 1922 letter at the top of this article is one of dozens. Alumni wrote in to cheer Lowell on. The Harvard attorney Morris Gray Jr. wrote Lowell in 1920 first approving of the way the university was “dealing with the Negro” and then noting that “the question of the Jew is, however, more difficult and, so far as Harvard is concerned, more serious.” The investor Hall Allen wrote that “a rigid restriction… against the Jews is not only necessary but most desirable.” The Crimson’s 2015 retrospective, “President Lowell’s Quotas,” reproduces the correspondence in context. The letters are not difficult to read. The letters are not difficult to understand. The letters do not require the modern reader to perform any acrobatic interpretive maneuver. The letters say what they mean.

What they mean is that the Harvard establishment of 1920 wanted to exclude Jews and was casting around for a politically viable mechanism by which to do it. Lowell’s first proposal was the open numerical quota: a hard cap on the percentage of the entering class that could be Jewish. The Harvard Corporation rejected the proposal. They did not reject it because they disagreed with its purpose. They rejected it because they understood that a printed numerical quota was the kind of explicit discrimination that, in the post-Wilson era, would become a public scandal. The university could not afford to be seen to be running a quota. It could afford to be seen to be running an admissions process with subjective criteria. The difference between the two was a vocabulary problem, not a policy problem.

So Lowell and his administration solved the vocabulary problem. They invented holistic admissions.

The 1926 New Plan introduced the architecture every modern American applicant now navigates without knowing where it came from. Applicants were required to submit photographs. Why? Because photographs allow the admissions office to assess the applicant’s race and likely religion before reading anything she has written. Applicants were required to list “race and color,” “religious preference,” and the “maiden name of mother.” Why? Because a Jewish mother’s maiden name is the most reliable single indicator of an applicant’s Jewish heritage, more reliable than the applicant’s own surname (which might have been Anglicized) or the family’s contemporary religious practice (which might have been suppressed). Applicants were assessed on “character,” “leadership,” and “manliness.” Why? Because these categories were not defined anywhere, could not be falsified by any external evidence, and depended entirely on the admissions officer’s subjective evaluation, which the office trusted to track its own implicit ranking of who looked, sounded, and carried themselves the way Harvard wanted its students to look, sound, and carry themselves. Admissions officers visited secondary schools to interview applicants in person. Why? Because the in-person interview let the office observe the applicant’s accent, dress, manners, and home environment. Teacher recommendations became mandatory. Why? Because teachers in the secondary schools the Harvard pipeline drew from understood, often without being told, what the office was looking for, and would write the recommendations accordingly. A “geographic distribution” requirement was added. Why? Because the urban Northeast was where the Jewish applicants lived, and diluting the pool with boys from Iowa and Montana meant fewer Jewish applicants in proportion.

The classifications. New York applicants were filed under one of three codes:

• J1 — conclusively Jewish
• J2 — probably Jewish
• J3 — possibly Jewish

Based on surname and photograph. The admissions files are the source. Karabel reproduces them. The Crimson’s 2018 piece on the SFFA filings reproduces the legal record of their disclosure. The classifications were not folklore. They were administrative practice at the most prestigious educational institution in the United States, practiced openly within the walls of the office and, when discovered, defended as a necessary tool for managing “the Hebrew problem.” The phrase “the Hebrew problem” appears in Lowell’s own correspondence. The quotation marks here are around what he actually wrote.

The result was instant. Harvard’s Jewish enrollment fell from 28 percent in 1925 to approximately 15 percent by 1933. The percentage was held at roughly that level for decades. Yale and Princeton adopted parallel measures and produced parallel reductions. The architecture worked.

This is the foundational fact about modern elite admissions. The entire scaffolding that contemporary college counselors now coach students to navigate — the personal essay, the recommendation letters, the interview, the “well-rounded” extracurricular portfolio, the “leadership” narrative, the “character” assessment, the “demonstrated interest,” the “authentic voice,” the carefully cultivated portfolio of activities that signals depth rather than breadth — was engineered between 1922 and 1926 by a Harvard president who wanted to exclude Jews. Every category in the contemporary admissions vocabulary was named by Lowell or his administration as part of the apparatus for replacing the open numerical quota the Corporation refused to print.

Mark Blyth’s frame for institutional ideas is precise here. The vocabulary of holistic admissions is not a neutral description of a sorting process. It is an institutional weapon. It does political work for a specific class of beneficiaries by using language that makes the work invisible. The language was invented in 1922 to do political work. It is doing the same political work today. The beneficiaries have changed marginally — Jews, after the post-1960s opening, were largely allowed in, and the apparatus migrated its filtering toward other targets — but the structural function of the vocabulary has not changed. It still produces opacity. Opacity is still what the apparatus requires to function as a sorting mechanism. The apparatus still functions.

III. Legacy Preferences: A 1920s Invention

Legacy preferences are routinely defended by their beneficiaries as a venerable tradition. The tradition is precisely 100 years old, which in the life of Harvard (now 390 years old) is a quarter of the institution’s existence. And it was invented in the same crisis as holistic admissions, for the same purpose, by the same people.

The mechanics. Once Lowell’s open numerical quota was politically dead, the Harvard administration needed a non-numerical mechanism to guarantee Anglo-Saxon Protestant continuity. The legacy preference — explicit bonus weighting in the admissions process for the sons of Harvard alumni — solved the problem mechanically. If the father was Harvard ’02, the son got in. Jewish applicants of the 1920s, by definition, did not have Harvard fathers, because the Harvard student body of the 1890s had been almost entirely WASP. Legacy preference was therefore a non-numerical exclusion mechanism that produced the same outcome as a numerical Jewish quota, while permitting the institution to maintain that no quota existed. The political innovation was elegant. The moral content was identical to the numerical quota the Corporation had rejected on optics.

The 2018 Students for Fair Admissions v. Harvard trial in the District of Massachusetts surfaced the first independent statistical analysis of the preference’s contemporary weight. Peter Arcidiacono of Duke University, the plaintiff’s expert witness, was given access to six years of Harvard admissions data under the protective order of the federal litigation. He pulled the data, ran the analyses, and published the findings, first in his expert report and then in NBER Working Paper 26316 with Josh Kinsler.

The numbers are the part of the public record that the institution would most prefer the public not to remember. They should be presented, in sequence, without softening.

The acronym for the four bottom categories is ALDC — Athletes, Legacies, Dean’s List, and Children of faculty and staff. ALDC applicants make up about 5 percent of the applicant pool but 30 percent of the admitted class. Forty-three percent of all white admits to Harvard are ALDC. Less than 16 percent of admits in each of the non-white categories are ALDC. Arcidiacono’s regression models, run on the same data, estimated that roughly three-quarters of white ALDC admits would have been rejected if treated as non-ALDC white applicants. In plainer English: most of Harvard’s white students are at Harvard because of preferences specifically designed to benefit white students.

That sentence should not slide past on first reading. Most of Harvard’s white students are at Harvard because of preferences specifically designed to benefit white students. Not because of merit-blind admissions. Not because of meritocracy on the dimension that the SAT supposedly measures. Not because of holistic assessment of character. Because of named preferences whose statistical effect is large enough that, if those preferences were removed, three out of four of the white admits the preferences benefit would not be admitted at all. The preferences are not a tiebreaker. The preferences are the principal cause of admission for the population they benefit.

Justice Neil Gorsuch noticed. In his concurrence in the 2023 SFFA decision, Gorsuch wrote that Harvard’s ALDC preferences “undoubtedly benefit white and wealthy applicants the most.” The Court struck down race-conscious admissions on the theory that they discriminated against Asian Americans. The Court did not touch legacy preferences. It did not touch the Dean’s Interest List. It did not touch the recruited athlete pipeline. The preferences with statistical effect six or seven times stronger than the race-conscious admissions the Court abolished were left intact, because the plaintiffs did not sue over them. The Court’s ruling is a model of selective scrutiny: it took the smaller intervention favoring under-represented minorities and called it an unconstitutional violation of equal protection, while leaving the larger interventions favoring wealthy whites alone, on the grounds that it had not been asked.

Five states have since banned legacy preferences at their public universities: Colorado in 2021, then Virginia, Maryland, Illinois, and California in 2024. California’s ban extends to private nonprofits as well. Virginia’s ban passed both legislative chambers unanimously and was signed by Governor Glenn Youngkin, a Republican. Maryland’s ban covers any institution that receives state funding. The Ivy League is untouched by any of these bans, because the Ivy League is private and does not depend on state appropriations the way public universities do. As of the writing of this article, no Ivy League institution has voluntarily ended legacy preferences. Harvard maintained its preferences through 2024 over explicit pressure from the Biden administration and from a portion of its own faculty. The federal civil rights complaint filed by Lawyers for Civil Rights, arguing that legacy preferences violate Title VI by disproportionately disadvantaging non-white applicants, is pending.

The Niklas Luhmann reading of this situation is that the admissions code — admitted/rejected — cannot represent the upstream causes that determine which applicants enter the pool with what advantages. The code processes class as merit, not because the admissions officers are corrupt, but because the code itself has no slot for “your father donated $400,000” or “your mother is faculty” or “you have been playing squash with a private coach since you were eight.” The code processes the visible application and reports a decision. The structural blindness is required for the sorting to function as sorting. If the code could see the upstream causes, it would have to acknowledge that what it was sorting was not merit but inheritance, and the institution would no longer be able to maintain the fiction that produces its prestige. The fiction is the prestige. The opacity is the credential. Lowell understood this in 1922. Every admissions officer at every Ivy League institution understands it now. The conversation about it is held in private. The public-facing language is “character” and “leadership” and “well-rounded.”

IV. The Athletic Recruit Pipeline

If legacy is the front door, the athletic recruit pipeline is the side entrance no one talks about. Harvard’s recruited athletes are admitted at 86 percent — not 8.6 percent, 86 percent — against a non-recruited base rate of about 5 percent. The racial and economic composition of the recruited athletes is the part of the picture that the admissions office’s public statements most carefully avoid:

• 83 percent of Harvard recruited athletes are white — in a class that is 53 percent white.
• 46.3 percent of recruited athletes come from households earning $250,000 or more.
• 3.2 percent of white recruited athletes qualify as economically disadvantaged.

The mechanism is the sports themselves. Harvard does not primarily recruit basketball and football players, the sports that admit the broadest cross-section of American athletic talent. It recruits rowers, fencers, squash players, sailors, equestrians, lacrosse players, water polo players, golfers, and skiers. These are sports that require boats, horses, country clubs, private coaching from age eight, and in many cases boarding schools with dedicated facilities. They are not sports a child develops on a public playground. They are sports a child develops in households with the disposable capital and the parental time to subsidize fifteen years of private athletic instruction. The sports are means-testing in reverse. They select for wealth because only wealthy families can develop a child into an elite squash player by 17.

The credential laundering is then complete. A child who would never have been admitted on academic measures alone is admitted as a “recruited athlete” whose scholastic floor is well below the general applicant pool, on the strength of an athletic accomplishment that the family’s capital underwrote from kindergarten. The Harvard admissions office can now describe its admit as “a recruited athlete in squash” rather than “the daughter of a hedge fund partner who paid for twelve years of private squash coaching at a country club her family belongs to.” The first description is admirable. The second description is Brahmin. Both are accurate. The first is the one the public sees.

This is functionally identical to what Rick Singer was prosecuted for. The only difference between the legitimate athletic recruit pipeline and Singer’s “side door” is that in Singer’s version, the child could not actually play the sport, so Singer had to fabricate the credential through Photoshopped photographs and bribed coaches. In the legitimate version, the child can actually play the sport — but only because the family paid for twelve years of private instruction in it. Both versions produce admission to the Ivy League on the basis of an athletic credential whose underlying cause is the family’s capital. The legitimate version is legal because the family paid for the sport directly, rather than paying a middleman to fake it. The line between the two versions is the legal line between “genuine credential purchased through years of training” and “fake credential purchased through wire transfer.” The line, moralized, is that one took longer.

V. The Meritocratic Window: Conant, the SAT, the GI Bill, Pell

For roughly four decades in the middle of the twentieth century, the elite reproduction machine ran at reduced capacity. The story is told most fully in Nicholas Lemann’s The Big Test and in Karabel’s The Chosen. It is the story the postwar mythology of meritocracy is built on, and like most of the mythology it is partly true.

James Bryant Conant became president of Harvard in 1933, at the heart of the Lowell-quota era. Conant was a chemist by training, not a Brahmin, and he was genuinely repelled by the Lowell admissions machinery. He invoked Thomas Jefferson’s phrase “natural aristocracy” — a meritocracy of the intellectually gifted from every walk of life, educated to high standards and entrusted with the responsibility of governing. Conant’s instrument was the Scholastic Aptitude Test, originally developed by Carl Brigham (a eugenicist, awkwardly) from Army IQ tests. Conant commissioned Henry Chauncey to build a national scholarship program that would use the SAT to find brilliant poor kids anywhere in America and bring them to Harvard. In 1948, Chauncey opened the Educational Testing Service in Princeton. By the late 1950s, the SAT had become the dominant instrument for selective college admissions. Conant’s vision was that the test would be an aptitude measure — something that could not be studied for, that would reveal innate talent regardless of schooling.

The vision was sincere. It was also wrong about the test. We will get to that.

The GI Bill, the Servicemen’s Readjustment Act of 1944, opened the door from the other side. Approximately 2.3 million veterans attended colleges and universities under the bill; another 3.5 million received vocational training. The percentage of Americans with a bachelor’s degree rose from 4.6 percent in 1945 to roughly 25 percent a half-century later. Working-class veterans — Italian-Americans, Irish-Americans, Jewish-Americans — walked into classrooms their fathers had never imagined entering. The elite universities were never the primary beneficiaries (Columbia and NYU took more veterans than Harvard did, and CCNY and Brooklyn College did the heavy lifting in New York), but the composition of the educated American class shifted permanently. The previous article documented the racial caveats: Black veterans in the South were excluded from most state universities and channeled into underfunded HBCUs. Japanese-American veterans returning from internment were often unable to access benefits. The “universal” opening was partial from the start. But within the populations the federal government chose to serve, the opening was real.

Pell Grants in 1972, championed by Senator Claiborne Pell, extended need-based federal aid to low-income undergraduates regardless of veteran status. The maximum grant in 1975 covered roughly 80 percent of the cost of attending a public four-year university. By 2025, the maximum Pell covers approximately 25 percent. The opening of 1972 was followed by a slow closing across the next half-century, exactly contemporaneous with the state-appropriation collapse the previous article documented.

The sequence matters. Between roughly 1945 and 1975, the meritocratic opening at the prestigious American universities was real. It was not an illusion. A working-class kid with high SAT scores in 1960 had a meaningfully better chance of admission to Harvard than the same kid would have had in 1925 or has now. The Conant project moved the elite institutions, in measurable ways, in the direction of the natural aristocracy Jefferson described. The question that needs to be asked is not whether the opening happened, because it did, but how thoroughly and how durably. The honest answer is: not very, and not for long.

The opening was always conditional. It was conditioned on the SAT’s being a measure of innate aptitude rather than a measure of preparation, which it turned out not to be. It was conditioned on the elite institutions remaining willing to admit students whose social capital did not match their test scores, which they did unevenly and grudgingly. It was conditioned on a federal government willing to fund need-based aid at scale, which the federal government became progressively less willing to do after 1980. It was conditioned on the absence of a private apparatus designed to convert family capital back into admissions credentials, which is exactly what the next two decades produced.

VI. How the Window Closed: Test Prep, Consultants, Sticker Prices

The SAT survived as a putative aptitude test exactly as long as it took Stanley Kaplan to prove it wasn’t one. Kaplan started tutoring in his parents’ Brooklyn basement in 1938 and added SAT preparation in 1946. For decades, the Educational Testing Service insisted publicly that the SAT could not be coached. Kaplan kept raising scores. In 1979, the Federal Trade Commission investigated and confirmed that Kaplan’s methods produced an average gain of about 25 points on math and verbal — small as a per-student effect, large as a population-level marker that the test was teachable. The aptitude pretense was over. The test was a preparation test. The students whose families could afford preparation outscored the students whose families could not.

In 1984, Kaplan sold his company to the Washington Post Company for $45 million. By 2009, Kaplan, Inc. accounted for 58 percent of the Washington Post Company’s total revenue. A newspaper company had become, structurally, a test-prep company with a sideline in journalism. The point is not that the Washington Post became disreputable. The point is the magnitude. Test prep was such a profitable industry that it could absorb a major American newspaper as a corporate ornament. Princeton Review launched in 1981. The modern industry now spans tutoring franchises, one-on-one private tutors at $300 per hour, boot camps, online platforms, and residential summer programs. Affluent families in the Northeast and the Bay Area routinely spend $5,000 to $25,000 per child on SAT and ACT preparation, and sometimes much more. The geographic concentration of top scores follows family income and zip code with brutal linearity. The relationship is so strong it is visible on a map.

The independent educational consultant industry is the second layer. IECs charge an average of $200 per hour, with comprehensive four-year packages ranging from $850 to $10,000 for standard services and up to $200,000 or more for high-end concierge service at the top of the market. The industry is unregulated. The Independent Educational Consultants Association is a voluntary trade group. There is no licensing body. There is no continuing education requirement. There is no malpractice insurance norm. There is no public complaint process. Anyone can call themselves a college consultant. Anyone can charge $200,000 for a four-year package. The market sets the price. The market sets the price because the buyers have the money and they are paying for an outcome the seller cannot guarantee but can promise to optimize for.

What the IEC sells is the hidden curriculum of the Lowell admissions machine. How to frame the personal essay so that it sounds “authentic” in the way the admissions office is trained to receive as authentic. How to stage the extracurriculars across four years so that they read as a “narrative” rather than a list. Which teachers to ask for recommendations and how to seed the conversation so that the recommendation comes back saying what the office wants to hear. How to perform “character” and “leadership” in the ways Harvard has been scoring those categories for a century. How to identify the right activities, the right summer programs, the right service trips. How to build a profile that looks unmistakably like the kind of profile the office is hunting for, while sounding to the office’s reader as if it had emerged organically from a child’s spontaneous interests. A bright first-generation kid from a Title I high school does not know any of this. The IEC client does. This is the moat.

Anthony Jack’s ethnography The Privileged Poor: How Elite Colleges Are Failing Disadvantaged Students documented this from the inside at “Renowned University,” Jack’s anonymized name for Harvard. Half of the poor Black students Jack interviewed at the school had attended boarding, day, or prep high schools on scholarship before arriving. One third of the poor Hispanic students had similar pre-Harvard backgrounds. Jack called these students “the privileged poor” — low-income kids who arrived at Harvard already fluent in the hidden curriculum, because the prep school they had attended had taught them how to dress, how to speak to faculty, how to ask for help, how to navigate office hours, how to sit at a formal dinner. The students Jack called “the doubly disadvantaged” — poor students from distressed public high schools, who arrived at Harvard without that prep-school grammar — remained under five percent of Harvard’s student body. The opening that the Conant generation had built was admitting low-income students at a higher rate than at any time in the institution’s history. It was admitting them disproportionately from the prep school pipeline. The opening was real and the opening was narrow, and the narrowness was a function of the same hidden curriculum the Lowell apparatus had built.

The third layer is the sticker price. The cost of attending a private four-year college in the United States has risen at roughly 2.5 times the rate of inflation since 1980. Elite institutions are nominally “need-blind” in their admissions, and they are in fact need-blind in the technical sense that the financial aid office and the admissions office do not communicate during the application review. But the need-blind policy sits atop an application funnel that has already been filtered by zip code, by school district, by SAT prep, by IEC coaching, and by the long sequence of upstream advantages that the family’s capital purchased before the application was submitted. The applicant the need-blind office reviews is already the product of the apparatus the need-blind policy is supposed to be neutral toward. Need-blind admissions, in this context, means “admissions blind to the consequences of need for the small number of needy applicants who survived the upstream filtering.” It does not mean what the phrase sounds like it means.

VII. Development Cases: The Front Door Is a Revolving Door

Daniel Golden’s 2006 book The Price of Admission: How America’s Ruling Class Buys Its Way Into Elite Colleges — And Who Gets Left Outside the Gates is the definitive investigation of how donor children get into the elite schools. Golden, a Pulitzer-winning Wall Street Journal reporter, documented the category that elite universities internally call “development cases” — applicants whose parents have given, are giving, or are expected to give significant amounts of money to the institution.

The Kushner case is the clearest public example. In 1998, Charles Kushner, then a New Jersey real estate developer, pledged $2.5 million to Harvard, payable in $250,000 annual installments over ten years. He met with Harvard president Neil Rudenstine to discuss a scholarship program. Jared Kushner, whose high school record his teachers and administrators described to Golden as unexceptional, was admitted to Harvard the following year. Golden interviewed officials at Jared’s high school who were stunned by the admission. They had not expected him to be admitted to Harvard. They had expected him, on his academic record alone, to be admitted to a regional college of considerably lower selectivity. The donation preceded the admission. The admission followed the donation. Golden’s reporting documented the sequence. A Kushner Companies spokesperson called the allegation “false” without specifying which part. Harvard declined to comment.

The Kushner case is famous only because Jared later married the president of the United States’ daughter. The category is not rare. Golden documented dozens of parallel cases at Duke, Stanford, Notre Dame, and across the Ivy League. Universities maintain “watch lists” of promising donor families. Development offices coordinate with admissions offices, formally or informally. The Dean’s Interest List at Harvard — distinct from the Director’s List, which serves a different population — is populated largely by the children of donors, prospective donors, trustees, celebrities, and politically useful families. These applicants are admitted at 42 percent, seven times the base rate. The Dean’s List is not a meritocratic line item on the admissions spreadsheet. It is the development office’s instrument for converting prospective philanthropy into actual admission slots.

The crisp formulation. The Dean’s List is a side door that Harvard runs itself. The price of admission through the Dean’s List is $2.5 million. The price of admission through the side door Rick Singer was prosecuted for running was $250,000 to $500,000. The Singer side door was an order of magnitude cheaper, and thus, naturally, criminal.

VIII. Varsity Blues: The Arrest of the Wrong Criminal

On March 12, 2019, the United States Attorney for the District of Massachusetts unsealed indictments against 50 people in Operation Varsity Blues. The central figure was William “Rick” Singer, who had run a Newport Beach college counseling business called The Key. Singer pled guilty and cooperated. His 42-month sentence, handed down in January 2023, was the longest in the case.

Singer’s schemes had three components. First, he bribed test proctors to change answers on SAT and ACT exams, or to allow stand-in test-takers to sit the exams in place of the students. Second, he bribed university athletic coaches at USC, Yale, Stanford, Georgetown, UCLA, and Wake Forest to designate students as “recruited athletes” in sports the students did not play — water polo, sailing, tennis. He created fake athletic profiles. He Photoshopped photographs of students onto the bodies of actual athletes. In some cases he sent the fabricated applications through the coach’s recruiting pipeline so that the admissions office never knew the recruiting decision had been bought. Third, he ran a sham charity, the Key Worldwide Foundation, so that parents could deduct their bribes from their federal taxes.

Singer called his service “the side door.” In recorded phone calls the FBI played at trial, he explained the architecture to parents in his own words. The front door, he told them, was “you get in on your own.” The back door was “ten times as much money,” meaning legitimate development-case donations of the kind Charles Kushner had made. The side door was Singer’s service: cheaper than the back door, faster than the front door, with a guaranteed outcome. Singer’s framing was not invented for the prosecution. It was his sales pitch. He understood, more clearly than most of his clients, that the side door was a market arbitrage between the front door and the back door — that what he was selling was access to the same outcome the back door provided, at a discount, to families who could not afford the back door but could afford a quarter of a million dollars and a ten-year commitment to a sham charity.

This is the point of Operation Varsity Blues, and the federal prosecutors who indicted Singer were structurally incapable of seeing it. Singer was not running a different scheme. He was running the same scheme at a lower price point with worse documentation. Harvard’s Dean’s Interest List admits donor children at 42 percent and does not go to prison. USC’s water polo coach accepted $250,000 to recruit a non-athlete and went to prison. The only daylight between these two arrangements is that Harvard’s version is called “fundraising” and is reported on a 990, while USC’s coach’s version was called “bribery” and was reported on a wiretap. The structural function of the two arrangements is identical: a transfer of family capital to a university, in exchange for the admission of a child whose academic record would not have been competitive on its own. The legal distinction is that one of the transfers went into the institution’s general fund and the other went into a coach’s personal account. The moral distinction is that one is performed openly and the other was performed in the dark. The moral distinction matters less than the law assumes it does.

The sentencing disparity is the part of the public record that should be repeated, slowly, in the same room as the Dean’s List statistics. Felicity Huffman paid $15,000 to have her daughter’s SAT answers corrected by a Singer-bribed proctor. She pled guilty, expressed contrition, and served 14 days in federal prison. Roughly two-thirds of the Varsity Blues defendants served three months or less. Many served no time at all. The prosecutorial restraint was striking. Felony fraud convictions for wealthy white-collar defendants in an embarrassing celebrity case were treated as embarrassments to be wrapped up quickly with cooperation deals and short symbolic sentences.

Now compare two other mothers, neither of whom was a celebrity, neither of whom paid a college consultant a quarter of a million dollars, neither of whom had access to a Singer-grade bribery network. Both committed structurally identical crimes — using a false address to secure educational access for their children — and both were prosecuted by the state.

Kelley Williams-Bolar, a single mother in Akron, Ohio, used her father’s home address to enroll her two daughters in the Copley-Fairlawn School District, a mostly white, higher-performing district adjacent to her own. She was charged with two felonies for tampering with records. She was sentenced to nine days in jail plus probation. The felony convictions blocked her teaching credential, ending the career path she had spent years training for. Ohio prosecutors cited her case at Felicity Huffman’s sentencing hearing, telling the federal judge: “If a poor single mom from Akron who is actually trying to provide a better education for her kids should go to jail, there is no reason that a wealthy mother with the resources should not also go to jail.” The judge gave Huffman 14 days. The arithmetic is in the record.

Tanya McDowell, a homeless Connecticut mother, used a friend’s address to enroll her six-year-old son in a Norwalk kindergarten. She was sentenced in 2012 to five years in prison. (Her sentence also involved drug charges, but the enrollment fraud was the lead story of the prosecution and the lead headline of the local coverage.) Five years in prison for using a friend’s address to enroll a child in kindergarten.

The asymmetry is the story. Wealthy parents who paid to cheat the college admissions system received short symbolic sentences. Working-class mothers who committed structurally identical fraud — using a false address to secure a better education for their children — received multi-year sentences and lost their teaching credentials. The state treats class-jumping by the poor as a serious crime. The state treats class-laundering by the rich as a cultural embarrassment. Both responses are forms of state-enforced sorting. Both responses preserve the property line between the populations the educational sorting machine is built to favor and the populations it is built to refuse.

The Sandel reading of this is the carpenter/house distinction the series will develop fully in its synthesis article, but the seed of it is already visible. What the wealthy parents bought through Singer and what the wealthy parents buy through the Dean’s List is not instruction. It is position in a sorting mechanism whose value depends on the buyer not being publicly identified as a buyer. The 1922 Lowell letters had to be hidden for the sorting to function. The Arcidiacono numbers from the SFFA trial had to be obtained through federal litigation, against Harvard’s sustained resistance, before they could become public. The Singer wiretaps had to be played in court before the public understood that the side door and the back door were the same architecture at different price points. In every case, the value of the sorted credential depends on the opacity of the sorting. As soon as the public sees the mechanism clearly, the credential begins to lose its meaning. This is why the institutions defend the opacity so vigorously. It is not because they are evil. It is because the opacity is the credential, and they cannot afford to lose either.

IX. The Network Is the Product

The hardest thing to say about an elite undergraduate education is this: the instruction is not the point, and it is not even especially good.

You can read The Republic at Binghamton. You can learn calculus at Santa Cruz. The libraries at Michigan and Wisconsin are as good as anything in the Ivy League. The faculty pool overlaps significantly — professors rotate through postdocs at all the major research universities, and the median tenured faculty member at a flagship state institution holds a Ph.D. from one of the same dozen institutions her Ivy League counterpart holds. The intellectual goods that elite universities sell — access to learned faculty, well-stocked libraries, rigorous coursework, scholarly conversation — are available, in functionally similar form, at hundreds of less prestigious institutions across the country, often at a fraction of the cost.

Stacy Dale and Alan Krueger’s 1999 study, replicated and extended in 2014, found that once you control for the fact that selective colleges admit students with higher earning potential to begin with, the wage premium of attending a more selective institution essentially disappears for most demographic groups. The exceptions are interesting and important, and we will return to them in the steelman section. But the general finding is stark. Students who attended more selective colleges did not, on average, earn more than students of equivalent aptitude who applied to the same schools but attended less selective ones. The Harvard premium, on a strict apples-to-apples comparison net of selection bias, is small or nonexistent for the median upper-middle-class white applicant. The conventional story that “going to Harvard” produces a measurable financial return on investment turns out, in the careful empirical work, to be largely a story about who Harvard admits, not about what Harvard does to them after admission.

So what is the Harvard premium, if it isn’t instruction and isn’t the wage differential the popular story claims it is?

It is the network. It is access to alumni internship pipelines that the public university student does not even know exist. It is access to peer groups optimized for success, where the students sitting next to you in the dining hall already have the cultural fluency and the family connections you would otherwise have to spend a decade acquiring. It is the signal a Harvard diploma sends to a hiring manager who has been trained, through the same prestige hierarchy, to interpret that signal as evidence of worth. It is the investors who open their phones when a Harvard founder cold-emails. It is the romantic partners who assume a Harvard spouse is a safe bet. It is the hidden curriculum — how to dress, how to write a professional email, what to call adults, how to ask for an introduction, how to negotiate a starting salary, when to follow up on a job application and how to phrase the follow-up — that Anthony Jack identified as the actual content of elite college life for students who didn’t grow up with it.

Nicholas Christakis’s network science formalizes what Jack observed ethnographically. Social networks propagate behaviors, opportunities, information, and trust along their edges. A network is not just a list of names. It is a structure through which valuable things move. An elite college is a machine for manufacturing thick, persistent, high-trust networks among people who will hold consequential jobs, and who will continue to hold them for fifty years. The instruction is the cover story. The four years of classroom hours, the lectures, the seminars, the office hours, the weekly problem sets — all of this is the public-facing content the institution charges money for, and all of it could in principle be reproduced (and is in fact reproduced) at hundreds of other institutions. The classroom is the deliverable. The network is the product. The institution is selling access to the network and packaging it as access to the classroom, because the classroom is the part the buyer feels comfortable being seen to pay for. The network is the part nobody mentions in the brochure.

Raj Chetty’s Opportunity Insights project has spent the last decade quantifying the network effect at the level of the American polity. The findings are difficult to absorb in one reading. Only 0.8 percent of U.S. college graduates attend the “Ivy-Plus” institutions (the eight Ivies plus MIT, Stanford, Duke, and the University of Chicago). Yet Ivy-Plus graduates hold:

Less than one percent of American college graduates produce more than four in ten presidencies and seven in ten Supreme Court justices. The ratio is not a curiosity. It is the entire point of the institution. The Ivy-Plus universities are a pipeline for converting families with capital into families whose children hold consequential positions, and the conversion is dramatically efficient.

Children with parents in the top one percent of the income distribution are 77 times more likely to attend an Ivy-Plus college than children with parents in the bottom 20 percent. Among children with comparable SAT and ACT scores — which is to say, children whose objective measured aptitude is the same — those from the top 1 percent are roughly twice as likely to attend an Ivy-Plus college as those from middle-class families. The sorting is not random. The sorting is not aptitude-based. The sorting is class-based, with aptitude as a partial filter, and the class that benefits most from the sorting is the class that already had access to the network the sorting purchases.

And then the closing finding, which is the one this article wants to leave the reader with. Less selective institutions like SUNY-Stony Brook have comparable or higher rates of moving low-income students into the top income quintile. 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 nearly four times the upward mobility rate that Harvard does — not because Stony Brook is a better school, but because Harvard does not admit poor kids in the first place. The Ivy-Plus mobility rate is low not because the Ivy-Plus universities fail to lift their poor students, but because the Ivy-Plus universities have very few poor students for the lifting machinery to work on. Harvard admits the children of capital, gives them access to the network they already had, and then publishes statistics about its alumni’s lifetime earnings that confirm what the admissions office’s parents already knew: that going to Harvard correlates with success, because Harvard correlates with the population whose children were going to be successful anyway.

Stony Brook admits poor kids and gives them access to the network they lacked. The mobility math is in Stony Brook’s favor because the mobility math measures the marginal effect of the institution on the population that arrived without the network. Harvard’s population arrived with the network. The marginal effect Harvard can have on its students’ mobility, on average, is therefore small. The institution is mostly not lifting its students. The institution is mostly cataloguing students who were already in the air.

X. The Steelman

The strongest case for the current system is real, and it deserves to be stated honestly before the article reaches its conclusion. There are intelligent and thoughtful people who defend the elite reproduction machine, and the better of their arguments are not silly.

The case runs roughly as follows.

First, peer effects are real. A generation of economics research has found that who your classmates are shapes what you learn, who you become, and how much you earn. Elite colleges concentrate high-ability peers in ways that generate genuine learning gains for everyone in the room. Caroline Hoxby’s NBER work on selectivity documents a real (if contested) return to attending more selective institutions. Her estimates suggest lifetime earnings differences of $300,000 to $400,000 between the most and least selective colleges for comparable students. That return has to come from somewhere, and the most parsimonious explanation is the combination of peers, instruction, and network access. If you abolish the elite institutions, you do not redistribute the peer effects. You destroy them.

Second, resource concentration enables quality. Harvard’s endowment was approximately $52 billion in 2024. That endowment allows it to hire faculty no median institution can match, fund research facilities no median institution can match, and offer financial aid packages that, for the small percentage of low-income applicants who do gain admission, make Harvard cheaper to attend than the average state university. The concentration of wealth at the top of the system is also the concentration of capacity. Dissolving the top into a thousand equal institutions would not elevate the thousand. It would lower the top. In this view, elite reproduction is the tax we pay for preserving institutions capable of genuine excellence.

Third, the alternative counterfactuals are worse. European universities select almost entirely on single-test performance — the French baccalauréat, the British A-levels, the Indian Joint Entrance Examination for IIT admission. These systems produce their own elite reproduction. Oxbridge admits are disproportionately from a small number of British private schools (the “Clarendon schools,” Eton, Harrow, and the rest). French grandes écoles admits are disproportionately from Parisian classes préparatoires that are themselves concentrated in the same affluent neighborhoods. Indian IIT admission via the JEE is now dominated by the Kota coaching ecosystem, where families pay for year-long residential test preparation. Pure meritocracy based on a single national examination does not eliminate elite reproduction. It just relocates it upstream into who can pay for test preparation. The American “holistic” system, for all its abuses, at least leaves room for Pell Grant students, first-generation college students, and the surprising-poor cases that a strict examination regime would filter out before the application stage.

Fourth, legacy preferences plausibly sustain donor networks that fund financial aid. The defenders argue that legacy-linked donations cross-subsidize first-generation and low-income students — that the rich families paying to install their children create the financial cushion that allows the poor students to attend at reduced cost. (The best empirical evidence does not support this claim, it should be noted: studies tracking alumni giving at the top 100 U.S. universities from 1998 to 2008 found no statistically significant relationship between legacy preferences and giving rates. But the defenders continue to offer the argument, and it has some surface plausibility.)

Fifth, Dale and Krueger’s “no-effect” finding has a critical exception. Their own 2014 follow-up found persistent and large effects of attending more selective colleges for Black, Hispanic, and low-SES students — exactly the groups most likely to benefit from network access they did not bring from home. If the Harvard premium disappears when you control for selection bias among white upper-middle-class applicants, but it reappears for poor kids and kids of color, that is itself an argument that elite institutions do something real. They provide social capital to students who lack it. The students who benefit most from elite admission are precisely the students the elite admissions process most resists admitting.

Sixth, the Anthony Jack finding is two-edged. Jack’s documentation of how elite colleges fail the doubly disadvantaged is also documentation that they are trying. Twenty years ago there were very few Pell Grant recipients at Harvard. There are now roughly one in five. The institution is imperfectly integrating populations it excluded entirely for three centuries. That should count for something, even against the ALDC numbers.

The steelman, honestly stated, is this. The elite universities do produce something real. The alternative systems are different flavors of the same poison. The preferences are ugly but the outcomes for the students who get through are better than anything on offer at comparable scale. The answer is reform — ending legacy preferences, expanding need-based aid, restructuring the athletic recruit pipeline, opening the admissions process to public scrutiny — rather than dissolution. This is the argument the article should engage, not dismiss.

And the article does engage it. It concedes that the peer effects are real, that the resource concentration enables some genuine excellence, that the alternative counterfactuals have their own pathologies, and that the Dale-Krueger exception for low-SES and minority students is a meaningful argument for the elite institutions’ counterfactual value. The reform program the steelman describes is a serious program. A reformed Ivy League — legacy preferences eliminated, athletic recruitment opened to genuine cross-class competition, the Dean’s List shut down, the admissions process subject to external statistical audit — would be better for the country than the present arrangement, and it is worth fighting for.

What the steelman cannot explain is the present arrangement. The steelman cannot explain why 83 percent of Harvard’s recruited athletes are white in a class that is 53 percent white. The steelman cannot explain the Dean’s Interest List, which has no plausible meritocratic justification and exists for the sole purpose of converting capital into admission. The steelman cannot explain why Kelley Williams-Bolar served nine days plus felonies for using her father’s address while Felicity Huffman served fourteen days for paying $15,000 to corrupt her daughter’s standardized test. The steelman cannot explain why the same Court that struck down race-conscious admissions for Asian-American plaintiffs left intact preferences six and seven times stronger that benefit white wealthy applicants. The steelman cannot explain why the J1/J2/J3 classifications were filed in 1922, why the 86 percent athletic recruit admit rate was filed in 2018, why the side door and the back door are priced differently but engineered identically, and why the apparatus that produces all of this is the same apparatus an honest reform program would have to dismantle from the foundation.

The steelman, in other words, describes what a reformed elite reproduction machine could be. It does not describe what the existing elite reproduction machine is. The gap between those two things is the part of the reality the steelman cannot reach, and the gap is large.

XI. The Reproduction Pipeline, Whole

Walk the pipeline backward from the present. The Chetty data is at one end: 0.8 percent of American college graduates produce 41.7 percent of presidencies and 71.4 percent of Supreme Court seats. Children of the top 1 percent are 77 times more likely than children of the bottom 20 percent to attend an Ivy-Plus institution. The sorting is not random. The sorting is the point.

Behind the Chetty data is the Arcidiacono data: legacy admits at 33.6 percent versus non-legacy at 5.9 percent. Recruited athletes at 86 percent. Dean’s Interest List at 42 percent. ALDC at 30 percent of the admitted class drawn from 5 percent of the applicant pool. Forty-three percent of white admits in the ALDC categories. Three-quarters of white ALDC admits unable to clear the bar without the preferences specifically designed to admit them.

Behind the Arcidiacono data is Daniel Golden’s reporting on the development cases, the Kushner $2.5 million paper trail, the Dean’s List as the formal mechanism by which donor children are converted into admits.

Behind the development cases is the Anthony Jack ethnography, which shows that even the “poor” students at Harvard are mostly the privileged poor — low-income kids who attended prep schools on scholarship and arrived already fluent in the hidden curriculum. The doubly disadvantaged remain under five percent of the student body.

Behind the privileged-poor pattern is the IEC industry, the test prep industry, and the prep school pipeline — the apparatus that converts family capital into the specific cultural fluency the elite admissions office is trained to recognize as merit.

Behind the IEC industry is the meritocratic opening of 1945 to 1975 — the Conant project, the SAT, the GI Bill, the Pell Grant — which was real and which was narrow and which began closing the moment the elite institutions and their adjacent industries figured out how to convert family capital back into the credentials the SAT had briefly forced them to acknowledge.

Behind the meritocratic opening is the 1922 Lowell apparatus, the New Plan of 1926, the J1/J2/J3 classifications, the personal essay, the photograph requirement, the “character” interview, the recommendation letter, the geographic distribution requirement, and the legacy preference invented in the same crisis for the same purpose. The vocabulary of contemporary American admissions was engineered between 1922 and 1926 by a Harvard president who wanted to exclude Jews, and the vocabulary is still in use, and it is still doing the same kind of political work, and the work is still mostly invisible to the people who pass through it.

Behind the Lowell apparatus is the colonial college sequence, the Latin examinations, the curriculum-as-filter that for two and a half centuries reproduced a hereditary Protestant gentry through a sorting mechanism so seamless that no admissions office was needed. The sorting happened upstream, in who got tutored in Latin as a child. The Latin tutoring was the original IEC industry.

Behind the colonial colleges is Harvard 1636, “dreading to leave an illiterate Ministry to the Churches.” The institution was clerical reproduction from the moment it was chartered. Three hundred and ninety years later, the institution is presidential and judicial reproduction at a similar concentration ratio. The product has changed. The function has not.

This is the pipeline. It is continuous. It has no break in its operation. The brief postwar opening was a partial slowdown, not a stoppage, and the slowdown was over within a generation. The instruction the institution charges money for is not what the institution is selling. The instruction is the cover story for the network, and the network is the product, and the product’s value depends on the buyer not seeing the mechanism by which the network is sorted.

The Sandel reading is the one the series will develop fully in its closing article. The corruption is not that money changed hands. The corruption is that the value of the credential depends on its purchase being invisible. The opacity is the credential. The 1922 letters had to be hidden for the apparatus to function. The Arcidiacono numbers had to be obtained through federal litigation against the institution’s active resistance. The Singer wiretaps had to be played in court before the public could see what its own elite institutions were doing in plain sight. Every layer of the pipeline depends on the preceding layer remaining out of focus, and every layer’s defenders mobilize to keep the focus where they need it.

The Christakis reading is that the network is not optional. Networks are how anything important moves through human societies. Information, opportunities, trust, the whisper that gets you the interview, the introduction that gets you the meeting, the marriage that gets you into the family that has held the position for three generations. None of this is corrupt in itself. Networks are part of how human beings live, and any society that pretends they are not is lying to itself. The corruption enters when the institution that is supposed to be sorting on merit is in fact sorting on prior network membership, and is calling the sort by the name of the cover story rather than by the name of the actual mechanism. The Ivy League is not corrupt because it has a network. The Ivy League is corrupt because it sells admission to the network as if it were admission to a classroom, and the price of the misrepresentation is paid by every student who arrives at the classroom thinking she has won a fair contest.

And the Luhmann reading, finally, is that the admissions code has no slot for what it is doing. The code processes admitted/rejected. It has no field for “your father donated $400,000” or “your mother is faculty” or “you have been playing squash with a private coach since you were eight, on a court your family belongs to.” The structural blindness is not a defect of the code. It is a feature. The code requires the blindness in order to function as a credential-distributing mechanism. If the code could see the upstream causes, it would have to acknowledge that it was distributing inheritance rather than sorting talent, and the institution would no longer be able to maintain the fiction that produces its prestige. The fiction is the prestige.

The series will close, in its seventh article, with a sustained development of the carpenter/house distinction that is implicit throughout this article. For now, the relevant move is to leave the reader with the closing finding, which is the one Chetty’s data forces and which the rest of the apparatus this article has documented makes legible.

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 nearly four times the upward mobility per student that Harvard does. Not because Stony Brook is a better school. Not because Stony Brook’s instruction is superior. Not because Stony Brook’s endowment is larger or its faculty more decorated. Stony Brook produces more upward mobility than Harvard because Harvard doesn’t admit poor kids in the first place.

Read that sentence again. Then read it once more.

The institution that markets itself as the engine of American meritocracy admits the children of the population that is already at the top, gives them access to the network they already had, and reports the resulting lifetime earnings as evidence of its educational excellence. The institution that does not market itself this way admits the children of the population that is not already at the top, gives them access to a network they did not have, and produces a measurable jump in lifetime earnings that is rounded down by the popular conversation about which schools matter. The popular conversation is wrong about which schools matter, and the popular conversation is wrong because the popular conversation has been trained, by the apparatus this article documents, to confuse selection for sorting and prestige for excellence and inheritance for merit.

The machine works as designed. The design is the scandal. The scandal is not in the side door. The scandal is in the entire architecture of the building, the front door and the back door and the side door and the network of people who walk through any of them, and the four-hundred-year continuity of the institution that has been doing this since before the United States existed. The scandal has been documented. The receipts are in the federal record. The Chetty numbers are public. The Arcidiacono numbers are in NBER working papers anyone can download. The Karabel book is in the library of every research university in the country. None of this is a secret. The work of this article has been to put the receipts in one place and to read them in sequence.

That is the entire architecture, and that is the entire pipeline, and that is the answer to what the credential is actually for. The next article will turn from the structural analysis of the machine to the moral analysis of what the machine does to the people inside it. Michael Sandel calls it the tyranny of merit, and the tyranny is what it feels like, from the inside, to be sorted by an apparatus that has been telling you since you were five years old that the sort was a measure of your worth.