Field notes from inside the cycle.

№ 001 · The Quad · Essay

On the over-application problem.

The typical Class of 2030 applicant sent 14.3 applications. The typical Class of 2020 applicant sent 7. The compounding consequences of that arithmetic are the entire story of modern elite admissions.

8-minute read · Updated May 28, 2026

← the Common App made this trivially easy. cost of marginal application: 30 minutes & $80.

A decade ago, the average elite-bound senior applied to seven schools. Today she applies to fourteen. The math of that doubling — across roughly 600,000 applicants chasing the top 50 — produces a system in which 9.2 million applications now compete for about 350,000 seats. The selectivity numbers everyone reads in the spring are downstream of this single fact.

It is tempting to attribute the change to anxiety. Anxiety is real and the cycle absolutely produces it, but anxiety is the symptom. The structural cause is simpler: the marginal cost of one more application has collapsed. The Common App, the Coalition App, and the test-optional shift have removed nearly all friction. What used to require typing the same data fifteen times now requires a checkbox.

What the colleges experience on the receiving end is harder to picture. An admissions office that received 25,000 applications a decade ago receives 60,000 today, with the same staff. The composition of the pile is also worse: more strong-but-not-exceptional applicants applying to the same schools, more "throwing it at the wall" choices from students for whom the school is a stretch, more impossible-to-distinguish-from-each-other essays that AI assistance has flattened. The signal-to-noise ratio of any given file has gone down.

◆ ◆ ◆

Yield management is what the colleges do about this. A school like Brown, knowing it will enroll roughly 1,700 students from its admit pool, has to model how many students it can admit to land at that number. The yield rate — what percentage of admitted students enroll — is the entire mechanism. If yield is 60%, Brown admits about 2,800. If yield drops to 55%, it has to admit 3,100. The school's entire risk management is "predict yield within 2%."

Now imagine what 14.3 applications per applicant does to that prediction. The pool of students cross-admitted with Brown's peers has grown enormously. A student admitted to Brown is also, statistically, admitted to four to seven other schools of comparable selectivity. Brown cannot tell which of those students will enroll until they decide on May 1. So Brown does two things: it builds a waitlist three times larger than the admit class as insurance, and it tightens its acceptance criteria for "demonstrated interest."

★ the unspoken yield model: "where else did you apply?"

This is why your decision letter from Tufts arrived alongside a personal email from your regional AO and why Brown's deans court committed students aggressively at admit weekends and why everyone waitlists you. None of it is irrational from the colleges' standpoint. It's the only way to manage a system whose inputs have doubled.

The over-application problem is bad for everyone in it. It's bad for students, because the cycle becomes a high-noise lottery rather than a discernment process. It's bad for parents, because the cost of admissions consulting has gone up to chase the noise. It's bad for counselors, who now manage cohorts twice the size with the same time. It's bad for colleges, because their admit pools are riskier and their yield modeling more brittle. It's especially bad for the applicants at the edges — first-generation students, applicants from less-represented regions, anyone for whom the volume looks unwinnable. They apply to fewer schools. They lose.

What changes the system is information. The reason the cycle escalated is that no individual actor — student, parent, counselor — has any visibility into what their peers are doing or what's actually working. They apply to fourteen schools because everyone applies to fourteen schools, because the dominant strategy is to over-apply. A real picture of the cross-admit landscape — who got in where, with what profile, against what list — changes that calculation. If a student could see that the median applicant with her stats had three admits out of fourteen attempts, she might apply to seven schools with more care.

That is what this publication is for. Not to fix the system, which requires the colleges themselves to act, but to publish the picture honestly enough that the participants can stop guessing.

№ 002 · The Quad · Reporting

The yield trap: how Tufts waitlists 26,000.

Inside the operational reality of yield-protect admissions. What the deans are managing when they decide who waits, and why the waitlist exists in the shape it does.

6-minute read · Reporting · May 2026

"yield protect" — a phrase the colleges hate but absolutely use.

The number sounds like a typo. Tufts University, with an entering class of around 1,800, places roughly 26,000 applicants on its waitlist each spring. That is more than fourteen times its target. By any normal logic of waitlists as a backup mechanism — *we'll go back to you if we need to* — the math is absurd. The school will, in a typical year, admit fewer than 100 students off the waitlist. The other 25,900 receive a soft no dressed up as a maybe.

It looks insane from the outside. It is rational from the inside. Understanding why requires understanding what yield protection actually is.

Tufts, like Northwestern and Vanderbilt and NYU and a handful of other "yield-protect" schools, sits in a difficult demographic position. It is selective enough to attract students who treat it as a safety relative to the Ivies — strong applicants with HYPS-level stats who apply to Tufts in case nothing else works out. The yield rate on those admits is brutal: a student admitted to Tufts who also gets into Yale will enroll at Yale roughly nineteen times out of twenty. Tufts admits those students and they don't come.

◆ ◆ ◆

So Tufts has to do something. The most defensible thing it can do — and the thing every yield-protect school does — is to identify the applicants who are likely using it as a safety and to send them to the waitlist instead of admitting them outright. The signal an AO looks for is constellation: very high stats, full slate of HYPS-tier reaches, no demonstrated interest in Tufts specifically, no Tufts-fit narrative in the essay. The student probably will not enroll. Waitlisting them protects yield without rejecting them outright.

This is the yield trap in microcosm. The strongest applicants, applying to the strongest schools, get waitlisted at the schools they assume they'll easily get into. "Why did I get rejected from Tufts but admitted to Brown?" becomes the most-asked question on every counselor's spring forum.

The 26,000 number scales from this. Tufts cannot identify each yield-risk applicant perfectly. Many strong applicants would have enrolled if admitted. So the school casts a wide waitlist — wider than its actual capacity to admit off it — so that if its yield model misfires (more committed admits decline than expected), it has a deep pool to draw from. The waitlist is operational insurance against yield model error.

↓ this is why counselors say "show real interest at your reaches"

The signal: students who demonstrate clear, specific interest in a yield-protect school — visiting, attending sessions, writing a thoughtful "why us" essay, attending the regional info session — get admitted at meaningfully higher rates than equivalent applicants who don't. Demonstrated interest moves the needle more at Tufts than at Brown by a factor of three to four in our reading of the cycle. The colleges deny that this is so. The admissions data, where we can see it, says otherwise.

What follows from this for an applicant is not complicated. If you are applying to a yield-protect school as a safety, treat it like a target. Visit. Write the essay as if you mean it. Attend the session. Apply ED if you actually would attend (this is the cleanest signal of all). The school does not need you to swear loyalty; it needs you to give it any reason to believe you will enroll.

The 26,000 are the people who didn't.

№ 003 · The Quad · Methodology

A note on anonymity.

Where the line is, why it's there, and what we will not publish. The promise that makes the dataset possible — and that protects the seventeen-year-olds who fill it.

5-minute read · Editorial · May 2026

k-anonymity is the technical term. matters enormously here.

Every dataset built on personal data has the same temptation: to publish the granular details that make individual stories interesting, at the cost of the privacy of the people in it. Most consumer-data products eventually give in to this temptation. We have built The Record specifically to resist it.

Here is what we collect: bucketed academic statistics (GPA in a range, SAT in a range), broad demographic categories (region, high school type, broad intent), the list of schools an applicant applied to, the outcome at each, and optional qualitative reflections. We do not collect names, exact birthdates, exact stats, school district identifiers, or anything else that could re-identify a specific student.

The deeper protection is statistical. K-anonymity is a property of a dataset that says: any combination of attributes that returns fewer than K records will not be displayed. We use K=5 as our minimum. If you would be the only applicant in our dataset with your stats, your region, your high school type, and your school list — we don't render that combination. Anywhere. Not in aggregates, not in cross-admit charts, not in the personal dashboard view.

◆ ◆ ◆

This is conservative and it costs us interesting visualizations. The trade-off is correct. A seventeen-year-old who submits her cycle to us in May 2026 deserves to know that a journalist or admissions officer or her own high school cannot reverse-engineer her identity from anything we publish — not now, not five years from now.

The Record is the institutional memory of a cycle. It belongs to the people who lived it. It is not for sale.

We will never sell the dataset. We will never license it to colleges in a form that includes individual records. The aggregate intelligence product we may eventually offer to enrollment management offices will be exactly that — aggregates, k-anonymized, with no path to individual records, in a separate legal entity and with a separate data architecture from the applicant-facing product. This is not a promise made lightly; it's the architectural decision that defines the publication.

If you ever want your record removed, you can. Use your recovery code at any time to delete your submission. The corresponding aggregate cells will recompute within the hour. We are obligated to honor takedown requests within 48 hours, no questions asked.

The reason any of this works as journalism is that the people contributing to it trust us. That trust is the entire asset. Every editorial decision flows from defending it.