A response to the 2026 school choice moment: engaging the strongest pro-choice arguments, conceding what deserves concession, and asking why a reform this large is being run without instrumentation.
The short version: American school choice is being transformed from small, means-tested programs into universal entitlements at historic speed. The favorable research comes almost entirely from the small, targeted programs; the money is flowing to universal ones that resemble them in name only, and early data shows the subsidy moving up the income ladder. Every country that makes publicly funded school choice work does so with the regulation the American movement politically opposes. The public system provides goods — universal acceptance, integration, rural provision, the education of everyone else’s children — that appear in no test-score comparison and cannot be cheaply rebuilt once dismantled. The child-first position is neither side’s: choice where it is targeted, accountable, and measured, and serious investment in the universal system everyone else’s children will still attend.
Author’s note: I was educated in the Irish public system, including a DEIS-designated primary school — the kind of compensatory public program discussed below.
American public education, an institution built over more than a century and a half of legislation, litigation, and local practice, is being structurally overhauled in about half a decade. The Heritage Foundation, celebrating rather than cautioning, describes the K–12 system as being in the early stages of a dramatic transformation: participation in private choice programs up more than tenfold per capita over 25 years, 18 states adopting universal eligibility since 2020, and more than half of American students now eligible for a private choice program (Heritage, 2026). Arizona’s ESA program grew from roughly 12,000 students in 2022 to nearly 99,500 by January 2026 after going universal (RAND, via Queen Creek Tribune). Texas launches a $1 billion program this fall, expected to be the largest first-year choice program in the nation’s history.
Institutions of this age and complexity are Chesterton’s fences. Before removing a fence whose purpose you don’t understand, Chesterton advised, go and find out why it was put there. What strikes me about the current wave isn’t that it wants to move the fence. There are defensible reasons to. It’s that the fence is coming down at speed, with almost no instrumentation to detect what it was holding back.
New to this debate? A ninety-second primer. American K–12 education is run by roughly 13,000 local school districts, funded substantially by local property taxes, with children assigned to schools by home address. “School choice” is the umbrella term for policies that let public money follow a child somewhere else. The three main instruments matter to everything that follows. Charter schools are publicly funded but independently run, free to attend, and (when oversubscribed) required to admit by lottery. Vouchers give families public money toward private school tuition. Education savings accounts (ESAs) go further, depositing state funds into a parent-controlled account spendable on private tuition, tutoring, homeschool curricula, and other approved expenses. For thirty years these programs were mostly small and restricted to low-income families or students with disabilities. Since 2020, a wave of states has made them universal — open to every family regardless of income, including those already in private school. That shift from targeted to universal is the hinge of everything I argue below. For a fuller neutral overview of the policy landscape, see Ballotpedia’s explainer.
The speed is worth dating, because the movement’s history divides into three quite different phases. The idea is about seventy years old; Milton Friedman proposed vouchers in 1955. The policy experiment is about thirty-five. Milwaukee’s small, means-tested voucher program launched in 1990, Minnesota passed the first charter law in 1991, and for three decades “choice” meant modest, targeted, city- and state-level programs. Those programs generated virtually the entire research base. The transformation is barely five years old: eighteen states adopted universal eligibility after 2020, Arizona went universal in 2022, and nearly all of the per-capita participation growth of the past quarter-century has happened since the pandemic. Seventy years of ideas, thirty-five years of small experiments, five years of restructuring. The evidence comes from the middle phase. The money is flowing to the last one. Whatever this is, it isn’t conservatism in any Burkean sense. It’s a rapid, ideologically confident experiment on 50 million children, resting on an evidence base far thinner and less transferable than its advocates claim.
Any honest treatment has to start by granting the choice movement its nugget of truth, and I want to grant it in full.
The status quo genuinely fails poor children. In much of urban and rural America, the neighborhood public school predicts the absence of opportunity rather than guaranteeing it. No reasonable defense of “the public system” can pretend otherwise. In Ireland or Finland, no reasonable family anywhere in the class structure views public education as a cap on their child’s future. Large parts of the United States cannot say that. That failure, not Milton Friedman, is the engine of the movement’s moral energy.
The wealthy already have school choice. They exercise it through the housing market. Because US schools are funded substantially by local property taxes and assigned by catchment, buying a home in a good district is a voucher, available only to those who can afford the house. Choice advocates are right that the status quo is itself a choice system rigged in favor of homeowners.
Some choice programs demonstrably reach the families they claim to. EdChoice’s analysis of Pennsylvania’s tax-credit scholarships is the strongest recent version of this argument and deserves engagement rather than dismissal. Pennsylvania’s EITC program is income-targeted; recipient household incomes run roughly $56,000–$78,000 against a state median near $100,000, and Philadelphia scholarship families average about $63,000, in line with the county median (EdChoice / Kristof, June 2026). Around 101,000 scholarships were awarded in 2023–24 while roughly 69,000 additional applicants were turned away for lack of funds. That waitlist is revealed-preference evidence of demand from working-class families. It can’t be waved away as astroturf.
The best charter evidence is strong. Lottery-based studies of urban charter schools in Boston, New York, and New Orleans, serving overwhelmingly poor, Black, and Hispanic students, show some of the largest achievement effects in modern education research. Where oversubscribed, non-selective, accountable charters operate in poor neighborhoods, children measurably benefit.
Parents value things tests don’t measure. EdChoice is correct that safety, values, discipline, and fit matter to families, and that parental satisfaction in choice programs is consistently high (EdChoice, “What the Research Really Says,” 2026). A framework that treats test scores as the only outcome is impoverished.
And one concession about causation that critics of choice routinely dodge. The education depression Elorza describes (NAEP declines, widening gaps, absenteeism) largely predates the ESA wave and runs independent of it. Scores began sliding around 2013, well before universal programs existed, and the pandemic collapse happened inside a system that was still overwhelmingly traditional public schooling. Blaming school choice for the achievement crisis would be the same causal sloppiness I’m criticizing in the other direction. The public system’s decline is real, and mostly self-inflicted or externally driven, which is why the response to choice cannot be a defense of the status quo.
These concessions define the problem any serious alternative must solve: the American funding-by-postcode model has produced a public system that poor families are rational to want to escape.
EdChoice cites nearly 200 empirical studies across eight outcomes, and notes that 26 of 29 studies on competitive effects found benefits for students remaining in public schools. Taken at face value this sounds overwhelming. But that literature is built almost entirely on small, means-tested, oversubscribed programs: a few thousand students, low-income eligibility, private schools opting in at the margin. Those are the conditions universal ESAs abolish.
Worse, the scale relationship in the data runs the wrong way. As the Brookings Institution summarized before the ESA wave crested, the larger the voucher program, the worse the results tend to be, possibly because the supply of effective private schools is limited, possibly because vouchers let private schools choose their students as much as parents choose schools (Brookings, 2024). The largest statewide voucher evaluations on record (Louisiana, Indiana, Ohio, D.C.) found null or significantly negative test-score effects, among the worst results anywhere in the education literature. The programs most like what states are now building universally have the worst track record. Extrapolating from a 2,000-student pilot in Milwaukee to a universal statewide entitlement is not a conservative inference.
Brookings also makes a comparison the movement rarely invites: if the goal is improving public school outcomes, the evidence for simply funding public schools directly is far more plentiful than the evidence for competitive spillovers.
Choice advocates do have a serious answer to the Louisiana results, and I think it deserves a fair hearing. Louisiana’s program was unusually heavily regulated, with mandatory state testing, open admissions, and price controls, and researchers found that higher-quality private schools largely declined to participate. The program was left with financially desperate, under-enrolled schools. On this reading the regulation, not the choice, produced the bad results. It’s the movement’s best empirical argument, and it changed how I talk about Louisiana.
But look at what the argument actually establishes: selective regulation in a two-tier market drives adverse selection. When good private schools can prosper outside the program, charging full tuition to families who don’t need vouchers, regulation becomes a tax that only weak schools will pay. That’s an argument against half-measures, not against oversight. The Netherlands avoids the trap by making regulated public funding the only game in town; virtually all Dutch private schools take the state’s money and the state’s rules, because there is no lucrative unregulated tier to retreat to. Drawn properly, the Louisiana lesson is that a regulated choice system must be universal in its regulation, with uniform funding, no top-ups, and common standards for every funded school, or the quality supply will sort itself out of reach. The US movement has drawn the opposite lesson and deregulated everything, which “solves” the adverse-selection problem by abolishing the ability to detect quality at all. That isn’t a fix. It’s turning off the smoke alarm because it kept going off.
The first serious independent evaluation of a universal program, RAND’s 2025 case study of Arizona (funded, notably, by the pro-choice Walton Family Foundation), found:
Consider what this does to the Pennsylvania argument. The features EdChoice celebrates there — income targeting, geographic targeting to low-achieving schools, modest scholarship amounts that require family skin in the game — are the very features the universal ESA movement is eliminating everywhere else. Pennsylvania is being used as evidence for a model it does not resemble.
I’d call this the laundering problem: the movement’s evidence and the movement’s money serve different populations. The favorable research comes from means-tested programs, lottery-admitted charters, and working-class waitlists. The appropriations flow to universal entitlements whose users, in Arizona, were 74% ineligible under the rules that generated that research. No one has to impute intent here; the revealed preferences speak for themselves. A movement confident that its case rested on helping poor children would fight to extend the design features that produced its evidence — income caps, lottery admissions, participation testing, independent evaluation. Instead those features are being stripped out as the programs scale, and proposals to restore them are treated as attacks. The poor supply the studies and the sympathetic anecdotes, the subsidy travels up the income distribution, and the data that would let anyone check whether the bait matches the switch is the one thing legislatures reliably refuse to fund.
Stephen Moore’s syndicated column states the movement’s core premise with admirable bluntness: when stores compete, customers win, and this is “an inviolable law of economics” that applies equally to schools (Moore, June 2026). This premise carries the entire policy, and it is a declaration rather than a finding.
Education differs from retail in ways economists have documented for decades:
Moore’s own column illustrates the evidentiary standard in play. Florida choice students are described as outperforming public peers, with no mention that non-random selection into these programs makes raw comparisons close to meaningless. And his framing of routine Democratic budget decisions as a “war on school choice” sits awkwardly beside his own admission that even under robust choice, two of three kids will remain in public schools — the system his agenda would drain.
There’s no need to speculate about motives; the Heritage Foundation footnotes them. Its 2026 Index attributes the choice surge partly to awareness of classroom “politicization” and gender-related grievances, and its supporting citations include a 2022 Backgrounder literally titled “Time for the School Choice Movement to Embrace the Culture War” (Heritage, 2026). When a movement’s flagship think tank frames the policy as a culture-war instrument, the burden shifts. Student outcomes become the public-facing justification; the restructuring is the point.
The most reliable behavioral test is the response to accountability. If the goal were student outcomes, advocates would demand the testing, reporting, and evaluation that could vindicate their case. Instead, Arizona’s legislature has rejected guardrails three years running, and most universal programs require no comparable outcome data at all. You don’t refuse the scoreboard when you believe you’re winning. You refuse it when the score is not the point.
The strongest version of the pro-choice argument comes not from the right but from Jorge Elorza of Democrats for Education Reform, writing in the New York Times (March 7, 2026). I want to treat his case separately, because a fair-minded reader might find it persuasive, and because on close reading it concedes most of what I’m arguing here.
His diagnosis should be granted in full: proficiency stagnant at roughly a third of students, widening gaps, surging absenteeism, teacher attrition, collapsing trust, and a funding architecture tied to ZIP codes that insulates schools from their own results. He is also right that incremental reform-from-within has repeatedly stalled. This is the education depression that makes the choice movement possible, and defenders of the status quo have no adequate answer to it.
Four features of his argument undermine the policy it’s deployed to support, though.
First, his own evidentiary standard indicts the programs actually being built. Elorza’s scholarly anchor is Ashley Rogers Berner’s research on educational pluralism, which finds that students thrive when countries fund many school types while holding all of them to common academic standards. He stipulates that government must set guardrails on access and standards, and that where oversight is lacking, quality suffers. That is the Dutch model, and it is what Arizona, Texas, and the universal ESA wave have declined to build. Arizona has rejected quality metrics and outcome measurement for three consecutive sessions. The op-ed argues for regulated pluralism; the argument is being used as air cover for unregulated subsidy. By his own cited research, the flagship American programs should be expected to fail.
Second, the iPhone analogy proves the opposite of what it’s meant to. Elorza credits the App Store’s explosion of bottom-up innovation to openness. But the App Store is one of the most tightly governed marketplaces ever created. Every app passes centralized review, follows detailed rules, and can be removed at any time. Apple’s insight was that a platform generates trust and quality because of strict curation, not despite it. The education equivalent of the App Store is the Dutch inspectorate, not an ESA marketplace where any vendor can collect public money without review. And the deeper disanalogy remains: a bad app costs 99 cents and ten seconds to delete, while a bad school costs a child years that cannot be refunded, discovered only after the damage is done.
Third, the Pell Grant analogy carries a warning label he omits. Portable public funding plus lightly regulated private providers is the combination that produced the for-profit college scandal, in which predatory institutions harvested federal aid while handing poor students worthless credentials, until the government was forced into gainful-employment rules and mass loan forgiveness. Arizona’s post-expansion tuition inflation suggests the same subsidy-capture dynamic is already operating in K–12. The lesson of Pell is not that money following students works. It’s that money following students works only with aggressive quality regulation, a lesson learned at enormous human cost.
Fourth, “public education is a goal, not a fixed set of institutions” is elegant, and it elides what the institutions do. The formulation invites indifference about who runs schools so long as children thrive. But some functions inhere in the institutional form itself: the legal duty to accept every child, the civic commons where a society’s classes and creeds actually meet, guaranteed provision in rural counties no market will serve, and democratic governance of what the public pays for. A goal cannot enroll a disabled child the market has priced out. Arizona’s falling disability share is what goal-not-institutions looks like in practice.
One more thing in Elorza’s piece is worth naming. He celebrates the “interregnum,” the crisis window in which unthinkable reforms become possible, as an opportunity to rebuild from the ground up. That’s an honest description of the politics, and it’s the condition under which Chesterton’s warning applies with maximum force: crisis windows are when fences get torn out fastest and understood least. If Elorza’s guardrailed pluralism were what states were building, I’d be writing largely an endorsement. It isn’t, and the gap between the op-ed’s vision and the statute books is the whole dispute.
Choice advocates are correct that publicly funded private schooling can work at national scale. The Netherlands has run it for a century, and Belgium and Ireland fund non-state schools extensively. But every functioning example pairs choice with the architecture American universal programs omit: uniform per-pupil public funding, bans on tuition top-ups, regulated admissions to prevent cream-skimming, national curricula and examinations, and state inspection. Dutch “private” schools are, in economic and accountability terms, more regulated than many American public schools.
The two large natural experiments in lightly regulated universal choice are cautionary tales. Chile’s nationwide voucher system produced severe socioeconomic stratification with little aggregate achievement gain, and was substantially re-regulated in 2015, with copayments and selective admissions banned after decades of documented sorting. Sweden’s for-profit free-school reform coincided with systematic grade inflation — grades rose while measured knowledge fell, since Swedish teachers grade with full autonomy — and one of the sharpest PISA declines in the OECD, prompting rollback debates that continue today. (Fairness note: part of Sweden’s decline is compositional, with one study attributing about 29% of the 2000–2012 fall to changing student demographics, and the OECD also pointed at teacher training. The grade inflation and increased segregation, though, are directly tied to the competitive design.) The international evidence does not say choice fails. It says choice without regulation stratifies, and the American movement is adopting the choice while organizing politically against the regulation.
Ireland offers the inverse lesson. Its fee-paying schools, concentrated in the affluent east and south, purchase networks and prestige, and their own alumni are typically the first to say so. Nobody pretends they escape a failing baseline, because the baseline isn’t failing. Ireland achieves broad attainment through centralized per-pupil funding and explicit compensatory targeting: the DEIS program directs smaller classes, extra funding, home-school liaison, and literacy supports to designated disadvantaged schools, inside the public system, no market mechanism required.
Intellectual honesty, by the same standard I’ve applied to voucher research above, requires noting that the causal evidence on DEIS itself is contested — and that it splits by age in an instructive way. At primary level, the independent Educational Research Centre’s repeated assessments of urban DEIS pupils (2007, 2010, 2013, 2016) found significant improvements in reading and maths across all grade levels (ERC), though national assessments cautioned that in some grades the gains merely kept pace with national improvement and performance remained below national norms. At post-primary, the ERC reports a continued narrowing of achievement and retention gaps between DEIS and non-DEIS schools (ERC, Jan 2026), and Junior Cycle retention in DEIS schools has risen dramatically over the program’s life. But a 2025 regression-discontinuity study, comparing schools just above and below the eligibility cutoff (a far stronger design than trend comparisons), found no detectable impact of DEIS designation on post-primary examination performance, retention, or the shares of low and high achievers, despite widespread belief in its effectiveness (Evaluation and Program Planning, 2025). Gap-narrowing trends cannot separate program effects from Ireland’s economic recovery, preschool expansion, and demographic change — the same confound problem that afflicts raw voucher comparisons. If we dismiss selection-confounded evidence when EdChoice offers it, we can’t embrace it when the ERC does. The clearer signal at primary and the contested null at post-primary do fit the broader economics of education, though: returns to intervention are highest early and fall with age, and by fifteen a decade of accumulated disadvantage is hard for school-level supports to reverse. If DEIS teaches a design lesson, it is to front-load the investment.
Two things follow, and both strengthen the argument rather than weakening it. First, DEIS models the process the American reform wave refuses: two decades of continuous, independent, published evaluation of a public program, which is how a null result can even be discovered and debated. Arizona cannot have this argument, because it declined to collect the data. Second, if DEIS’s specific resource package matters less than believed, then Ireland’s equity comes from the system-level architecture itself: central per-pupil funding, a professionalized teaching corps, socioeconomically mixed schools, and no respectable mass exit route for the middle class. The fence, in other words, not any single ornament hung on it. Either way, the result is a society in which no reasonable family experiences public education as a ceiling, which is the condition that makes exit unnecessary and keeps the middle class invested in the shared system.
The deepest risk is not that ESAs will fail on test scores. It’s that the observed benefits of choice may be artifacts of the mixed system that choice is now dismantling.
A century-and-a-half-old institution is being overhauled in five years, and the overhaul is designed so that its effects cannot be measured until they cannot be reversed. That is the opposite of the epistemic caution the reform’s own political tradition claims to honor.
I want to add one more argument, addressed directly to the parent whose honest, decent motive is my child’s welfare — the parent watching classroom behavior deteriorate and wondering why their kid should pay the price. That motive deserves respect, not lectures. Behavior really has worsened: 64 percent of teachers this year say classroom behavior got worse over the past twelve months (EdWeek/The Hill, March 2026), elementary teachers nationwide report escalating and increasingly severe problems even among the youngest children (Hechinger Report, Feb 2026), and children whose early years were disrupted by the pandemic show measurably more emotional problems. Wanting out of that is not selfishness. It’s parenting.
But follow the mechanics of exit one step further than the school gate. Difficult children do not disappear when other families leave; they concentrate. Private schools, and in practice many choice options, can expel or quietly counsel out. Public schools legally cannot. So each round of behavior-driven exit distills the remaining schools into a higher concentration of the hardest-to-serve children, taught by the least experienced teachers (turnover is worst where behavior is worst), funded by a budget shrinking faster than costs, and stripped of the well-regulated peers whose presence is one of the best-documented moderating influences on behavior. The concentration effect itself is well established: Irish researchers found a substantial “social context effect,” in which individual student achievement falls as the school’s concentration of disadvantage rises, over and above the effect of the student’s own background (Sofroniou, Archer & Weir, replicated in the ERC’s DEIS evaluations). Sorting doesn’t just move problems around; it multiplies them where they land. In one American school’s discipline data this year, close to half of all referrals came from the same twenty students (EdWeek, March 2026). A sorting system takes that concentration and compounds it, school by school, year by year, until the residual institution is exactly the “terrible school for difficult children” that the next wave of families flees in turn.
And here is where the fence reveals its original purpose. Universal education was never sold to America as a service for your child alone. Horace Mann pitched the common school to wealthy taxpayers in the 1840s as insurance, not charity: the children you decline to educate alongside your own do not vanish, they grow up. They share your child’s labor market, neighborhood, jury pool, and ballot. The externality of abandoning them is not avoided by exit; it is deferred and compounded, collected a generation later through crime, dependency, and social distrust, at a price no comparison of school fees will ever capture. A system that lets each family individually escape difficult children is a textbook collective-action failure — individually rational, collectively ruinous — and the families who exit first still have to live in the society the residual system produces. You can move your child away from those classmates. You cannot move your child away from the adults they will become.
One caution keeps this argument honest, and I want to state it clearly. It cannot be used to conscript poor children as peer-effect providers in chaotic classrooms; that is the ugly version, and choice advocates are right to attack it. The alternative to sorting is not tolerating disorder. It is resourcing behavior seriously inside the universal system: smaller classes (teachers’ own first-choice remedy in that same survey), behavioral specialists and counselors at scale, and structured alternative settings that stay within public accountability rather than expulsion into nowhere. Countries with orderly classrooms and strong outcomes do not achieve them by sorting the difficult children out of sight. They achieve them by treating every child’s behavior as the system’s problem to solve, because every child’s future is the society’s future to live in.
The vendor ecosystem invites grift. ESAs fund not just tuition but a sprawling marketplace of curricula, tutors, and “educational services.” In Arizona only about 60% of ESA funds go to tuition, and 28% of awarded funds in 2023–24 went unspent, rolling over indefinitely (EdWeek). Unvetted vendors plus minimal spending oversight is the for-profit college pattern in miniature, and documented cases of dubious ESA purchases in Arizona suggest it’s already running. Any funded marketplace needs vendor accreditation and spending audits. More App Store, less flea market.
The religious liberty trap cuts against the movement’s own culture warriors. Under current Supreme Court doctrine (Carson v. Makin, 2022), a state that funds private education generally cannot exclude religious schools — any religious schools. Texas discovered this when its initial attempt to keep Islamic schools out of its program drew immediate legal challenge (EdWeek). States adopting universal choice are committing, constitutionally, to fund worldviews their own majorities may find objectionable. Few culture-war advocates on any side have priced this in, and it’s one more reason the “guardrails later” strategy is a fantasy: the guardrails available later are narrower than the ones available at design time.
Rural conservatives were the original skeptics, and their reasons are worth hearing. Texas vouchers were blocked for two decades not by unions but by rural Republicans, whose districts are often the county’s largest employer and social anchor, and where no private school market exists or ever will. Their objection was Chestertonian avant la lettre: they could see what the fence was doing. The movement finally overcame them through primary challenges rather than persuasion, offering rural families a subsidy they largely cannot spend, which says something about whose interests universal choice serves. Any honest design would include a rural test: what does this program do for a child in a county with one school and no market?
If the goal is the best sustainable outcome for children rather than an ideological win for either coalition, I think the international and domestic evidence converges on a program neither American tribe currently champions:
I want to say something plainly about what has happened to this debate. Education, a domain that for most of the twentieth century commanded broad bipartisan investment, has been turned into a front in the culture war, and the escalation has been asymmetric.
The larger share of the politicization has come from the right, and that isn’t a partisan characterization; it’s the movement’s self-description. Heritage’s flagship strategy document is titled “Time for the School Choice Movement to Embrace the Culture War,” and its 2026 Index credits the choice surge to classroom grievance politics as much as to any educational outcome. Stephen Moore’s syndicated columns frame routine budget decisions as Democrats “declaring war” and casually indict an entire profession as union “masters” pulling puppet strings (Moore, June 2026). The comment sections beneath those columns, where public schools are described flatly as socialist indoctrination camps, are the base being cultivated. When the argument for restructuring how 50 million children are educated leads with enemies rather than evidence, the evidence was never load-bearing.
Honesty requires the second clause. Teachers’ unions have contributed their share, at real cost to their credibility. Union-aligned opposition has fought high-performing, non-selective urban charters, the one choice instrument with unambiguous evidence of helping poor children, with the same intensity it fights unregulated vouchers — a failure of discrimination that mirrors the movement it opposes. Prolonged pandemic closures in union-strong districts, defended past the point the epidemiology supported, did more to collapse public trust and fuel the exit wave than any Heritage white paper. And blanket resistance to accountability gives choice advocates their most honest talking point. The asymmetry is real, but it’s an asymmetry, not an innocence.
The casualty of this pincer is the child-first middle, and if you’ve read this far with some sympathy for school choice, you’re probably standing in it. So here is the caution, offered in good faith: your sympathy is the product being harvested. The moral case that moves you — the kid trapped in a failing school, the working-class waitlist in Philadelphia, the charter lottery that changed a life — is real. And it’s being systematically attached to policies that don’t serve it: universal subsidies flowing three-quarters to the already-departed, programs stripped of the targeting and accountability that generated the evidence, a fiscal architecture that defunds the schools where two-thirds of children, and nearly all of the poorest, will remain. One test separates an ally from an agenda, and it’s the test I’ve applied throughout this piece: an ally of the child accepts the scoreboard. Support choice where it’s targeted, accountable, and measured. Withhold your sympathy from anyone, on either side, who refuses to be measured, because whatever they’re optimizing for, it isn’t the thing they told you.
The choice movement’s nugget of truth is real. American public education, funded by postcode, fails poor children in a way that peer nations’ systems do not, and the families on Pennsylvania’s 69,000-person waitlist are not ideologues; they’re parents responding rationally to that failure. But the policy being built on that truth is not the one the evidence supports. The research comes from small, targeted, accountable programs while the money flows to universal, unregulated, unmeasured ones. The countries that make choice work do so with the regulatory architecture the American movement treats as apostasy. And the institution being dismantled provides civic goods — universal acceptance, integration, rural infrastructure, a shared commons — that will register in no dataset until they are gone.
Chesterton’s point was never that fences must stand forever. It was that the reformer who cannot explain the fence’s purpose has not earned the right to remove it. The current wave can explain the fence’s failures fluently; it shows remarkably little curiosity about its functions. Children deserve reformers who want to know both, and who are willing to be measured.
Sources engaged: EdChoice (“What the Research Really Says,” 2026; “Pennsylvania Shatters National Narratives,” June 2026); Heritage Foundation, 2026 Index of Culture and Opportunity; Stephen Moore, syndicated column, June 2026; RAND Corporation, Case Study of Arizona’s K–12 ESA Program (Nov 2025), via press coverage; Brookings Institution (2024); Education Week (Jan 2026); Houston Public Media / REACH (Jan 2026); Save Our Schools Arizona (Jan 2026); Jorge Elorza, “We Need a New Educational Operating System,” New York Times (March 7, 2026); Center for New York City Affairs, “The Paradox of Choice”; Educational Research Centre (Ireland), DEIS evaluation series (2007–2026); “Reducing educational disadvantage across post-primary schools,” Evaluation and Program Planning (2025); Education Week State of Teaching survey coverage (March 2026); Hechinger Report (Feb 2026).