The cracks in American education don’t start in the classroom—they begin at the property line. Math scores have plummeted in a staggering 83% of school districts since 2009 and reading scores are down 70%. Smartphones and pandemic-era lockdowns have contributed to the decline, but well before the iPhone our modern education became trapped inside a geographically segregated housing market. Wealthy families are systematically buying their way into elite, walled-off school districts, leaving middle and lower-income students stranded in underfunded classrooms.

Welcome to The Dividing Line from American Inequality.

These visuals are all interactive and available with more information below.

See the visual story

Education is not considered a ‘fundamental right’ in America

Before the mid-1900s, property tax was mostly a relatively flat tax and public school funding was highly localized, small-scale, and supplemented by various state funds. But after World War II, the GI Bill and massive federal highway investments triggered a suburban boom, fueled by redlining and the systematic exclusion of minority families. Wealthy White homeowners concentrated their growing home equity into self-governing suburbs, creating tax-revenue goldmines for their local schools while hollowing out urban and rural budgets.

By the late 1960s, parents and civil rights lawyers realized that this funding system was wildly unfair. In San Antonio, Texas, parents in the low-income, predominantly Mexican-American Edgewood school district pointed out a stark reality: even though they taxed themselves at a much higher rate than their wealthy neighbors in Alamo Heights, they could only raise $37 per student, while Alamo Heights raised $413 per student.

They sued, arguing that this violated the Equal Protection Clause of the 14th Amendment. The case went all the way to the U.S. Supreme Court as San Antonio Independent School District v. Rodriguez (1973) and in a devastating 5–4 decision (nearly 20 years after Brown v. Board of Education mind you), the Supreme Court ruled against the parents. The Court held that:

• Education is not a “fundamental right” explicitly guaranteed by the U.S. Constitution.

• Socioeconomic status (being poor) is not a “suspect classification” (unlike race), meaning the government didn’t have to prove a compelling reason for the funding disparities or protection under the 14th Amendment

• Local control over schools was a legitimate state interest, even if it resulted in massive funding gaps between rich and poor neighborhoods.

The Rodriguez decision effectively shut the door on using the federal courts to force equal funding across school districts. It institutionalized a zip code trap for education, ensuring that states were legally permitted to let a child’s educational funding be determined by the market value of the houses around them.

State funding closes the gap, until it can’t

Defeated at the federal level, civil rights lawyers pivoted to state courts. In 1990, the dam broke open when a court found that Kentucky’s entire public school system was unconstitutional because it failed to provide an adequate education to poor children. Civil rights lawyers used this “adequacy” strategy in dozens of other states, allowing more state tax dollars to flow into low-income districts, helping to overcome the inadequate funding from local property taxes. NBER data shows this specific influx of cash directly caused a gradual, steady rise in reading and math scores for disadvantaged students throughout the decade.

By the early 2000s, state governments were providing the largest single share of public education funding (roughly 49%), while local property taxes had dropped to about 43%.

For over two decades (1990 to 2013), American math and reading scores marched steadily upward. But between 2013 and 2015, that progress abruptly flattened, turned negative, and began a decade-long slide that left eighth-grade students roughly 60% of a school year behind in reading and 40% of a school year behind in math compared to a decade prior.

Researchers point to three major structural shifts during the mid-to-late 2000s and early 2010s that explain this decline:

1. The Financial Aftershocks of the Great Recession (2008)

When the housing market crashed in 2008, local property taxes and state revenues plummeted. Because schools operate on a lag, the deepest cuts to education budgets didn’t hit until roughly 2010 to 2013. School districts across the country executed massive layoffs, increased class sizes, eliminated counselor positions, and cut academic intervention programs. Students in the mid-2010s were learning in severely under-resourced classrooms compared to students in the early 2000s.

Even though state budgets for education had increased dramatically, when the housing bubble burst state income and sales tax revenues fell off a cliff. Facing massive budget deficits, 31 states slashed their state education aid per student by an average of nearly 10%. By 2014, the share of education funding coming from local property taxes had climbed right back up to its pre-1990s highs.

Student performance since 2009 has declined in almost every state, with math scores down in 83% of school districts and math scores down in 70%. Well before phones fully penetrated the life of children and households, math and reading scores really plummeted in large part when the housing market fell away.

2. The Dismantling of Accountability Standards (2015)

During the 1990s and 2000s, federal policies like No Child Left Behind (NCLB) forced school districts to hyper-focus on standardized test scores. While NCLB was heavily criticized for causing teachers to “teach to the test,” the intense federal pressure kept math and reading proficiency scores rising.

In 2015, Congress replaced NCLB with the Every Student Succeeds Act (ESSA), which dramatically scaled back federal penalties and test-based accountability, handing power back to the states. The data shows that the moment this strict, test-based federal pressure evaporated, student performance in core subjects began to drift downward.

3. The Screen Explosion and Social Media Disruption

The collapse of student focus directly correlates with a massive cultural shift in the mid-2010s: the absolute saturation of smartphones, the rise of algorithmic social media (like Instagram and Snapchat), and the school-led push toward 1-to-1 personalized laptops in classrooms.

Higher property values, higher grades

The Educational Opportunity Project at Stanford recently released new data on student testing performance across the US and the results are concerning. Almost every state in the country showed a decrease in testing performance, but poorer communities have been hit much harder.

Student performance has declined in almost every state, but local districts reveal a more complex picture. Only 9 states since 2009 have shown an increase in scores, and all improving states are still below the national average. The regions that saw the smallest decline in performance were also the wealthiest regions. The wealthiest districts declined less than middle-class and poorer districts. This highlights a long occurring trend in the American education system that hinders upward mobility and strongly relates income to educational success.

Historically, White families moved from urban neighborhoods into suburban ones, but as minority residents have joined the middle class, the phenomenon of White flight has shifted and segregation continued. Instead, wealthy White families are moving out of more diverse suburban regions into less diverse ones. The demographics of the top 1% of districts are striking. The communities are 72% White, 12% Hispanic and 5% Black. By contrast, the bottom 1% of districts, as measured by school funding, are 51% White, 25% Hispanic, and 15% Black.

Between 2000 and 2010, 150 of the largest metro areas lost at least 20% of their White populations. Areas with strong home values and median incomes show the wealthiest White families leaving as they become more diverse. Over time, these disparities have compounded and the schools with the most funding became havens for majority-White communities. Since 1988, segregation between White and Black students has increased 64 percent in the 100 largest school districts, and 50 percent by economic status since 1991.

The Lower Merion district located in the suburban Philadelphia Main Line boosted its education funding 87%between 2000 and 2015 to more than $23,000 per student. That’s more than double the amount that Philadelphia, one of the poorest cities in America, spent on its students. The school funding gap between a top 1% district and an average-spending school district at the 50th percentile widened by 32% between 2000 and 2015.

Sprawling away from low-income communities

Do rich families just have more means to move to areas where better teachers are already located? Are the schools established in a place and then wealthy families congregate around those institutions? The data shows a bit of a chicken-and-egg problem. More money definitely improves how those schools perform, but wealthy families are constantly monitoring where good school districts are and moving to those.

A 2025 study analyzed a 2011 “information shock” in Chicago, when the city publicly released new school safety and support ratings for the first time. In the very next quarter, home prices zoned to the highest-rated schools experienced an immediate 9% to 21% price premium, while incoming buyer incomes in those exact zones jumped by 16%. This indicates that affluent families actively track school quality and possess the immediate capital to outbid middle-class buyers for homes the moment a neighborhood school is verified as “elite.”

In a complementary study from the Federal Reserve, researchers found that parents were willing to pay a 2.5% premium on home prices for every 5% increase in elementary school test scores.

Urban neighborhoods aren’t avoiding school segregation either. Young affluent White families in recent years have moved into urban neighborhoods as well, but urban public schools still have high poverty rates and lower test scores than suburban schools. This is primarily due to wealthy families in urban cities sending their kids to private schools at higher rates than wealthy suburban families. Private school enrollment for high-income families living in urban regions since 1970 has relatively stayed the same, but middle-class families are disappearing.

Even in gentrifying neighborhoods, demographic change does not equate to integration. One study found that among urban neighborhoods that experienced gentrification between 2000 and 2014, student enrollment declined rather than becoming more integrated, primarily when incoming residents were White. In New York City, the White school-aged population in the fastest gentrifying neighborhoods increased from 10% to 29%, yet White elementary school enrollment only increased from 5.7% to 10.4%, suggesting many incoming affluent families remained disconnected from neighborhood schools.

Research also shows that schools with less diversity have worse test scores than more diverse schools. Integrated classrooms encourage critical thinking, problem solving, and exposure to new ideas. Students in integrated schools are more likely to enroll in college, less likely to drop out, and more likely to seek out integrated settings in adult life.

Private Schools are Increasingly for the Ultra Rich

In 1970, 13% of middle-class families enrolled their children in private schools. By 2011, that number plummeted to just 7%. Private schools have become exclusively for the ultra-wealthy, but historically private schools were for the ultra-religious.

K-12 private school fees have increased 553% since the late 1990s, while average incomes grew only 217%, causing the tuition-to-income percentage to rise dramatically for families.

85% of elementary school students in the US who were enrolled in a private school in 1970 were attending a Catholic school. With the decline in religious attendance since then, non-sectarian private schooling increased, along with the price. Between 1970 and 2011 and adjusting for inflation, the average cost for Catholic private schools increased from $873 to $5,858 in 2011 as they lost enrollment and funding, and non-sectarian schools rose from $4,120 in 1979 to $22,611 in 2011.

High income families increasingly secure educational advantages either through expensive suburban housing or elite urban private schools, while middle-class families are excluded from either option and pushed towards more affordable suburban districts. Lower-income students are left concentrated in underfunded school districts with fewer opportunities and limited upward mobility.

The problem is bigger than the phones

We have allowed public education and the real estate market to become so intertwined that it has begun to choke learning outcomes for our children. The classroom crisis in America is now so strongly mapping to the housing-wealth crisis in America, which is actively starving public education based entirely on zip codes.

We cannot continue to blame smartphones or pandemic hangovers for a systemic crisis that we have legally institutionalized. Before screens ever fractured student attention spans, the 2008 housing crash delivered a far more devastating blow to American education, tying test scores directly to a collapsing real estate market.

America needs to ensure that home property values and student success are not attached at the hip. Boston, Denver, New York and other school districts have started using a system where families rank preferred schools. This is put into an algorithm along with schools’ enrollment priorities, student type and random lottery numbers. The output matches students to their highest ranked school with constraints that help to increase integration.

The wealthy are hoarding public resources behind exclusive zoning laws and private school tuition moats all while the middle class is squeezed out and low-income students are left stranded. True educational equality will never be achieved until excellent schools are treated as a fundamental right for every child.

Subscribe now

Wrap the pipes and cover the lawn chairs. It’s winter in edtech. Among other announcements and legislation nationwide, Los Angeles USD announced its draft plan last week for screens in classrooms. They’re planning to limit screen time for kids in grades 4–5 to 30 minutes per day, grade 2–3 to 20, and grade 1 and below to—nothing. Exceptions for testing and accessibility, but beyond that—nothing.

Every edtech executive with >1,000 followers on LinkedIn (including yours truly) has published their #NotAllEdtech post arguing that we should focus less on screen time and more on screen value. These companies all argue we should focus on how kids use their products rather than how much.

Many of those executives should be careful what they wish for. Indeed, to observe how many products are used in classrooms would only validate the fears of many parents—screens that pacify rather than challenge; screens that isolate rather than connect; screens that decrease rather than increase a teacher’s visibility into a student’s learning.

Show Your Cards

I’d like each of my colleagues to show their cards. I’d like to see from everyone engaged in this discourse: one (1) stationary video of a full classroom session.

Speed it up if you want, but that video will do more to illuminate the real-world social and cognitive impact of these products than any company-funded research study, customer testimonial, or LinkedIn post. Just show us how the pieces fit together, how the humans, ideas, and technology add up to more than the sum of their parts.

Our Cards

Here is a video of NYCPS teacher Liz Clark-Garvey teaching an Amplify Desmos Math lesson called Sand Dollar Search. As a treat, I have coded each segment of the video for “what the students are paying attention to.”

How Our Education Technology Works

At Amplify, we know that, yes, certain enterprising students can learn quite well from an LLM or a library card. But most students benefit enormously from the motivation, accountability, and support they receive from their teachers. We also know that if you ask students “why do you put up with school?” the vast majority of them will say, “Because it’s where my friends are.”

So we use technology as a loom and weave together people and their ideas.

In Liz’s class, you’ll see students on their devices together. You’ll see them use those devices for short intervals—none longer than 8 minutes—13 minutes of screen time total. The devices first stir their thinking, letting them play with math in ways that are impossible with pencil and paper. Then the devices make that thinking visible to Liz who uses it in conversation with the whole class—calling kids to the board to elaborate their ideas, contrasting several ideas together, noting their similarities and differences, never speaking for longer than 90 seconds without checking in with students.

You’ll see students come to realize their work matters and react accordingly: working harder, participating actively, and learning more.

How Most Education Technology Works

Every edtech executive on LinkedIn seems willing to stuff at least one education technology into the wicker man and light it on fire. Everyone seems to agree that unrestricted access to YouTube is bad, for example. Everyone hopes this controlled burn will divert attention from their technology. Me, I hope the light from the fire helps everyone pay more attention.

With lots of education technology, students spend too much unaccountable time on their devices. Everyone works on different things. The dashboard gives teachers limited visibility into that work. Kids know that teachers can’t easily check up on them. They come to realize their work doesn’t matter and react accordingly: drifting off task, onto other tabs, and out of any state parents would recognize as “learning.”

I’m an edtech developer and a parent of elementary school-aged students and I welcome greater scrutiny of our industry. After winter comes spring—a time of growth and renewal. Many edtech companies will try to survive this winter by warming themselves next to a fire that is right now consuming several of their peers. But they should show their cards—show a stationary video of a single classroom—and let parents decide whether or not to use their products for kindling as well.

Parents value the human relationships that schools produce, relationships that support student learning and human flourishing. Everyone in edtech should show their cards. Are they weaving those relationships together or pulling them apart?

A lot of the emails I get from founders are now written in a hard-hitting journalistic style. I know they're written by AI, because no founder ever wrote this way before. And once you realize something is written by AI, it's hard not to ignore it.

I have never knowingly finished reading an email signed by a human but written by AI. It feels like being lied to, and who would stand for that?

[...] It makes me think less of the author. It means they can't write well unaided (or feel they can't), and that they're trying to trick me.

It's not impressive to use AI to write stuff for you; any teenager can do that.

— Paul Graham

Tags: writing, ai-misuse, paul-graham, generative-ai, ai, llms

My Son’s Math Homework Is Essentially Just Pokémon. “As I watched my son play Prodigy, it became clear there wasn’t much learning happening. In about 10 minutes of gameplay, he spent less than 30 seconds answering math questions.”

“Social engineering” sounds like something out of a conspiracy thriller, charged with totalitarian control and fringe paranoia. More mundanely, it’s come to be associated with phishing and other scams, in which fraudsters manipulate people into disclosing personal information.

Yet the concept is older and more benign: it is the deliberate shaping of human behavior, often at scale. It predates silicon—and became pervasive, and ungoverned, especially once its practitioners learned to hide it. Authoritarian regimes and more recently scammers and big companies have profited from it. To defend ourselves from bad actors, and to benefit from social engineering’s good side, we need to reclaim the name, and govern it prudently.

The roots of engineering

In 1894, Dutch entrepreneur Jacques van Marken urged companies to hire “social engineers” to manage human systems such as insurance, education, and profit sharing for workers as carefully as they did mechanical ones. Fifteen years later, reformer William H. Tolman published Social Engineering, describing how U.S. industrialists optimized workers’ conditions alongside manufacturing methods. If industrialists could shape steel and electricity on demand, why not society itself?

By the 1920s, that confidence had spread. The architect Le Corbusier declared that dwellings were “machines for living in,” imagining cities as orderly lattices where people moved like parts on a conveyor belt. Civilization would run like a Swiss watch.

The idea soon darkened. Authoritarian regimes pushed it to extremes, promising to fashion “the New Man.” In Nazi Germany, engineer Fritz Todt founded Organization Todt, a vast state engineering enterprise that emerged from the autobahn highway system and later operated concentration camps using slave labor.

In the Soviet Union, leaders adopted U.S. scientific management techniques to plan factory-worker movements and classify populations through centralized records, feeding both rapid industrialization drives and the gulag system of forced labor. The same tools and managerial methods used to build highways and enact five-year plans worked for repression and mass control.

By the 1950s, “social engineering” had become a contaminated phrase. The revelations of Nazi and Soviet abuses, along with Cold War critiques of grand social planning turned the term from a progressive slogan into a warning label. Banishing the words pushed the practice underground, making it harder to recognize when it resurfaced in new forms—such as organizational psychology and systems management that still relied on classification and behavioral influence techniques but under softer, less loaded labels.

Social engineering’s more subtle spread

In the postwar years, the new social-engineering lexicon included “human factors” and “urban planning,” all promising integration rather than command. As computing advanced, the language shifted again: “customer journey mapping” to track interactions, “user experience” to script them. Engineering, which began as a means of reshaping physical space, set its sights on shaping behavior. Digital design features embedded in our smartphones now target our attention and desire.

Language helps conceal these modern forms of social engineering. “Data analytics” sounds neutral beside “surveillance.” “Personalization” flatters individuality while still sorting users into predictable categories. “Behavioral nudges” guide decisions without the sense of intrusion. We attach “social” as a favorable modifier to sciences, capital, and media, yet recoil when it meets “engineering.”

That discomfort is a clue. Engineering implies control, and control prompts us to ask who directs whom, toward what ends, and with whose permission.

Not all social engineering these days is hidden. Hackers don’t need to break a firewall if someone hands over their password. Romance scammers cultivate intimacy the way farmers cultivate crops. They succeed not through force but by exploiting trust. If even these obvious attacks work, the invisible kind, with roots in social engineering, are a shoo-in.

Most of the social engineering we encounter is proprietary and beyond our control. Firms build recommendation algorithms tuned to boost engagement and profit with no hearings or right of appeal. Browser and cookie defaults decide what data we surrender. A single autoplay toggle can cost users hours and build unhealthy habits. These are acts of engineering as deliberate as laying a road or redrawing an electoral district. They create a kind of curated itch by which boredom never settles, and satisfaction never arrives. The results are predictable—users click on targeted ads, make purchases, form habits, and lock in opinions.

Consent has transformed along with it. Once straightforward and revocable, it is now subtle and persistent, buried in defaults or opaque terms of service too quickly accepted. You remain free to opt out, much as you are free to refuse roads or electricity. Consent has become the preselected setting of modern life.

When social engineering operated more in the open, citizens could contest it, at least in societies with responsive government. Today’s invisible version diffuses accountability so thoroughly that scrutiny becomes hard to direct. Despite recent congressional hearings on social media’s impact on youth mental health and juries agreeing that firms are knowingly designing algorithms that cause harm, pinpointing responsibility remains elusive. When the mechanism is buried inside a system used by billions, we cannot easily point to a single decision-maker or trace the precise moment of manipulation.

Today’s social engineering is less overt and theatrical than its predecessors. Earlier versions arrived on public posters and loudspeakers for mass audiences. Today’s version is more intimate, delivered through personal devices and constant feeds tailored to the individual. The model succeeds because participation feels like freedom, not control.

Not all social engineering is dystopian. Well-kept parks foster community, accessible buildings extend dignity, vaccines and seatbelts save lives. Even in the digital realm, positive examples exist: browser extensions that automatically block hidden trackers, search engines that refuse to build personalized surveillance profiles, and decentralized social platforms that give users greater control over their own data and feeds.

The term “social engineering” still unsettles, though. But “asocial” engineering, which ignores human consequences entirely, is worse. Recognition of the human dimension to engineering is the beginning of repair. Only by seeing the machinery clearly and naming it honestly can we decide who engineers what and why. The machinery will not dismantle itself. Once named, it becomes subject to choice. That negotiation of purpose, power, and process are the defining political questions of any real democracy. We cannot ensure that social engineering serves and sustains society so long as we dodge the words.

Age verification laws have been spreading fast, and we have been keeping tabs on them for a while now. California's Digital Age Assurance Act (AB 1043) was the first to land, signed in October 2025, with Colorado following with its own version (SB26-051).

Neither made any concessions for open source software in the original language, which left Linux distributions and other community-run projects in a very uncomfortable dilemma.

Both have since moved to fix that, with Colorado having wrapped it up earlier this month and California heading for a full Assembly vote.

What's California doing?

AB 1043 required OS providers to collect a user's age or birth date at account setup and share it with apps through a real-time API, starting January 1, 2027. Open Source projects got no special treatment in the original text, which is something we wrote about when the bills started drawing attention.

Assembly Member Buffy Wicks, who authored AB 1043 herself, introduced AB 1856 in February to address that.

After four rounds of revisions, the bill has rewritten the definition of "operating system provider" to exclude anyone distributing an OS under terms that let recipients copy, redistribute, and modify the software.

Most Linux distributions under permissive or copyleft licenses fall cleanly within that.

In tandem, another change covers the application side, where software that is not offered as a standalone executable through a covered app store is no longer treated as an "application" under the law.

The bill passed the Appropriations Committee 11-0 on May 14. It was ordered to third reading on May 19 and is awaiting an assembly vote. Interestingly, Buffy is the chair of that committee.

What about Colorado?

Colorado's path here involved some direct community legwork. Carl Richell, the founder of System76, spent some considerable time working with Senator Matt Ball, one of SB26-051's co-authors, to get open source exclusions written into the bill.

The bill exempts OS providers and developers distributing software under terms that permit copying, redistribution, and modification. It also adds a requirement that exempt software have no platform-imposed technical or contractual restrictions on installing modified versions.

The extra clause is aimed at tivoization, where manufacturers lock down hardware to block modified software from running even when the source code is freely available.

Beyond that, code repository providers, containerized software distributions, and applications from free, publicly available code repositories are explicitly excluded too.

The law also has a narrower scope, only applying to OS providers that operate a covered app store or ship one pre-installed. An OS provider with no app store involvement does not come into scope at all.

Besides that, SB26-051 is now set to take effect on July 1, 2028.

Some closing words…

Neither state got here automatically. The open source exemptions did not exist in either bill to start with, and it took sustained community pressure and direct legislative outreach to get them added.

This is something that can be applied to many other issues, of course. Though, when the representatives are more interested in serving certain interests (say due to pressure from certain lobbies) than their constituents, disruption tends to be the only way out.