In 1898, a Black man in Louisiana walked into a registrar’s office to vote. He was handed a section of the state constitution and told to read it aloud. Then he was told to interpret it. The registrar — a white man with no legal training and absolute discretionary power — decided whether the interpretation was satisfactory. It never was.
The literacy test was not a test of literacy. It was a technology of exclusion dressed in the language of competence. The man could read. That was never the point. The point was that someone else controlled the definition of “literate enough,” and that definition could move whenever a Black person met it.
This is the oldest trick in the American playbook. Not the denial of opportunity outright — that’s too visible, too legally vulnerable, too easy to name. The trick is the gate: a requirement that appears neutral, meritocratic, even reasonable, but whose function is to sort people along lines that were drawn long before anyone walked up to the door.
The gate has been redesigned many times. The language changes. The mechanism updates. The function holds.
Gate One: The Literacy Test
From the 1890s through the 1960s, literacy tests were the primary instrument for denying Black citizens the right to vote across the South. The tests were not standardized. They were not applied equally. They were administered at the discretion of local registrars who had the power to pass or fail anyone they chose — and who chose, with near-perfect consistency, to fail Black applicants.
Some tests required applicants to copy a section of the state constitution by hand. Others required interpretation of legal clauses that law professors would have struggled with. In some parishes, Black applicants were asked how many bubbles were in a bar of soap. White applicants in the same office were waved through.
The genius of the literacy test was its deniability. It never said “Black people cannot vote.” It said “illiterate people cannot vote” — in a country that had spent two centuries making Black illiteracy a matter of law. First you deny a people education. Then you punish them for not having it. And at every step, you call the punishment a standard.
The Voting Rights Act of 1965 killed the literacy test. But it did not kill the gate. The gate just moved.
Gate Two: The Credential
In the postwar economy, the college degree became the new sorting mechanism. A bachelor’s degree was the price of entry into the professional middle class — the gateway to careers in management, finance, education, law, medicine, and public administration.
On its face, this was progress. Education, unlike race, was something an individual could theoretically acquire through effort. The credential economy promised meritocracy: anyone who earned the degree could walk through the door.
But the door was not standing in an open field. It was standing at the end of a path paved by the GI Bill, which channeled homeownership and educational benefits overwhelmingly to white families. It was standing behind tuition costs that rose faster than Black household income for fifty consecutive years. It was standing past admissions processes shaped by legacy preferences, SAT preparation industries, and feeder networks that concentrated advantage in communities that already had it.
By the turn of the century, the credential had become what the literacy test had been: a nominally neutral requirement whose distribution tracked racial lines with devastating precision. The language said “qualified.” The structure said “excluded.” And the people locked out were told, again, that the problem was their preparation — not the architecture of the gate itself.
Gate Three: The Résumé Filter
As the job market digitized, a new gatekeeper emerged: the applicant tracking system. By the 2010s, most major employers used software to screen résumés before a human being ever saw them. The software scanned for keywords, degree requirements, employment gaps, and formatting conventions. It sorted applicants into piles. Most never made it past the algorithm.
The résumé filter introduced something new to the gatekeeping tradition: automation. For the first time, the gate did not require a human registrar, a biased admissions officer, or a prejudiced hiring manager to operate. It ran on its own. And because it ran on data — historical hiring data, which reflected decades of discriminatory practice — it reproduced those practices at scale, at speed, and without a face to hold accountable.
A name that sounded Black. An address in a redlined zip code. A degree from an HBCU that the algorithm had not been trained to weight as highly as a predominantly white institution. A gap in employment that reflected caregiving, incarceration, or the simple economic precarity that disproportionately affects Black workers. Each of these became a signal — not of incompetence, but of difference — and the filter read difference as disqualification.
The résumé filter did not say “we don’t hire Black people.” It said “we hire the most qualified candidates.” The gate was the same. The language was cleaner.
Gate Four: The Algorithmic Score
Now the gate has moved again.
In 2026, algorithmic scoring systems determine who gets a mortgage, who gets approved for a small business loan, who gets flagged for insurance fraud, who gets recommended for a job interview, who gets admitted to a program, who gets paroled, who gets seen by a doctor first. These systems are trained on historical data. They optimize for patterns. They produce outputs that carry the authority of mathematics and the appearance of objectivity.
But the data is not objective. It is a record of the world as it has been — a world in which Black people were denied loans at higher rates, arrested at higher rates, diagnosed later, hired less, paid less, and surveilled more. The algorithm learns these patterns. It does not interrogate them. It extends them into the future, efficiently and at scale, and calls the result a “score.”
The algorithmic score is the literacy test’s great-grandchild. It operates on the same principle: establish a standard that appears neutral, apply it uniformly, and let the accumulated weight of structural inequality do the sorting. The test-taker never sees the registrar’s face. The applicant never meets the hiring manager. The borrower never sits across from the loan officer. The gate is invisible. The exclusion is automated. And the language — data-driven, objective, evidence-based — makes it almost impossible to name what is happening as what it is.
Gate Five: AI Fluency
And now, one more turn.
As artificial intelligence reshapes the economy, a new competency is emerging as the gateway to economic participation: AI fluency. The ability to use AI tools productively. The ability to work alongside automated systems. The ability to prompt, evaluate, direct, and integrate AI outputs into workflows that generate value.
Employers are already sorting for this. Job postings increasingly require “experience with AI tools” or “comfort with generative AI workflows.” Companies are restructuring teams around human-AI collaboration. Entire career paths are being rerouted through AI proficiency. The workers who have it will advance. The workers who don’t will be automated, sidelined, or replaced.
This is, in many ways, the most insidious gate yet — because it operates under the banner of empowerment. Learn AI and thrive. Upskill and adapt. The future belongs to those who embrace the technology. The message is optimistic. The subtext is familiar: if you fall behind, it’s your fault.
But who has time to learn AI tools when they’re working two jobs? Who has access to the premium-tier platforms when the free versions are limited? Who gets employer-sponsored training when they work in the gig economy? Who has the professional networks that translate new skills into new opportunities? Who gets the venture capital to build AI-powered businesses? Who gets the second chance when an algorithmic score has already marked them as high-risk?
The gate is open, technically. Anyone can walk through. But the path to the gate runs through the same uneven terrain it has always run through — and the people standing at the back of the line are the same people who have been standing there for four hundred years.
The Pattern
Here it is, stripped bare:
1890s. You must be literate to vote. We made it illegal for you to learn to read.
1950s. You must have a degree to advance. We excluded you from the institutions that grant them.
2010s. You must pass the algorithm to get hired. We trained it on data that reflects your exclusion.
2020s. You must be AI-fluent to compete. We built the industry without you and priced the tools beyond your reach.
The language changes. The technology updates. The gate remains.
And at every stage, the same move: the burden of overcoming a structural barrier is placed on the individual standing in front of it. Learn to read. Get a degree. Optimize your résumé. Upskill for AI. The instruction is always addressed to the person, never to the system. The gate is never questioned. Only the people failing to get through it.
Breaking the Pattern
SOLE refuses the premise that the answer to structural exclusion is individual adaptation.
Individual adaptation matters. AI literacy matters. Skills matter. But if the history of Black economic life in America teaches anything, it is that literacy without structural change is a treadmill. Black people learned to read, and the test moved to interpretation. They earned degrees, and the credential was devalued. They optimized their résumés, and the filter changed its keywords. They adopted AI tools faster than any demographic in the country, and the industry still hasn’t hired them, funded them, or governed itself with their interests in mind.
The pattern does not break from the bottom up. It breaks when the gate itself is redesigned — when the systems that sort, score, filter, and credential are built with the people they affect, governed by the communities they impact, and held accountable for the outcomes they produce.
That is not an AI problem. It is a democracy problem. And it is the oldest one we have.
