Joe Liemandt on How Alpha School Speeds Up Learning

A basketball coach never tells a player who loses the ball one time in five to move on to dunks. Twenty percent is a crisis, not a passing grade. Joe Liemandt, principal of the Alpha School network, opens with that image because it captures what he thinks is broken in American classrooms: we let a student finish fourth grade with an 80 percent understanding of the material and send them straight into fifth grade content built on top of the missing 20 percent. In sports, gaps get fixed before you move on. In school, they compound for twelve years.

Liemandt did not start Alpha School. It was founded twelve years ago by Mackenzie Price, and Liemandt became a parent first, a skeptic second, and principal fifth. His pitch is not that children need more screens. It is that the thing actually limiting learning today is not access to information. Anyone can watch an MIT lecture on YouTube. Almost nobody becomes an MIT graduate by doing that. What is missing, Liemandt argues, is structure and motivation, the two things a good school still has to supply.

The Gap Nobody Named

A century ago, most of the world had no schooling at all. The Prussian model that spread afterward solved a real problem: it gave everyone access to a teacher and a room. Liemandt's complaint is that access and learning got confused for one another. "We have universal access. What we are trying to do is bring universal learning," he says (04:34). A rich district and a poor one can both put a teacher in front of thirty kids. What happens inside those twelve years of instruction can differ enormously.

The theoretical ceiling for closing that gap has been known since the 1980s. Educational psychologist Benjamin Bloom published a study showing that one-on-one tutoring, delivered to mastery, could move an average student into the top 2 percent of performance, a result researchers call the "two sigma" effect (30:18). The catch, which Liemandt says gets left out of the story, is that both halves matter: individualized instruction and mastery. Skip mastery and you lose most of the gain. Historically, one-on-one tutoring for every child was not affordable, so society settled for classrooms of twenty or thirty.

What changed, in Liemandt's account, is that AI tutoring software can now generate individualized lessons cheaply, at the level each student actually needs, and track whether they actually mastered it. Alpha students spend an average of 121 minutes a day on these apps (31:28) and, according to Liemandt, learn more in that time than they would in six hours of traditional class plus homework. He credits the technology less than the loop it creates. A teacher standing in front of a room gets noisy data: who was paying attention, who had the prerequisite knowledge, who was distracted. An AI tutor watching one student's answers in real time gets a clean signal, which Liemandt compares to what the microscope did for biology or the analytical balance did for chemistry (44:22).

The clearest illustration involves a very unglamorous idea: memorizing multiplication tables. Liemandt describes a student stuck with a low 710 math SAT score despite being told, repeatedly, to just be more careful. The actual problem was that she had never memorized her times tables to fluency, so every multiplication step during a word problem consumed one of her limited working memory slots. "If I know seven times eight is fifty-six, that takes nothing, right? I, you just know it. I have it to fluency," Liemandt says (53:32). Once she went back and drilled the basics, her score rose to 790. The lesson he draws is blunt: struggling students in algebra usually don't have an algebra problem, they have an unaddressed arithmetic problem, and piling on more algebra instruction won't fix it.

Alpha's own internal data pushed Liemandt to revise his views mid-course. He initially assumed spaced repetition alone, without requiring 95 percent proficiency on first pass, could substitute for old-fashioned drilling (49:05). That held true for eighth through twelfth grade, where worked examples carried most of the weight, but scores dropped in fourth through seventh grade until the school reinstated an initial mastery requirement. The finding that median students learn almost nothing academically between eighth and twelfth grade, based on NWEA MAP assessment data, is the backdrop for all of this: "High school, the median student in America doesn't learn anything" (62:10).

None of this works, in Liemandt's telling, without solving motivation first, which he estimates accounts for 90 percent of the problem. A student who is three years behind hears an impossible number and gives up. Liemandt's fix is reframing: a gap of three years sounds hopeless, but the same gap described as 20 to 30 hours of focused work per missing grade level (64:39) sounds doable. "You're twenty-two hours behind. It's twenty-two hours," he says of one reframed conversation (66:48). New students transferring into Alpha, even from $50,000-a-year private schools, arrive testing an average of 2.2 grade levels behind (63:50), which suggests the gap he's describing is common, not exceptional.

Where Liemandt gets more provocative is on incentives. Alpha has paid students directly, in cash, to break through a belief that they cannot succeed at something. A group of middle school girls who assumed they weren't "math and science people" were offered money to reach the top 1 percentile; his own daughter filled a $1,000 shopping cart in advance as motivation and later hit the target. Struggling seventh graders were paid $100 per grade level to take remedial tests they were embarrassed to take, working backward from third grade until their knowledge caught up (78:01). Liemandt argues the payment is a bridge, not the destination: once a student experiences the success, the external reward gives way to a revised sense of what they're capable of. Public schools, he notes, are legally barred from paying students cash, which he considers an unnecessary obstacle.

Builders, Not Just Test-Takers

The two hours of core academics are only half of Alpha's model. The rest of the day is built around what Liemandt calls "philosopher builders," a phrase meant to combine rigorous thinking with the habit of actually making things. Examples pile up quickly in conversation: a high schooler who trained a language model on dating advice and will be, in Liemandt's telling, the first Alpha student to be lead author on a paper published in Nature (100:15); a founder-school track that promises students a shot at $1 million in real business profit or a tuition refund; a student who built what became the best bike park in Texas.

Liemandt situates all of this against a broader anxiety about consumption. Many students arrive from what he calls a dopamine loop of TikTok and Fortnite, and the school's answer is a program called Dream Launcher, which asks students to track how they actually spend their hours against the person they say they want to become. The gap between the two, he says, is often the first real motivator a disengaged student encounters.

The obvious rebuttal to all of this is that Alpha is an expensive private school serving a self-selected, often affluent population, and that its results say more about who enrolls than about what works. Liemandt doesn't dodge this. He points to Alpha's Texas Sports Academy, funded through the state's new voucher program for families earning under $65,000 a year (141:39), which drew more than 120,000 applications for roughly 1,000 spots and is now, by his account, the largest recipient of that voucher category in the state. He also notes that family income remains the single strongest predictor of standardized test performance in the current system, ahead of IQ and even the personality trait of conscientiousness (131:48). Whether AI-personalized tutoring can actually break that link at scale is now the subject of a research partnership with MIT's Blueprint Labs, which Liemandt says will run randomized controlled trials rather than rely on Alpha's own numbers.

The honest version of Liemandt's argument isn't that every child can become MIT-eligible by middle school, a goal he reserves for a separate gifted track. It's narrower and more testable: that the frontier of what any given student can learn is farther away than the current system assumes, and that the tools to close that gap, cheap individualized instruction plus a reason to care, now exist outside of elite tutoring. Whether that promise survives contact with the ordinary public school system, with its budgets, unions, and legal restrictions on paying kids in cash, is the experiment Liemandt says he is only a few years into running.