How to Accelerate Learning & Improve Education | Joe Liemandt

Why does it matter? Because the gap between a 710 and a 790 SAT score was third-grade multiplication tables.

A high school student with a 710 math SAT kept making what everyone called careless errors. Joe Liemandt's team at Alpha School found the real problem: she'd never memorized her multiplication tables, so she was burning finite working memory on arithmetic that should have been automatic years earlier. She drilled them. She scored a 790.

That story contains the entire argument of this episode.

• Every non-automated foundational skill silently burns a working memory slot — capping advanced thinking regardless of effort or IQ • The average American student gains exactly one standardized test point from 8th to 12th grade, because advancing at 80% mastery compounds into catastrophic foundation gaps • One grade level of K-8 curriculum takes 20 to 30 hours to close with level-appropriate practice — "three years behind" is really 60 to 90 hours • Between 40 and 60 percent of Alpha School kids say they prefer school over vacation — driven by maximally high standards, not low ones

Fluency is a working memory hack — automating the basics frees the cognitive slots needed for hard thinking

The problem in complex subjects is almost never at the level where the struggle is visible. It is two or three layers below, buried in basics that never reached automaticity.

Liemandt explains the mechanism through cognitive load theory: working memory has a fixed number of slots, roughly proportional to IQ. But fluency operates independently of IQ. "If you memorize something to fluency, when you use it, it doesn't take a working memory slot." The inverse is the hidden tax on every student with unautomated basics. "If I have to calculate 7 × 8, I'm actually using my working memory slots. The problem is I only have a few of them."

A multi-step algebra problem needs those slots free for the algebraic reasoning. If two are occupied by arithmetic the student never automated, the harder task becomes cognitively impossible — not because the algebra is too hard, but because the computational budget is gone. "When kids get to algebra and they start to struggle, the issue is not the algebra. The issue is the third grade math facts they didn't learn."

The cascade runs in both directions: chemistry is algebra with word problems; algebra requires fast fraction manipulation; fractions require automated multiplication tables. A gap anywhere propagates upward invisibly until the student hits what feels like a sudden, inexplicable wall. Huberman connects this to working memory research on smartphone proximity — a phone face-down on a desk degrades complex thinking not through distraction but through unconscious cognitive load. The mechanism is identical: cognitive resources are finite, and anything not fully automatic draws on the same limited pool.

Audit the foundational skills in any domain you want to master. If you still have to compute a basic, you are burning cognitive RAM that should go to the hard parts. Drill to true automaticity — not 80% familiarity.

Bloom proved any student can reach top-2% outcomes — but the two conditions that produce it are the same ones every school explicitly rejects

Give any student individual tutoring at a mastery level, and they learn at two standard deviations above the classroom mean — top 2% outcomes, regardless of starting point. Benjamin Bloom demonstrated this in the 1980s. It is one of the most replicated results in educational science and one of the most consistently ignored.

The part nearly every account leaves out: both conditions must be present simultaneously. "One sigma came from the tutor. One sigma comes from mastery." A tutor who doesn't require mastery before advancing produces one sigma. A mastery system without individualized instruction produces one sigma. "We all accept mastery in athletics, but we explicitly reject it in education." A coach who moves a player on while they're losing the ball 20% of the time is considered negligent. A school that promotes a student at 80% is considered normal.

The replication problem was always cost. One-on-one tutoring at 30:1 student-teacher ratios is not affordable at scale. AI changes that arithmetic. Alpha School students average 121 minutes per day in adaptive, mastery-gated apps and outlearn peers by more than 2x — in roughly one-third the conventional instructional time. Liemandt estimates the effective learning rate at 5 to 10x standard schooling depending on subject and grade.

Two questions reveal whether any learning tool is worth the investment: does it adapt to the learner's exact current level? Does it require mastery before advancing? Both must hold simultaneously. One without the other produces half the effect.

The average American student gains one measurable test point across four years of high school — because 80% mastery compounds into a complete learning stall

The NWEA MAP test runs on a 300-point scale. The average 8th grader scores 233. The average 12th grader scores 234. Four years. One point. Liemandt cites this not as a COVID effect or a teacher quality problem — it is the structural output of a system that advances students on 80% mastery.

"In America, we're like, 'You're getting a B, man. Move on. We're moving on to dunks. You just finished fourth grade at 80%, let's get you into that fifth grade stuff.' And the answer is no. Because then all those problems compound."

A basketball coach watching a player lose the ball on 20% of possessions does not say that's good enough and move to alley-oops. The 20% is unacceptable. The 80% is not a passing grade — it is a liability that will compound under game conditions. Academic promotion works the same way, invisibly.

By 8th grade, Liemandt says, a typical classroom contains students who functionally need 3rd, 4th, and 5th grade instruction. The 8th grade teacher, by law, delivers 8th grade content. The gap is uncrossable from within. "Chemistry is basically algebra with cool word problems. And if you don't know algebra, the time to try to learn algebra is not in the middle of a chemistry problem."

Students transferring into Alpha from $50,000-per-year private schools arrived 2.2 grade levels behind on assessment — not because those schools were failing, but because the system was designed for time-based advancement, not mastery.

When you hit a wall in any subject, pushing harder at the current level is almost always the wrong move. Diagnose which foundational layer is missing — often two or three levels back — and rebuild from there.

'Three years behind' produces paralysis; '22 hours behind' produces a plan

Twenty to thirty hours. That is what it takes for an average K-8 student, working through level-appropriate individualized practice, to close one full grade level of curriculum in a single subject. Liemandt leads with this number because it transforms how students respond to bad diagnostic news.

"When you're like, I'm three years behind, you're like, I'm out. It's too far." Three grade levels sounds like a verdict. Translate the same deficit into hours and the frame collapses: "When you're like, you're 22 hours behind — totally different message." At one hour of focused daily practice, three grade levels is 60 to 90 days.

Students entering Alpha from elite private schools arrived an average of 2.2 grade levels behind on assessment. Liemandt told parents not to panic, then explained why: in hours, the gap was far smaller than the grade-level framing implied. "We are taking kids who are years behind and catching them up because they see it and say, 'I can do an hour of homework a day to go fix that problem.'"

Perceived distance from a goal is the primary variable controlling whether someone engages or quits. Naming an hour count rather than a grade count compresses that perceived distance to something the brain can evaluate as achievable.

Before abandoning a hard skill, calculate the actual hours required to close the gap with focused, level-appropriate practice. Name the number. The felt distance between current ability and competence almost always shrinks when made concrete.

Alpha kids prefer school over vacation because standards are highest there — challenge paired with a caring adult is the mechanism of intrinsic engagement

Between 40 and 60 percent of Alpha School students say they would rather attend school than go on vacation. The number shifts by session and by how strong that period's workshops were — but it stays above 40 under any conditions Liemandt has measured. He presents it as the central design proof. "The key to your child's happiness is high standards. And I can't run a school that they love more than vacation if I have low standards."

The mechanism runs through a specific developmental loop Alpha borrowed from child development research: kids struggle and fail, sometimes cry, on their road to success, supported by a caring adult. Kindergarteners climb a 40-foot rock wall. Some stop partway. Their parents want to intervene. The guide coaches from the ground. The child finishes, rings the bell, comes down beaming. Eighty-five percent of Alpha parents reported in the past year that they saw their child do something they previously believed was impossible for that age.

Parents readily accept this loop in youth sports. They already know a great coach sets an apparently impossible standard and then walks athletes through failure. That loop produces engagement precisely because it produces real accomplishment — and the same mechanism works academically and almost never gets deployed there.

When engagement drops in any learning environment, lowering difficulty is almost always the wrong response. Ask first: is the challenge paired with genuine support? Has the person experienced a recent win? Is the goal meaningful to them, not just to you? Those are the actual levers.

Identity blocks obstruct learning more than cognitive limits do — and the only fix is engineering a real experience of success

Three middle school arts-and-theater girls — none of them self-described math students by any measure — all scored 790 on the SAT math section. They got there because Liemandt understands a specific property of identity-based learning blocks: you cannot argue someone out of them. They require disconfirming evidence.

"These kids have a block that they can't do it. You do whatever you can including paying them to break through the block. They then see themselves as a different person, which then creates the intrinsic motivation going forward."

Liemandt paid his own daughter $1,000 — she filled a shopping cart with what she planned to buy — to reach top 1% in math and science. She got there. She told her mother afterward: "I didn't think I was as smart as my older sister, but now I realize I can do everything she can do if I just put the effort in." The money stopped mattering the moment the identity updated.

The self-determination theory concern — does external reward kill intrinsic motivation? — applies to a different situation. The incentive is not sustaining long-term engagement. It is getting someone past the first threshold of disconfirming evidence so the self-model can update. After that, sustained effort follows from the new identity, not the incentive. "Motivation is 90% of learning these days."

When you or someone you know believes they cannot learn something, treat it as an identity claim requiring evidence, not an ability ceiling requiring encouragement. Design for one real experience of success, by whatever means necessary. The rest follows.

The only metric that reveals actual educational value-add is growth rate — achievement tells you mostly who walked in the door

"The single biggest indicator of your performance is your family income." Liemandt opens with this to make a measurement point: high-income families send high-performing students to high-performing schools. What the school actually contributes — the value-add — is nearly invisible inside that correlation.

Alpha School benchmarks on growth rate instead: 2x learning in two hours, measured independently of starting level. Crucially, Liemandt specifies this holds across the entire percentile distribution. Top 1st percentile students at Alpha learn twice as much in two hours as top 1st percentile students at conventional schools learn across a full day. The 2x figure is not a function of who enters.

"What we should be measuring schools on is development. Schools are supposed to develop kids, not just select them."

The selection-effect critique of elite schools is widely acknowledged and rarely acted on. Liemandt's version is sharper: a school that recruits exceptional students and produces average growth is a fundamentally different institution from one that produces 2x growth at any starting level. Achievement metrics make them look similar. Only the slope of the line reveals the difference.

The question extends to any environment — job, lab, training program, gym. What growth rate does it produce, and is that rate independent of who enters? A program that selects for high performers and produces mediocre development is not a high-performance environment. It is a selection filter. The two are easily confused and rarely distinguished.

The industrial classroom was engineered by Prussia to produce compliant factory workers — fixing education requires replacing the architecture, not improving the content inside it

The Prussian model that became universal education was not a compromise or a rough draft of something better. It was built to specification. "We need people who follow rules, who do what we're told." Students lined up in rows, passively receiving information, were being trained for industrial factories — and a century of that design has produced its intended behavioral output.

Liemandt's argument is structural rather than conspiratorial. A system optimized for passive compliance produces passive consumers. "Our education system was built to create consumers." Kids caught in dopamine loops on TikTok and Fortnite are not failing the system. They are running the program the system installed.

The fix cannot come from better content delivered through the same architecture. Six hours of passive instruction with improved material is still six hours of passive instruction. AI that compresses those six hours into two gives students back four — and time is the structural input that makes builders possible. Alpha School's afternoons are the proof: kids who finish mastery-gated apps in the morning spend afternoons starting food trucks, raising $25,000 in music festival sponsorships, building bike parks that SpaceX noticed, publishing in Nature.

"Give kids their time back." If passive consumption is your default mode — absorbing without creating, scrolling without building — treat it not as a character flaw but as the trained output of a system designed for another era. Protecting time to build something is a structural correction. The architecture has to change first.

The performance gap was always mostly a delivery problem — AI just made the fix affordable at scale

What Liemandt is describing is not primarily an education reform story. It is a delivery story. The gap between elite outcomes and average outcomes is not mainly explained by differences in raw human capability. Bloom showed in the 1980s what the ceiling looks like when every student gets individualized, mastery-gated instruction. The conditions existed. The cost was prohibitive. That was the only reason the gap persisted.

When broad access arrives, the gap should compress — because it was always mostly a delivery failure, not a talent distribution.

The most actionable thing Liemandt says: you are 22 hours behind. Not three years. Count the hours.

---

Topics: education, mastery learning, working memory, cognitive load, AI tutoring, motivation, learning science, educational reform, neuroplasticity, Alpha School, Bloom two sigma, fluency, foundational skills, growth rate, identity blocks