Mastery Learning

Mastery learning is an instructional approach in which a student advances to the next topic only after demonstrating mastery of the current one — typically 90 percent or better on an assessment — inverting the traditional model where time is fixed and understanding is allowed to vary.

Conventional education holds the calendar constant: the class spends two weeks on a unit, sits a test, and moves on together — carrying whatever gaps the test revealed into material that builds on them. Mastery learning, formalized by Benjamin Bloom in the late 1960s, flips the variables: understanding is the constant, and time flexes to reach it.

How it works

The loop is simple and unforgiving:

1. Teach a well-defined unit.

2. Give a formative assessment — a check that exists to locate gaps, not to assign grades.

3. If mastery is shown, advance. If not, deliver corrective instruction targeted at the specific gaps — a different explanation, more practice, a different angle — and assess again.

4. Repeat until mastery, then move on.

The consequence is that grades stop measuring ranking and start measuring nothing at all — everyone eventually reaches roughly the same standard; what varies is how long each unit took. In cumulative subjects like mathematics and engineering this compounds powerfully, because every unit stands on genuinely solid ground rather than a 72-percent-understood foundation.

Why it matters

Bloom's research group produced one of the most cited findings in education: the 2-sigma problem. Students taught one-on-one by a tutor using mastery techniques performed about two standard deviations better than students in conventional classrooms — the average tutored student outscored 98 percent of the conventional class. Bloom framed it as a challenge: tutoring plus mastery is astonishingly effective but was economically impossible to scale, so what combination of methods could approach it at group cost? Mastery learning alone, in his data, recovered roughly one sigma of the two. Fifty years later, AI tutoring makes the other sigma look reachable — which is precisely why the 2-sigma paper is enjoying a second life.

A worked example

Two engineers learn SQL joins. In a fixed-pace course, both get one week; Engineer A scores 95 on the quiz, Engineer B scores 70 — and both proceed to query optimization, where B's shaky join model quietly poisons everything downstream.

Under mastery learning, B's quiz triggers correctives: the diagnostic shows outer joins specifically are the gap, B gets a Venn-diagram-free explanation plus ten targeted exercises, retests at 93 two days later, and then advances. B spent nine days where A spent five — and both actually understand query optimization when they get there. The week was never the point.

How Miatz uses it

Mastery learning is Miatz's operating system, not a feature. Admission through the DSAT establishes a baseline; from there, progress through the 28 competencies is gated by demonstrated mastery — coding reps, scheduled reviews, and Saturday assessments decide advancement, never the calendar. The AI tutor grounded in each learner's personal memory engine is Miatz's run at Bloom's challenge: tutoring economics finally cheap enough to pair with the mastery standard.