The Multiplier

On Thursday, July 9, Alpha School premiered a short documentary called Teachers 2.0 to a live online audience and, in the same week, kept restating the claim that has followed the school into every headline about it.¹ Students in an Alpha classroom, the school says, learn twice as fast as students in a traditional one. Some marketing pages say 2.6. A newer platform note claims 10. The documentary itself was less a data release than a mood, a set of educators talking about the role of the human in a room where the software does most of the direct instruction.¹ But it arrived at a moment when the school is no longer a curiosity. Enrollment has grown from about two hundred students to more than a thousand across twenty-two campuses in two years. This fall the network will expand into Puerto Rico, Chicago, Atlanta, North Carolina, and the Seattle suburbs, past thirty-five locations in total.²

The same week the documentary premiered, a smaller headline ran under it. AI private schools promise twice the learning. Experts cannot verify that claim.³ That sentence is the story of AI in schools this July.

What The Number Is Made Of

The 2x claim is built from NWEA MAP scores. A student takes the assessment. Her observed growth is divided by a projected growth for demographically similar peers. If observed is double projected, that is 2x. Alpha's own report on the winter of 2025 to 2026 places the average student well beyond the ninety-ninth percentile of national norms, a number so unusual that the arithmetic alone tells you something about who is enrolling and who is not.⁴ Independent scholars have asked, patiently and repeatedly, for outside access to the raw score distributions, the standard deviations, and the way the sample was selected. They have not received it.³

A public education researcher writing on the National Education Policy Center's blog earlier this year noted that Alpha's parent platform, Timeback, advertises gains of up to ten times faster than in traditional classrooms without linking to a single supporting document.⁵ The Canadian Broadcasting Corporation, reporting on the model in April, quoted a Vancouver-area superintendent who is not opposed to AI in classrooms but noted the plain thing. Some children need someone. A tool cannot always tell which ones.⁶

None of this is a takedown. Alpha is doing something serious and consequential. Two hours a day of adaptive practice, a room of adult guides rather than instructors, workshops in life skills, and a growing number of families willing to pay forty to seventy-five thousand dollars a year for the arrangement.² The families are choosing something. The industry is watching. The claim is what carries the story into the next city.

What A Score Cannot Show

The more interesting question is not whether Alpha's number is real or inflated. It is why the number is the only thing anyone outside the building has to work with. MAP is a good instrument. It cannot see the fifteen minutes a student spent stuck on a problem before the software escalated her hint. It cannot see whether her answer came from the reasoning she built or from an option the tool suggested. It cannot see the third revision she quietly deleted because it was easier to accept the fluent one. It cannot see the moment she understood, or the moment she pretended to.

Two schools with the same MAP delta can be doing utterly different things. In one, a student worked harder than she ever has. In the other, a student was expertly walked to the correct final keystroke on a test aligned to the software that taught her. The score does not distinguish. Nobody in the industry yet has an instrument that does. And so the country is being asked to buy a claim that its measurement tools, by design, cannot confirm or deny.

Where The Record Comes In

This is the part of the debate we spend our days thinking about at Koan. Not as a rival to Alpha, and not as a defense of the classroom as it was. We think a great deal about what the record of a child's thinking would have to look like for anyone, teacher or parent or superintendent or investor, to answer the question the multiplier keeps dodging. Did the student learn, or did the tool arrive at the answer for her.

The record does not need to be exotic. It is the pause before a word is chosen. The paragraph tried once and then thrown away. The moment a hint was asked for, and the moment it wasn't. The seven small revisions between the first sentence and the final one. Those are the traces of thinking, and they are the only honest evidence that thinking happened. A summative score, in isolation, is compatible with almost any story about how it was produced.

The Documentary And The Instrument

Teachers 2.0 will be watched, and Alpha will keep growing. That is a real event, and it deserves a real response. The response cannot be a slower documentary from the other side. It has to be an instrument. If a school claims that a student is learning twice as fast as she would have learned anywhere else, the country now needs a way to look at the work she did and see, plainly, whether she did it.

The industry has spent this year arguing about the multiplier. The interesting argument is the record underneath it.

If a school hands you a number, and the process behind it is invisible, is what you have bought a school, or the number?