How much can a young brain learn under ideal conditions?

No intervention in the history of education research has produced a comparable effect size.

On the critique

Recent reanalyses of Bloom's 1984 paper have argued that the two-sigma effect was not typical of tutoring in practice. We take that seriously. This study is designed to find the real ceiling, measured carefully, with pre-registered methods, strong controls, and open data.

The methods

Each child receives a dedicated tutor focused only on them for the full year. Tutors are hired as a team, not in isolation. They meet often to share what is working, to teach each other the techniques that move a child forward, and to swap students when a different match will help more.

The control group receives typical classroom instruction. We will publish our benchmarks, evaluations, curriculum, and results in full.

Why now

Now that AI exists, the cost curve of personalized instruction has broken. The data and methods this study produces can inspire builders to recreate these outcomes for every child. Our hope: an education that used to cost $100,000 a year, in every kid's pocket, for under $100.