Your Review: Alpha School

[Guest post - a 2025 book-review-contest finalist, scored on the same axes.] A thorough, balanced, first-hand investigation of Alpha School (the '2-hour AI learning' program): the author moved his family to Austin for a year to test it on his own kids. Across seven parts it strips the PR from the reality - the platform is 'a turbocharged spreadsheet plus spaced-repetition,' not generative AI; 'guides' (not teachers) run a 5:1 ratio; the undiscussed 'secret sauce' is an incentive system (Alpha bucks) grounded in Fryer's pay-kids-to-read research and Ericsson's deliberate-practice motivation stages; the '2.6x' is a MAP-growth figure with real teaching-to-the-test caveats - and situates it in the education literature (Caplan's signaling/range-restriction, Chetty's neighborhoods, Polgar's chess prodigies, Bloom's 2-sigma tutoring problem), framing Alpha as software-mediated tutoring that narrows Bloom's gap at scale.

Quality 80 · Excellent. An exceptional guest review: original first-hand field research nobody else did, balanced and PR-skeptical, well-structured, and genuinely illuminating about a hyped and poorly-understood program, with a strong synthesis situating it in the tutoring/2-sigma literature. A notch below the host's best reviews on polish and concision. Excellent.

Claude’s paradigm shift 50 · Moderate. Reviews an existing program and existing research (Caplan, Fryer, Ericsson, Chetty, Polgar, Bloom); the fresh contribution is the first-hand reporting and the 'Alpha as scalable Bloom-tutoring, incentives as the real secret' synthesis, not new ideas. Moderate.

Real-world impact 2 · Minor. An exceptional guest review built on original first-hand field research (the author moved his family to test Alpha School) that genuinely illuminates a hyped, poorly-understood education program. Real topical relevance, but a review of niche reach with no material change — low RWI.