Rob Race argues that the classic programming tutorial model - build a pristine sample app, show implementation steps, and expect readers to translate that work into their messy, long-lived codebases - is dying. Revenue trends and blunt reports from long-running education businesses like GoRails suggest AI has already disrupted demand. The core problem is the hidden translation step: tutorials show what worked in a specific environment, and developers historically had to integrate those patterns into apps with different models, auth, naming, and accumulated technical debt. Capable coding agents change that by inspecting a real repository, running tests, locating appropriate places to implement changes, and iterating until things pass. Small examples remain useful for learning single APIs or concepts, but production problems - authentication, billing, webhooks, retries - are repo-specific, and agents live on the side where the work actually happens.
That shift changes what learners need and what educators must produce. Beginners gain more execution power from agents but not the experiential judgment to spot architectural pitfalls, so instruction must emphasize decisions, trade-offs, failure modes, and review questions rather than step-by-step builds. Educational products should include an “execution layer” for agents: repository inspection instructions, implementation constraints, architectural contracts, invariants, verification steps, and explicit boundaries of what must not change. The educator’s job becomes richer - documenting reasoning and creating agent-ready artifacts - rather than simply providing static implementation steps.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.