Research paper designs and A/B tests a new Rust ownership pedagogy, lifting learner scores from 48% to 57% on a standardized concept inventory (N=342, p<0.001).
Key Takeaways
Paper identifies the core gap: learners recognize surface borrow-checker errors but cannot reason about what undefined behavior would occur if ill-typed code were allowed to run.
Researchers modeled ownership around three explicit permissions per variable: readable (R), writable (W), and ownable (O), visualized as they change statement by statement.
An Ownership Inventory was built from the most common Rust ownership questions on StackOverflow, covering dangling pointers, overlapping borrows, illegal borrow promotion, and lifetime parameters.
A/B test against the TRPL baseline (N=342) showed the new chapter improved Inventory scores from 48% to 57% (d=0.56); comprehension questions on the conceptual model reached 72% accuracy.
Google Android reported zero memory vulnerabilities across 1.5 million lines of Rust, framing the stakes for closing the ownership teaching gap.
