White-box-then-black-box sequencing: transparent study of a method before opaque delegation¶

Strategy · All strategies
Evidence · no claims cited

Description¶

Adopting Buchberger's principle from computer-algebra education, the article prescribes that "a method should be studied transparently before it is invoked opaquely, and yesterday's white box becomes tomorrow's trusted primitive." In the author's courses each method class gets a white-box phase—students build a fifteen-line Euler integrator, watch it fail at large dt, and implement it on an AI-free quiz—before the course moves to solve_ivp and later to directing agents. The phase is sized for conceptual residue, not professional fluency, and the author calls whether a brief phase suffices the framework's central untested assumption.

Design Implications¶

Context¶

Requirements¶

• The white-box phase must be deliberately scheduled, since the practice the workflow no longer provides incidentally must be manufactured deliberately.

Constraints¶

• The dose is untested: the article notes pilots' manual-flying skills decayed under training doses far larger than a brief white-box phase, and the constructionist tradition may yet demand more building.

Target Learners¶

• undergraduate physics students in computational courses

Target Learning Goals¶

• conceptual working model of numerical methods sufficient to catch agent errors

Affordances¶

• Five Pillar Tool Invariant Computational Framework

Related Strategies¶

• Error-injection exercises: diagnosing subtly wrong simulations as graded practice

Examples¶

-

Key Sources¶

• Larry Engelhardt. (2026). A Tool-Invariant Framework for Teaching and Assessing Computational Methods in the Age of Agentic AI. https://arxiv.org/abs/2607.15518