Progressive complexity in training design

Most technical training fails in a specific, predictable place: the jump from layer one to layer two. Layer one — the guided introduction — is usually fine. Then the course leaps to realistic complexity and learners fall into the gap, because the intermediate layer was never designed. It was assumed.

Load-bearing layers

The design principle: each layer must be load-bearing for the next. A layer is load-bearing when a learner who has completed it can predict what the next layer will need to address — when the limitations of the simple model are felt before the complex model is introduced.

1. Layer 1 — the working caricature. A deliberately simplified model that produces correct behavior in the common case. Wrong in known ways, useful immediately.
2. Layer 2 — the breakage tour. Exercises designed so the caricature fails. This is the layer everyone skips. The learner must experience the failure, not be told about it.
3. Layer 3 — the real model. Introduced as the resolution to felt problems, not as a wall of upfront completeness.
4. Layer 4 — judgment. Cases where the right answer is "it depends," with the depends made explicit.

Objectives before content

For each layer, write the learning objective as an observable capability — "can diagnose X given Y," never "understands X." If you can't describe the observable behavior, you don't have an objective; you have a topic. This is the same forcing-function logic as product specs: the structure exists to make design decisions happen, not to hold content.

Why experts design layer two badly

Expertise compresses. Once you know something deeply, the intermediate confusions become invisible — you genuinely cannot remember what it was like to hold the caricature. Which is why layer two should be designed from learner failure data (questions asked, errors made) rather than from the expert's reconstruction of their own path.