Hypothesis

Statement

For mixed-expertise technical audiences, a short linear conceptual overview linked to a canonical non-linear reference layer will outperform either layer alone on combined orientation and retrieval outcomes.

Mechanism

The overview builds a global schema; the reference graph supports precise lookup without forcing experts through tutorial prose.

Predictions

Novices will orient faster than with reference-only material; experts will retrieve exact requirements faster than with overview-only material; drift between layers will erase the advantage.

Evidence that would support it

Randomized task study measuring initial mental-model accuracy, delayed retrieval, critical interpretation errors, and assistance by expertise.

Evidence that would contradict it

No improvement over a well-designed single layer, or increased errors caused by conflicting abstraction levels.

Tests performed

Desk application to technical instruction and research-handoff cases only.

Results

Untested outcome hypothesis; architecture is plausible but not validated.

Falsification attempts

Identified link chasing, maintenance drift, and duplicated definitions as credible mechanisms of harm.

Current assessment

Proposed at Medium (0.52 judgment).

Next experiment

Compare overview-only, reference-only, and linked two-layer documentation on a versioned API task.