The Orientation Statistic Read From Learner Records Is a Difficulty Sort¶

Claim · All claims
Evidence · 1 study · theoretical r? · q1
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly
Evidence quality · This claim rests on one research release that has not been peer-reviewed. Its research and writing were carried out by an AI system (Claude Code) under one human author, and its scope audit and reproduction were made by AI agents with no part in the work; no research team independent of the author has reviewed or replicated it. Treat it as a lead to check, not an established finding.

The statistic that orients a pair from asymmetric conditional failure points, whenever some learners fail both skills, from the skill fewer learners fail to the skill more learners fail. It is therefore right exactly as often as the true edges run from an easier skill to a harder one, the property known as sortability.

Subclaims¶

Restated 2026-10-14 to match the article. The earlier version read orientation accuracy as a property of the prerequisite mechanism; the identity shows it is a property of how difficulty runs along the edges, which the mechanism helps set.

• With fₐ and fᵦ the failure rates on a and b and j the rate of failing both, the statistic equals j(fᵦ − fₐ)/(fₐfᵦ), so its sign is the sign of fᵦ − fₐ. In all 30 cells of the simulation sweep it oriented every scored edge exactly as a difficulty sort did.
• A prerequisite itself makes the dependent skill fail more often, so deeper skills look harder even where latent difficulty does not rise with depth: where it is independent of depth the statistic oriented 0.883 [0.854, 0.910] of edges at hardness 1.0 and 0.773 [0.736, 0.809] at 0.0.
• Where any one of a skill's prerequisites suffices and later skills are easier, it was right on only 0.009 to 0.020 of the edges into skills with two or more prerequisites.
• On ASSISTments, with grades assigned by a name-matching rule no person checked, the skill from the higher grade was harder in 0.576 of pairs: difficulty rises only weakly with grade there.

Evidence¶

Porcaro 2026¶

David S. Porcaro. (2026). What learner records can say about a prerequisite edge. Research release learning-graph-budget-2026 1.0.0, Learning Design Alliance. https://doi.org/10.5281/zenodo.23480842

q1 · i? · theoretical · r?

An algebraic identity for the orientation statistic and a simulation sweep with one generator (Section 5.1), with difficulty placed against grade order on two real corpora. Observation record: lazuli-graph-simulation-2026.

Discussion¶

What this narrows. Sortability has to be measured in each corpus before any direction is read. An estimator that uses variation in the order learners met two skills is not tied to difficulty, but it needs learners who met the skills in both orders and an order that ability did not choose; in simulation such an estimator was fooled when abler learners chose the order.

Scope. The identity covers one statistic. The sweep is simulated, with one generator.

Related Claims¶

• Learner records rarely establish that a prerequisite edge exists. — the existence limit
• Prior difficulty collections mostly lack fine-grained classification of student difficulties
• Learning-curve plateaus were caused by learners' failure to overcome specific difficulties at critical stages, not by the order in which habits developed — related