Prospective validation protocol¶

Status: protocol, not a completed study. Existing All Souls mappings are preserved external intellectual proxies. Imported dates and freeze claims are not newly verified evidence of unseen evaluation.

Freeze before exposure¶

Archive the architecture, teaching attachments, source assignments and scoring rubric with immutable hashes. Record who designed the curriculum and who selected the evaluation tasks. The selector records prior exposure and keeps task content outside the public development tree until administration.

Register population, prerequisites, accommodations, safety limits, sampling rules, exclusions, primary outcomes and stopping criteria before opening the task set. Distinguish a public benchmark used in development from a genuinely unseen test. Any accessed prompt is contaminated for that developer and cannot be relabeled.

Heterogeneous evidence¶

Use technical reconstruction and failure analysis; PH argument and counterexample; HW source-packet comparison; EX unseen interpretation and complete-work relations; ML delayed transfer; DR changed briefs and observed user consequences. Include practical life tasks only within consent, competence and supervision boundaries.

All Souls specialist and General papers may contribute to some intellectual dimensions. They do not validate health practice, relational care, craft, ecological stewardship or longitudinal agency. Link official papers rather than redistributing copyrighted examination content without permission.

Administration and scoring¶

Separate independent acquisition/reconstruction, permitted augmentation and human defense. Keep tool access and assistance logs. Two qualified examiners score independently before adjudication, using the calibration protocol. Record disagreements, accommodations, attrition and invalid administrations.

Report raw denominators and uncertainty, not only average scores. Analyze failure as missing facts, prerequisite gaps, missing intellectual families, weak instruction, assessment ambiguity, unsafe practice or access barriers. Do not turn every failure into a new module or excuse every failure as learner effort.

Saturation and outcomes¶

Architecture freeze requires repeated heterogeneous rounds with no new load-bearing family under a declared decision rule. The rounds must use separately selected material and cannot be recycled after the curriculum has been changed for it.

Retest retained performance after an agreed delay and in changed contexts. A small pilot can expose implementation failures; it cannot establish population efficacy. Comparative or causal claims require an appropriate design and independent review. Longitudinal functioning remains sensitive to health, opportunity, institutions and choice. No study can establish a universal scalar maximum of human potential.