HCD.7:11 - SoTA-Echoing

The working question is: How should a practitioner establish that the people, access, tools, AI, and other support needed by a development use are actually available at the required competence, time, capacity, and consequence boundary?

The selected best-known line treats support as a configuration of distinct provider, access, coordination, and checking contributions rather than as a provider label. Brooks and colleagues’ 2024 qualitative study of learning-health-system communities (SOC-05) makes people, platform, facilitation, leadership, and resource contributions separately visible; it supplies arrangement questions, not a causal capability effect. Matsuo and Tsukube’s 2020 cognitive-apprenticeship review (SOC-02) supports distinguishing modelling, coaching, scaffolding, articulation, reflection, and exploration while leaving their sequence and workplace transfer conditional. HCD.7 adapts those contributions into the allocation and service tests in §§4.2–4.5.

The serious ordinary alternative is label-and-total planning: name a teacher, mentor, community, apprenticeship, or platform, total the available hours, and assume the functions will be supplied. It is cheaper to plan, but it hides competence, individual work, peak load, access, privacy, assessment independence, and fallback. HCD.7 rejects the provider label and aggregate total as sufficiency evidence while retaining them as inputs to a complete contribution account.

Tullis, Goldstone, and Hanson’s 2015 experiment (COG-11) is direct limit evidence that assistance improving practice can impair later unassisted performance and that support scheduling matters in its laboratory motor task; it does not prescribe a universal fading schedule or occupational arrangement. Wiese, Kilty, and Bennett’s 2018 clinical realist synthesis (SOC-03) shows that supervised participation and graduated access depend on task risk, trust, learner position, and service pressure; its clinical governance does not transfer wholesale. Dell’Acqua and colleagues’ bounded field experiment (AI-03) supports independent error challenge under specific task and model conditions, not a stable universal AI frontier. These limits produce the support-state, human-check, and changed-support boundaries in §§4.3–4.7.

The selected line costs more coordination than naming providers and adding hours, but it preserves the exact result, peak feasibility, authority, privacy, human judgement, and fallback that the cheaper plan loses. Reopen it when a provider, task, capacity window, access rule, AI condition, evidence use, or serious source changes, or when a lower-effort arrangement test preserves the same decision-bearing values.