Supervised Practice¶

Graduated practice process — instantiates Competence Calibration Feedback

A staged process in which a person performs real work under a supervisor's observation, earning independence one demonstrated case at a time rather than by assertion.

Supervised Practice develops and calibrates competence in the real context, with the downside capped. Its defining move is graduated release: the person does genuine work — not a simulation — under a supervisor who is at once the evidence source and the safety net, and autonomy is extended incrementally as observed competence accumulates, especially on progressively harder and less familiar cases. It differs from a Simulation or Case Test in that the work is real, and from Decision Rights by Competence in that it is the practice that produces the evidence, not the rule that grants the authority. Independence here is earned one demonstrated case at a time, never asserted.

Example¶

A newly graduated electrician doesn't wire a building alone on day one. They work under a licensed journeyman through a staged apprenticeship: at first, doing straightforward runs while watched closely; then, as the journeyman observes them handle real jobs correctly, taking on more independently. Panel work and anything with unusual load conditions still require sign-off — deliberately, because those are the unfamiliar, higher-stakes cases that test whether the apprentice's competence transfers beyond the routine circuits they've mastered. Each checkpoint is a real gate: the journeyman releases more autonomy only on evidence of correct, independent work on progressively harder tasks. The apprentice's own confidence is continually checked against what the supervisor actually observes, and the building never carries the risk of an unearned solo mistake.

How it works¶

Its distinguishing structure is real work plus live observation plus graduated release. The supervisor plays a double role — generating first-hand evidence of competence while catching failures before they cause harm — which is what lets a novice be exposed to real stakes safely. Autonomy is titrated upward through checkpoints, and the difficulty ramp is deliberate: progressively less familiar cases are introduced to test transfer before independence is granted on them. Learning and evidence-gathering happen together, in the context where the competence will actually be used.

Tuning parameters¶

Autonomy step size — how much independence each checkpoint releases. Big steps are efficient but risk unearned autonomy; small steps are safe but slow.
Observation intensity — over-the-shoulder versus sampled review. Intense observation catches more but constrains the trainee and consumes supervisor time.
Case-difficulty ramp — how quickly novel and harder cases are introduced. Faster probes transfer sooner; too fast overwhelms and teaches nothing.
Checkpoint criteria — what evidence a checkpoint requires before releasing more autonomy. Explicit criteria keep checkpoints real gates rather than formalities.
Safety-net responsiveness — how quickly the supervisor can intervene, which sets how much real risk the trainee can safely be exposed to.

When it helps, and when it misleads¶

Its strength is that it develops competence and gathers evidence in the real context while capping the downside, so autonomy comes to track demonstrated rather than asserted ability — a direct counter to unsafe overconfidence. The staged fading of support as competence grows is scaffolding in the developmental sense.^[1]

It misleads when the checkpoints go soft. Supervision can decay into rubber-stamp sign-off that certifies time served rather than competence; over-supervision can stunt the very independent judgment it's meant to build; and a difficulty ramp that never leaves the familiar case certifies a competence that won't transfer when it matters. The discipline is to tie every autonomy release to observed evidence on progressively harder cases, and to keep checkpoints as genuine gates rather than attendance milestones.

How it implements the components¶

Supervised Practice fills the real-context, observed-development slice:

performance_evidence_set — observed real-work performance is first-hand evidence of competence, gathered where the skill is actually used.
supervision_checkpoint — the review gate where accumulated evidence is judged and more autonomy is, or isn't, released.
transfer_context_probe — progressively harder, less familiar cases test whether competence generalizes before independence is granted on them.

It does not set the formal authority rule (decision_rights_boundary, Decision Rights by Competence) or supply the abstract standard being judged against (performance_benchmark, Competency Framework); it produces the observed evidence those consume.

Instantiates: Competence Calibration Feedback — develops competence and gathers evidence in the real context while capping the downside of unearned autonomy.
Consumes: the authority boundary from Decision Rights by Competence, which sets what each demonstrated level unlocks.
Sibling mechanisms: Decision Rights by Competence · Simulation or Case Test · Skills Assessment · Competency Framework · Benchmarked Feedback · Calibration Conversation · Confidence Rating Scale · Calibration Exercise · Exemplar Comparison · Peer Review · Reflective Error Log

Notes¶

Supervised Practice produces the observed evidence; Decision Rights by Competence sets the rule that turns it into authority. Keeping the two separate lets the practice ramp be tuned — faster observation, harder cases — without re-litigating the governance policy, and lets the policy be applied consistently across people whose practice paths differ.

References¶

[1] The zone of proximal development (Vygotsky) is the band of tasks a learner can do with guidance but not yet alone; "scaffolding" is the support a more capable other provides within it, faded as the learner becomes independent. Graduated supervised practice is that fading made into a structured process, with the supervisor's checkpoints controlling the rate. ↩