A turn in which a user deliberately hands a discrete cognitive sub-task, such as enumeration, drafting, or calculation, to an AI system to reserve attention for other reasoning, observable as an explicit delegation request.
One occurrence is coded per user turn whose primary act is delegating a bounded sub-task to the AI (an imperative request to produce, compute, or list something the user could do themselves), excluding clarification or evaluation turns.
Proposed measurement protocol (not yet empirically validated): Rater-coded delegation count per 10 user turns using a turn-level codebook (delegation vs. clarification vs. evaluation); inter-rater agreement could be reported as Cohen's kappa on a double-coded subset.
https://andreasehstandlicenseofclarityloc.github.io/augmanitai-periodic/#topic-cognition-thought