HCD.21:4.2 - Construct and compare feasible compositions

First exclude combinations that cannot perform the intended work under actual access and timing. Then compare the remaining options on the useful differences, common preparation, existing relationships and coordination cost. For a small cohort, an explained manual comparison can suffice. For a large one, an algorithm can propose groups from explicit criteria and weights; inspect the result against the original task and constraints.

Inspect the whole cohort as well as each attractive group. Moving one participant can improve a receiving group while removing a needed contribution from the group they leave. A cohort average or aggregate fit can hide that loss. Protect task-specific minimum inputs for every group before trading preferences or changing their weights.

When two desirable properties conflict, state the sacrifice. Keeping colleagues together may preserve confidential access while reducing cross-project variation. A later exchange of permitted abstractions can supply some variation without disclosing the project. Moving one prepared participant can add a needed perspective while avoiding complete regrouping. If every plausible group would require extensive preliminary teaching, compare a different task, a preparation branch or a supported individual route.

Keep continuing support relationships separate from exercise membership. A learner can retain a trusted mentor or study partner while changing partners for one comparison. The question is what each relation contributes, not which single group must contain all of learning.