Deterministic and Infeasible (D-I)
Task nature where a correct answer exists but exact computation is impractical at scale
Deterministic and infeasible (D-I) is a task-nature class: a theoretically correct answer exists, but computing it exactly is impractical at the scale, latency, or resource budget the organisation needs.
Examples: large-scale combinatorial optimisation, vast search spaces, computationally hard planning, routing at national scale with hard constraints.
D-I is not P. The answer is not inherently unknowable — it is expensive or slow to derive exactly. Approximation, heuristics, or validated domain-specific models (VM) may be warranted — with outcome severity governing how much human oversight and calibration the bind requires.
Organisational implications:
- At catastrophic or significant severity: validated models plus mandatory human at the commitment boundary for high-stakes cells — see the calibration matrix in article 19.
- Rule 1 applies to D-F at execution only — do not treat D-I as licence for a general LLM where a feasible deterministic refactor exists.
- Contested D-I (experts disagree whether the problem is truly infeasible vs unpublished D-F structure) overlaps false uncertainty — classify before installing collective machinery.
Sibling classes: D-F, P. Full matrix: Decision Framework map.