Apply Best Practice Tree

Recursively apply the best-matching skill to a problem, match each resulting sub-problem to a best practice, and repeat until every sub-problem is covered or fully resolved.

Why This Is Best Practice

Adopted by: Divide-and-conquer is foundational in computer science (Knuth, 1968) and is the basis of virtually every structured problem-solving system — used by Google's SRE incident analysis, McKinsey's issue-tree methodology, and NASA's fault-tree analysis. Case-Based Reasoning (CBR), the formal basis for "use the most similar practice when no exact match exists," is deployed in IBM's Watson, medical diagnosis systems (Aamodt & Plaza, 1994), and legal precedent reasoning. Impact: Aamodt & Plaza (1994) demonstrated that CBR reduces problem-solving time by 40–60% in domains with partial-match knowledge bases, by reusing prior solutions rather than solving from scratch. Divide-and-conquer reduces cognitive load by bounding each decision to a tractable sub-problem — the primary mechanism behind McKinsey's documented success with MECE issue trees (Rasiel, The McKinsey Way, 1999). Why best: Flat skill matching — "find one skill for the whole problem" — fails on complex problems because no single skill covers all dimensions. MECE decomposition (the alternative in plan-best-practice-solution) requires full domain expertise upfront before any skill is applied, which is unavailable when the problem space is new. Recursive skill decomposition uses the skills themselves as the decomposition engine, allowing the knowledge base to guide the breakdown without requiring prior expertise. It also degrades gracefully: when no exact skill exists, it applies the closest match with an explicit adaptation note rather than halting.

Sources: Knuth, The Art of Computer Programming (1968); Aamodt & Plaza, "Case-Based Reasoning: Foundational Issues" (1994); Rasiel, The McKinsey Way (1999)

Steps

Step 0: Problem clarity check (silent)

Before classifying, check if the problem is defined enough to decompose: