Competing Hypothesis Matrix

Competing Hypothesis Matrix

Multi-hypothesis management — systematically generate alternative explanations competing with the primary hypothesis, design key predictions that can distinguish them, and build a structured comparison matrix to avoid confirmation bias.

Orchestration Intent

The most dangerous bias in scientific reasoning is holding only one hypothesis. This tactic forces CC to first systematically construct competing hypotheses before settling on the "best" hypothesis, then design predictions that can distinguish them.

The three steps cannot be reordered: first generate competing hypotheses (skipping not allowed), then design discriminating predictions (not allowed to only compare without testing), and finally build the comparison matrix (not allowed to only enumerate without quantifying). The final output is not "which hypothesis is correct" but "what experiment can distinguish them."

Available SOPs

Orchestration Pattern

Simplified (S tier, 1 primary hypothesis)

• Sequentially execute all 3 SOPs; generate ≥3 competing hypotheses; design ≥2 discriminating predictions; build comparison matrix
• Applicable: a single primary hypothesis requiring competitive-thinking scrutiny

Standard (M tier, 2-3 primary hypotheses)

• competing-hypothesis-generation generates competing hypotheses independently for each primary hypothesis; discriminating-prediction-design designs discriminating predictions across primary and competing hypotheses; hypothesis-comparison-matrix includes all hypotheses (primary + competing)
• Applicable: multiple candidate hypotheses already exist, requiring unified management and comparison

Deep (L tier, complex hypothesis set)

• All 3 SOPs execute; competing-hypothesis-generation additional requirement: at least 1 competing hypothesis comes from a completely different theoretical framework; discriminating-prediction-design additional requirement: each discriminating prediction annotates the required experiment scale and difficulty; hypothesis-comparison-matrix additional output: recommended experiment priority (most discriminating predictions ranked first)
• Applicable: a complex hypothesis space requiring direct input to experiment design

Minimum Yield

• ≥3 competing hypotheses (explaining the same phenomenon as the primary hypothesis but with different mechanisms)
• ≥2 discriminating predictions (each prediction produces different expected outcomes for at least 2 hypotheses)
• Structured comparison matrix: hypotheses × predictions, each cell annotating expected outcome direction (support/contradict/irrelevant)

Yield Report