JSIFS Methodology Group · 2026-05-06
Long-running programmes accumulate explanations faster than they retire them. Compression addresses the accumulation directly: enumerate candidates, merge those that predict identical observables, and require a disconfirming observation as the condition for staying in the set.
Two rules keep the exercise honest. Conventional instrumentation causes are enumerated first, because a set that begins with exotic mechanisms will rank them relative to each other rather than against the mundane baseline. And any candidate without a feasible disconfirming measurement leaves the ranked set for the speculative register — retained, visible, but not competing for resources.
The resulting rank order expresses discriminating power and measurement feasibility. It is not a probability, and presenting it as one inverts the purpose of the exercise.