When One Support Bears Too Much

When a person or a path rests its steadiness on a single dominant support, the way it breaks reveals which support was lost, and it breaks hard. When steadiness is spread across several independent supports that each carry real load, the loss of any one rarely produces a clean break, and you can no longer read backward from the collapse to the missing support. So the familiar idea that you can diagnose where continuity was held by watching how someone fails is true only in the single-dominant-support case. Support concentration is the hinge that decides both how hard the fall is and whether the fall is legible: redundancy buffers the fall and blurs the diagnosis at the same time. Traditions that survive a long time tend to overdetermine their supports on purpose, which is why their failures are usually muddy rather than clean, and why their practitioners are hard to break by removing one thing.

The two halves are each well established on their own. That redundancy of supports protects against shock is standard across ritual theory, the sociology of plausibility structures, social-network diversity research, engineering defense-in-depth, and biological robustness. That an overdetermined outcome cannot be traced back to a single cause is the textbook structure of causal overdetermination and of redundancy in biology. The distinct move here is to make support concentration a single hinge that controls both facts at once, and to use it to bound a popular claim: you can read which support failed from how someone breaks only when one support carries most of the load. Under genuine near-equal support, removal is both survivable and non-diagnostic. This is narrower and more mechanical than calling the predictor vague or unfalsifiable: it names the exact regime where the prediction holds and the exact regime where it inverts, and it points to the one case (single dominant support) where breakdown is both most severe and most readable.

An adversarial check weakened the stronger version of this finding. The first instinct, that deliberate multiplexing defeats failure-type prediction outright, is refuted by Hakuin's Zen sickness: a multiplexing tradition still produced a clean effort-type collapse because concentrated striving was the dominant holder. So the inference-defeat applies only under near-equal-weight supports, and the result is a friendly amendment that strengthens the original predictor by specifying its valid regime, not a refutation that destroys it. Novelty is therefore modest, near 0.45, because both component ideas are mature and the same model family has already faced a falsifiability attack. The human side is also asymmetric: modern collapse tends to converge on the same downstream syndrome (depression, withdrawal, exhaustion, raised mortality) whether the lost support was a spouse, a role, or a source of worth, so even the legibility-under-dominance claim is shakier in people than the buffer claim. Contingent-self-worth research suggests domain specificity is partly readable, but the widowhood effect is confounded by anticipatory decline. Finally, Shinran's relocation of the single support to grace shows a single-locus design that the fragility claim does not touch, because the holder is never absent.