Con-C.2.4 — Fitness Analogy Breaks: No Natural Scalar for Civilizational Influence#
Severity: C (Serious) | Sphere: Se1, Se2 | Target: ax19, th6
The fitness analogy for ax19 succeeds for existence but fails for uniqueness. Evolutionary fitness has a natural scalar definition: \(w = E[\text{offspring in next generation}]\), with unambiguous outcome variable, biologically fixed time horizon, and nature performing the multi-dimensional projection. For ax19’s causal influence, no natural answers exist to four critical questions:
Time horizon? One year, one generation, one Jubilee cycle, eternity? Different horizons yield different h* values.
Probability distribution over futures? “The future of civilization” is a distribution over infinite-dimensional outcome space, with no canonical measure.
Discount rate? Must distant futures count equally?
Metric on world-histories? Without a metric, “total deviation in probability distribution” is undefined.
The measure-zero argument depends on scalar projection, but causal influence is a vector in infinite-dimensional space. Pearl’s do-calculus (2009) computes \(P(Y \mid \text{do}(X=x))\) for each outcome variable Y separately. “Total influence” across all Y-variables simultaneously is a vector, and the measure-zero argument does not apply without a specific projection.
The Tolstoy objection survives. If causal influence is power-law distributed, the gap between #1 and #2 may be vanishingly small while the aggregate influence of the bottom 99% vastly exceeds #1. Uniqueness is technically true but practically irrelevant.
Academic support: Pearl (2009), Causality, ch. 3: do-calculus defines intervention effects for specific outcomes, not “total influence.” Harsanyi (1955), JPE: aggregating utilities across individuals requires strong axioms — analogous axioms for causal influence are not stated. Gabaix (2009), Annual Review of Economics: empirically, social influence follows power-laws where the 1st–2nd gap is typically small.
(Source: C2.4 from OOv1 Critique Round 2.)