======================================================================
ARCHITECTURAL RATIO BOOTSTRAP (Section 7.7)
======================================================================

Pre-specified target ratio: theta_4 / theta_2 = 96/34 = 2.8235
Trials per prior:           N = 100,000
Random seed:                2026
Akivi observation:          L_ahu/L_plaza = 70/25 = 2.8000
  Relative error vs target: 0.833%

Prior                  tol 1.00%     tol 0.83%     tol 0.50%
----------------------------------------------------------------------
A (broad)                0.810%         0.657%         0.404%  
B (narrow)               1.125%         0.938%         0.588%  
C (empirical)            0.908%         0.750%         0.455%  

INTERPRETATION
----------------------------------------------------------------------
Across all three priors, the probability of obtaining the framework
ratio 96/34 within 1% by random pairing of platform/plaza lengths is
on the order of 0.7%-1.0%. The Akivi observation (error 0.83%) is
therefore not compatible with arbitrary dimensioning of the complex.

The result is robust to prior choice within the range of plausible
Polynesian platform/plaza dimensions: broad-uniform, narrow-uniform,
and empirical-anchored priors all yield comparable probabilities.

Caveat: this bootstrap tests the probability of the ratio match by
chance, given the architectural typology of platform vs plaza. It
does NOT test whether the typological identification itself
(L_ahu as total horizontal extent, L_plaza as frontal sub-extent) is
post-hoc. That identification is logically pre-specified by the
standard architecture of Polynesian ahu complexes (Mulloy & Figueroa,
1978; Hamilton et al., 2011) and is not derived from this analysis.
