What does Bicameral AI Architecture conclude, in numbers, about “The Empirical Loop: Field Tools and Checker Specifications”?
Answer: Chapter 8 gives the bicameral ledger, the distilled Bicameral Error-Orthogonality Gain and State-Sync Packet Checker, a checker specification table grouped by kit, and a pre-registered test: lowering the leak from 0.030 to 0.015 needs 1,534 tasks per arm at power 0.80. Marked boxes separate published evidence, runnable Python calculations (23 programs), classical sources and the author's hypotheses.
Scope of the numbers: Its parameters are illustrative; all calculations are generated by its programs; "bicameral" is an engineering metaphor and no model is claimed to have awareness; the book is not personalised business or engineering advice. Marked boxes separate published evidence, runnable Python calculations (23 programs), classical sources and the author's hypotheses.
Data table
The numbers below are printed verbatim by the chapter's Python program and can be reproduced by running it. standard-library Python onlyprinted values that match the answer: 0.015, 0.030, 0.80, 1534
A. leak 0.030 (now) vs 0.015 (target); two-sided 0.05 power 0.80: 1,534 tasks per arm power 0.90: 2,053 tasks per arm halving a 0.010 leak to 0.005 at power 0.80: 4,673 per arm B. SPRT for H0 leak 0.015 vs H1 leak 0.030, alpha 0.05, beta 0.20 accept H1 when log-LR >= 2.7726; accept H0 when <= -1.5581 each leak adds +0.6931; each clean task adds -0.0153 expected tasks: about 284 if H0 holds, 323 if H1 C. register before looking: P0, P1, alpha, power, n, and the rule that the protocol is not changed before n is reached
Show the full program
# Chapter 8. p82: pre-registered sample sizes for the leak test.
# H0: the current set-up leaks at P0; H1: the heterogeneous set-up
# leaks at P1. Fixed-n two-proportion test (normal approximation)
# and Wald's sequential probability ratio test with its boundaries
# and approximate expected sample sizes. Rates are illustrative and
# must be replaced by the user's own baseline before registering.
import math
from statistics import NormalDist
ND = NormalDist()
P0, P1 = 0.030, 0.015
def n_two_prop(p0, p1, alpha=0.05, power=0.80):
za = ND.inv_cdf(1.0 - alpha / 2.0)
zb = ND.inv_cdf(power)
pb = (p0 + p1) / 2.0
top = (za * math.sqrt(2.0 * pb * (1.0 - pb))
+ zb * math.sqrt(p0 * (1.0 - p0) + p1 * (1.0 - p1))) ** 2
return math.ceil(top / (p0 - p1) ** 2)
def sprt(p0, p1, alpha=0.05, beta=0.20):
a = math.log((1.0 - beta) / alpha)
b = math.log(beta / (1.0 - alpha))
up = math.log(p1 / p0)
dn = math.log((1.0 - p1) / (1.0 - p0))
e0 = p0 * up + (1.0 - p0) * dn
e1 = p1 * up + (1.0 - p1) * dn
n0 = ((1.0 - alpha) * b + alpha * a) / e0
n1 = (beta * b + (1.0 - beta) * a) / e1
return a, b, up, dn, n0, n1
if __name__ == "__main__":
print("A. leak {:.3f} (now) vs {:.3f} (target); two-sided 0.05"
.format(P0, P1))
for pw in (0.80, 0.90):
print(" power {:.2f}: {:,} tasks per arm".format(
pw, n_two_prop(P0, P1, power=pw)))
print(" halving a 0.010 leak to 0.005 at power 0.80: {:,} per arm"
.format(n_two_prop(0.010, 0.005)))
a, b, up, dn, n0, n1 = sprt(P1, P0)
print("B. SPRT for H0 leak {:.3f} vs H1 leak {:.3f}, alpha 0.05, "
"beta 0.20".format(P1, P0))
print(" accept H1 when log-LR >= {:.4f}; accept H0 when <= {:.4f}"
.format(a, b))
print(" each leak adds {:+.4f}; each clean task adds {:+.4f}"
.format(up, dn))
print(" expected tasks: about {:,.0f} if H0 holds, {:,.0f} if H1"
.format(n0, n1))
print("C. register before looking: P0, P1, alpha, power, n, and the")
print(" rule that the protocol is not changed before n is reached")Single source and citation
Fan, Y. (2026). Bicameral AI Architecture: Fusing Long-Context Holographic Strategy Models with Local Deterministic Execution Agents (Version 1.0.0), Chapter 8: The Empirical Loop: Field Tools and Checker Specifications. 5D Library. DOI pending
DOI pending · monograph page goes live after publication · 5DL-B-410 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026410,
author = {Fan, Yafu},
title = {Bicameral AI Architecture: Fusing Long-Context Holographic Strategy Models with Local Deterministic Execution Agents},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}Other chapter answers in this book
- Chapter 1 · Correlated Errors: Where the Second Model's Gain Comes From
- Chapter 2 · Diversity Measures and the Ambiguity Decomposition: How Different Is Different Enough
- Chapter 3 · The Three-Stage Pipeline: Turning Errors into Intercepted Errors
- Chapter 4 · The Intent Translation Protocol: Slots, Ambiguity Entropy and Upper Bounds
- Chapter 5 · State-Sync Packets: Hash Chains, Version Vectors and Idempotency Keys
- Chapter 6 · Two-Way Correction: Gain, Noise and the Limits of Self-Correction
- Chapter 7 · Eastern and Western Classical Sources: Temple Reckoning, Breadth with Precision, and Friction