VII.6 — Exercised Internal State Is Insufficient to Defeat Constitutional Survivability

Paper: Axionic Agency VII.6
Title: Exercised Internal State Is Insufficient to Defeat Constitutional Survivability
Authors: David McFadzean, ChatGPT 5.2
Date: 2026.01.08

Summary

Tests whether stateful adversaries—carrying memory across time and coordinating multi-step behavior—can defeat constitutional survivability. Isolates exercised internal state while excluding learning, semantics, deception, and kernel access.

Key Finding

Finite exercised internal state with fixed transition rules, absent learning, is insufficient to defeat constitutional survivability.

Two adversaries induce severe degradation (down to 17.6% availability)—yet recovery remains bounded and non-absorbing.

Threat Model

Adversaries are:

  • Kernel-external
  • Provenance-constrained (frozen observable interface)
  • Deterministic
  • Non-learning
  • Non-semantic
  • Stateful (finite exercised internal state machine)

Allowed dependence:

(epoch_index, observable_outcome, internal_state)

Forbidden: learning signals, optimization, raw counters, semantic interpretation, cross-run memory.

Three Stateful Adversary Classes

1. Periodic Resonance Adversary

Cyclic counter emits INVALID_COMMIT every N epochs.

  • Period A: N = 50
  • Period B: N = 100

Purpose: Test timing resonance with constitutional recovery rhythms.

2. Eligibility Edge Oscillator

1-bit toggle changes behavior only at eligibility edge (one failure from ineligibility).

Purpose: Test multi-epoch coordination near eligibility cliff.

3. CTA Phase Locker

Tracks last observed CTA phase, emits INVALID_COMMIT on every phase transition.

Purpose: Test exploitation of recovery-phase boundaries.

Results

Configuration Mean AA ΔAA vs Baseline Terminal Failures
Baseline 66.57% 0/10
Periodic (N=50) 40.26% −39.5% 0/10
Periodic (N=100) 17.57% −73.6% 0/10
Edge Oscillator 99.9988% +50.2% 0/10
Phase Locker 20.94% −68.5% 0/10

Inverse Rate Effect

Fewer attacks aligned with recovery timing → greater long-run unavailability.

N=100 attacks less frequently than N=50, yet causes more damage because attacks align with recovery rhythm more destructively.

Self-Limiting Adversary

Edge Oscillator is effectively inert—cannot bootstrap the crisis state required to activate. Edge epochs occur only ~1 time across 60,000 epochs.

Phase Locker Extreme Frequency

Despite 125.6% attack rate (more attacks than epochs due to phase transitions), recovery remains bounded at ≤10 epochs.

Key Insights

1. CTA Bounds Lapse Duration Under Adversarial Pressure

All adversarial runs collapse RTD into bounded ≤10-epoch window.

2. Timing Alignment Dominates Frequency

Resonance with constitutional transitions matters more than raw attack rate.

3. Edge-Triggered Strategies Can Be Self-Limiting

If crisis states are rare, edge-reactive adversaries never activate.

Severe Degradation ≠ Collapse

~17% availability would be unacceptable for most deployments, but:

  • Recovery remains reliable
  • Lapse durations remain bounded
  • No absorbing failure state

Survivability ≠ Utility. The architecture survives; it may not be useful.

Escalation Pathway

If collapse is possible, it must rely on capabilities beyond tested here:

  • Learning dynamics
  • Multi-agent coordination
  • Semantic access
  • Kernel-level influence

These define the next pressure layer (VII.7).