ASP-PROTO-01 · v0.1.0-protocol · Pre-Empirical · No Claims Made

Adaptive Survival
Protocol

A Measurement Protocol for Survival Under Model–Reality Divergence
Revision, Action, and Survival Under Conflict

"Survival under model–reality divergence is not explained by the
absence of divergence, but by the capacity to revise and act under it."
7
Variables
3
Candidate Mechanisms
4
Falsification Levels
7
Test Environments
3
Nested Models
25
Frozen Files

Distinguishing Survival from Prediction Accuracy

A widespread intuition is that systems survive model–reality divergence because they predict well. Under this view, revision is merely a correction of prediction errors. The protocol proposes a different framing: prediction accuracy is one candidate explanation, but not the only one. Survival may depend on the capacity to revise and act under divergence.

MECHANISM 1

Model Revision (R^K)

The system changes the structure of its internal model, not only its surface values. A structural change that reduces objective conflict and persists beyond a pre-specified interval.

MECHANISM 2

Policy Revision (R^Ω)

The system changes its policy, goal, or action space. Model revision and policy revision are distinct processes, and they may fail independently.

MECHANISM 3

Action Capacity (A^cap)

The system has a viable path, sufficient resources, and authority to execute. Capacity is not the same as outcome: a system may possess full capacity and still fail due to environmental interference.

What the Protocol Changes

The protocol separates variables that are often conflated, specifies conditions of possibility that are often ignored, and pre-specifies all decision rules before data collection.

Outcome ≠ Mechanism

Survival is not defined by revision or action. Revision and action are candidate mechanisms whose contribution to survival is to be tested.

Capacity ≠ Opportunity

Opportunity structure (signal, authority, execution) is a condition of possibility, not a mechanism. Absence of opportunity is not a systemic failure.

Capacity ≠ Outcome

A^cap and A^out are distinct. A system may possess full capacity and still fail due to environmental interference.

Failure ≠ Binary

Failure is classified by the first layer that fails: detection, revision, action, or environment. Each has different implications.

Pre-Empirical · No Claims Made

Conceptual architecture Frozen
Central hypothesis (H0 / H1) Frozen
Variables and operational definitions Frozen
Causal pipeline Frozen
Measurement and thresholds Frozen
Experimental design Frozen
Statistical models (M0, M1, M2) Frozen
Falsification criteria Frozen
Structural audit (153 checks) Passing
Model contract checks (54 checks) Passing
Empirical data Not collected
Simulation Not run
Statistical results Not available

View Protocol Status

The dashboard shows the current status of the protocol: frozen layers, audit results, and what is not yet available. No empirical values are fabricated.

Open Dashboard →