Epistemics

Scope

Epistemics defines how claims earn the status of knowledge within the initiative. Every modular in the Economic School — and any cross-domain work that touches economic reasoning — inherits these standards.

Axiomatics

1. Falsifiability as Entry Condition

A claim that cannot be tested, even in principle, has no standing in this framework. Every assertion must specify the conditions under which it would be wrong. Simulation results qualify as evidence only when the model's assumptions are stated explicitly enough to be challenged independently.

2. Evidence Pluralism

No single evidence type is sufficient. The field spans a transition that is partly historical, partly in-progress, and partly anticipatory. Valid evidence includes:

• Empirical observation: measured economic data, labor statistics, market behavior.
• Historical analogy: prior scarcity-shifts (mechanization, electrification, digitization) used as structural parallels, with the limits of the analogy stated.
• Formal models: mathematical and computational frameworks with explicit axioms and derivable consequences.
• Simulation output: agent-based, system-dynamic, or other computational experiments. Simulation demonstrates model behavior under stated conditions. The gap between model and real system must always be characterized.
• Case studies: documented instances of AI-driven economic transformation at firm, sector, or national scale.

Each type has known failure modes. Cross-validation across types is the baseline standard. Reliance on a single type requires explicit justification.

3. Assumptions Are Load-Bearing

Every model encodes assumptions. Those assumptions must be:

• Stated in full before results are presented.
• Traceable to specific axiomatics in the tier hierarchy.
• Tested for sensitivity - if small perturbations in an assumption produce large changes in output, the conclusions are conditional on that assumption holding.

Unstated assumptions invalidate a modular's conclusions within this framework.

4. Temporal Stratification

Claims about the automated economy operate on different time horizons, each carrying different epistemic weight:

• Observed dynamics (present): highest confidence. Testable against current data. Subject to standard empirical challenge.
• Trajectory claims (near-term extrapolation): moderate confidence. Testable as time-series data accumulates. Must specify what data would confirm or refute them, and by when.
• Structural claims (long-run, endpoint): lowest default confidence. Often testable only through simulation or formal derivation. Must be labeled accordingly. Structural claims gain confidence through convergence — multiple independent models or evidence types reaching the same conclusion.

Mixing temporal strata without acknowledgment is a category error.

5. Reflexivity Disclosure

AI systems are simultaneously the subject of study and instruments of analysis within this initiative. The tools used to model the automated economy are themselves products of the transition being modeled.

Any modular that uses AI-generated analysis, data synthesis, or simulation must disclose this and assess whether the AI's involvement biases the result. The operative question: would a materially different result emerge without the AI tool?

6. Physical Grounding

Claims about AI capability — what AI "can" or "will" do — require grounding in physical and computational constraints. Compute, energy, data, memory, interconnect bandwidth, and supply chain dynamics are real bottlenecks. Claims that treat AI capability as unbounded or costless fail this standard.

The same standard applies to claims about AI limitations, since the boundary of the feasible shifts with time. Current evidence is required in both directions.

7. Cross-Domain Coherence

The Economic School operates alongside Governance, Ethics, and other initiative domains. Claims here must be consistent with claims in adjacent domains at the same tier level. When a contradiction arises, it must be resolved at the lowest tier where both claims originate.

Two domains may reach different policy conclusions from shared axiomatics — that is expected. Conflicting axiomatics indicate a foundational error that must be traced and repaired.