Transfer Claim Checklist

This is a validation tool, not a reading assignment. It takes five minutes. Any claim that transfers a principle from a canonical model to a new domain must pass all five steps. If it cannot, the claim is not ready to make.

Print this page, hand it to a colleague, or run it mentally before committing an emergence-based argument to a slide deck, a paper, or a policy recommendation.

The Five Steps

1. Name the model. Which canonical model are you invoking? State its setup in one sentence (what the agents are, what the interaction structure is) and its rule in one sentence (how agents update their state). If you cannot do this, you are working from an impression, not a model.

2. Name the mechanism. What specific causal chain operates in the target domain? “Emergence” is not a mechanism. Name the actual one: threshold cascade, symmetry breaking, preferential reinforcement, nonlinear congestion, stigmergic coordination, delay-wave propagation. The mechanism must be something that can, in principle, be measured or disrupted.

3. State what transfers. What domain-independent principle applies? State it precisely enough that someone could test it. “Complex behavior arises from simple rules” is not precise enough. “When agent density exceeds a critical threshold, delay waves propagate backward through the system” is.

4. State what does not transfer. Where does the analogy break? What assumptions of the canonical model are violated in the target domain? Every model simplifies. Name the simplifications that matter: homogeneous agents when the real population is heterogeneous, synchronous updates when real interactions are asynchronous, complete mixing when real contact networks are structured.

5. State what would falsify the claim. What observable evidence would disprove the transfer? If you cannot name a falsifier, the claim is not precise enough to be useful. A falsifier must be specific: not “if the system behaves differently” but “if the distribution of event sizes follows an exponential rather than a power law” or “if reducing the contact rate below the computed threshold does not halt spread.”

Worked Example: PASS

Claim: Clinical practice adoption follows SIR epidemic dynamics.

  1. Model: SIR/Epidemic. A population partitioned into susceptible, infected, and recovered compartments. Individuals move S to I upon effective contact with an adopter, I to R upon full adoption (no longer actively persuading peers). Rule: spread rate is proportional to the product of adopters and non-adopters.
  2. Mechanism: Threshold contagion. Each clinician’s probability of adopting a new practice increases with the number of peers who have already adopted. Spread is contact-dependent and self-limiting as the susceptible pool depletes.
  3. What transfers: There exists a critical threshold analogous to R_0 = 1. Below it, adoption dies out locally. Above it, adoption curves follow the characteristic shape: exponential rise, peak, decline. Interventions that target high-contact clinicians (opinion leaders) have disproportionate effect, analogous to targeted vaccination.
  4. What does not transfer: Real clinicians are not homogeneous. Adoption is not binary — there are partial adopters, backsliders, and resistors. The contact network is structured (departments, hospitals, specialties), not well-mixed. Institutional mandates can force adoption independent of peer contact, which has no analogue in the SIR model.
  5. Falsifier: If adoption curves in a measured population show no sensitivity to the fraction of current adopters — if adoption proceeds at a constant rate regardless of peer exposure — then the contagion mechanism is not operative, and the SIR transfer fails.

Verdict: Pass. The claim is specific, the mechanism is named, the limits are stated, and the falsifier is testable.

Worked Example: FAIL

Claim: “The economy is a cellular automaton.”

  1. Model: Cellular automata (unspecified — Conway’s Life? Wolfram’s Rule 110?). No specific CA is named. Setup and rule are not stated.
  2. Mechanism: None named. The claim implies that economic agents follow local rules and produce emergent macro behavior, but no specific causal chain is identified. What is the rule? What is the neighborhood? What is the state space?
  3. What transfers: Unstated. “Simple rules produce complex behavior” is not a transferable principle — it is a restatement of the definition of emergence.
  4. What does not transfer: Not addressed.
  5. Falsifier: None offered. The claim is unfalsifiable because it is not specific enough to generate predictions.

Verdict: Fail. The claim stalls at Step 1. It is a metaphor, not a transfer. It may be suggestive, but it cannot do analytical work in its current form.

Common Failure Patterns

Analogy without mechanism. The target system “looks like” the model, but no causal chain is specified. Resemblance is not transfer. Two systems can produce similar-looking outputs through entirely different mechanisms.

Metaphorical transfer claimed as formal. The language of the model is applied to the target domain without checking whether the formal conditions hold. Calling a market a “sandpile” because it has cascading failures does not establish that the market satisfies the conditions for self-organized criticality.

Unfalsifiable claims. If no observation could disprove the transfer, the claim has no empirical content. This is the most common failure in popular writing about emergence — and the most dangerous in policy contexts.

Descriptive overclaim. A model that describes an observed pattern is presented as predicting future behavior. Descriptive fit is necessary but not sufficient. The SIR model fits many adoption curves, but fitting a curve is not evidence that the SIR mechanism is operating — other mechanisms produce similar curves. Prediction requires specifying the mechanism and testing it under intervention.

For a full treatment of these failure modes, see Critiques and Failure Modes.