Critique: The Simulation Tautology (Circular Validation)

Critique and response context for The Simulation Tautology (Circular Validation) in AffineDrift’s control-affine golf-swing framework.

Critique: The Simulation Tautology (Circular Validation)

Summary of Concern

The manuscript (specifically Part V and the Simulink documentation) claims that the numerical equality \(F_{\text{total}} - F_{\text{ZTCF}} = F_{\text{input}}\) observed in simulation serves as a “strong validation” of the theory. This is a logical fallacy. Since the simulation is built using the exact same equations of motion (\(\dot{x} = f(x) + g(x)u\)) as the theory, this result validates only the algebraic consistency of the code, not the physical validity of the AffineDrift framework. It proves the math was typed correctly into MATLAB, not that the math represents reality.

Location

  • Page: articles/theory-part5.qmd (Simulink Model) and articles/affine-nature-golf-swing.qmd (Section 5, Numerical Computation).
  • Claim: “A key empirical result emerged: … matching the analytical identity predicted by the force taxonomy. This serves as a strong validation of the drift–input decomposition…”
  • Section: “Zero Velocity Counterfactual computation and validation”

Nature of the Issue

  • Logical Circularity: The author uses a model constructed from the theory to validate the theory.
  • Confirmation Bias: The simulation is guaranteed to satisfy the affine decomposition by definition (unless coded with bugs).
  • Overstated Significance: Framing a unit test (algebraic verification) as a scientific discovery (“empirical result”).

Why This Is a Problem

  1. Undermines Credibility: A sophisticated reviewer will immediately spot that the “validation” is tautological. It suggests the author confuses “internal consistency” with “external validity.”
  2. False Confidence: It implies the theory has passed a stress test when it has only passed a syntax check.
  3. Distraction from Real Issues: It draws attention away from the actual verification challenges (parameter sensitivity, noise, unmodeled dynamics).

Evidence / References

  • Oreskes, N., et al. (1994). “Verification, validation, and confirmation of numerical models in the earth sciences.” (Science). Models can only be confirmed, not validated; internal consistency does not imply truth.
  • Roache, P. J. (1998). “Verification and Validation in Computational Science and Engineering.” (Distinction between verification (solving equations right) and validation (solving the right equations)).

Severity

  • Medium: It frames a necessary sanity check as a major scientific result. It requires reframing, not identifying a fatal flaw in the math itself.

Suggested Remedies

  1. Rephrase “Validation” to “Verification”: Explicitly state that the Simulink model verifies the algebraic correctness and numerical stability of the subtraction, rather than validating the physical theory.
  2. Clarify the Purpose: Frame the simulation as a “Numerical Stress Test” for the algorithms (integration, discrete sampling, kill-switch logic), ensuring they don’t break under stiff dynamics.
  3. Remove “Empirical Result” Language: Do not call the identity \(F_{tot} - F_{ZTCF} = F_{in}\) an empirical finding. Call it a “Numerical Confirmation of Identity.”
  4. Acknowledge the Tautology: Add a note: “Since the simulation enforces the control-affine structure, this equality is expected by design. Its value lies in demonstrating that discrete-time integration errors do not corrupt the causal subtraction.”