The most important theoretical insight of the project did not come from a library. It came from a conversation at Alpe di Siusi with a German tourist.
The claim under discussion had this geometry: A is false + B is true + the causal link A→B is undemonstrated. The specific formulation was: “The migrant emergency is causing the collapse of the healthcare system.” The manipulation lies in presenting A as a necessary or sufficient condition for B, when it is neither.
Validating this claim correctly required a distinction that the taxonomy did not yet make explicit: causal validation is structurally asymmetric. The framework can falsify an incorrect causal link. It cannot replace it with the correct one — doing so would reproduce the very epistemic problem it is designed to expose.
This is not a weakness. It is the epistemically honest response, and it is what distinguishes a rigorous tool from one that pretends to more than it can demonstrate (Hume, Popper, Pearl, Woodward).
The result was a three-outcome procedure for all D4.3 (Causal-Mechanism) claims:
- Falsified — the proposed link is absent, inverted, or inoperative
- Non-corroborated — the link cannot be established with available evidence
- Beyond framework scope — the claim requires external intervention not provided
This distinction was formalized as an Addendum to Coding Manual v1.0 (February 2026) and later integrated into Coding Manual v2.0 and Taxonomy v2.1.
Documents: Coding_Manual_Addendum_D4_3.docx, tassonomia_completa_v2_2.docx (v2.1 notes)