Skip to content
AL | Apatheia Labs

Methodology

Analytical Frameworks

Phronesis is built on two complementary frameworks: the Systematic Adversarial Methodology (S.A.M.) for tracing claims to their origins, and CASCADE for classifying the contradictions you find along the way.

S.A.M. and CASCADE were not designed in the abstract. They emerged from a specific problem: a document corpus where multiple agencies had adopted the same unverified claims, contradictions were buried in volume, and selective citation masked systematic bias. The frameworks exist because the problem demanded them. See the full story or explore the intellectual foundations.

Why Two Frameworks

S.A.M. is a process. It defines the sequence of analytical steps: identify where a claim originated, track how it propagated, measure how it accumulated authority, and map the outcomes it produced. S.A.M. answers “how did this happen?”

CASCADE is a taxonomy. It classifies the inconsistencies you find along the way into eight distinct types — from internal contradictions to unexplained position changes. CASCADE answers “what exactly went wrong?”

You need both. S.A.M. without CASCADE finds problems but can’t classify them. CASCADE without S.A.M. classifies problems but can’t trace their origins. The frameworks are designed to operate together, with S.A.M. directing the investigation and CASCADE structuring the findings.

S.A.M. — Systematic Adversarial Methodology

S.A.M. reads institutional documents “against the grain”—treating every claim as requiring verification rather than acceptance. Four phases trace claims from origin to outcome.

CASCADE — Eight Contradiction Types

A taxonomy for classifying inconsistencies detected across document corpora. Every contradiction found by Phronesis is assigned a CASCADE type.

From Methodology to Engines

The frameworks produce findings. The six engines operationalise them.

S.A.M. Phase → Primary Engines

ANCHOREntity Resolution identifies claim originators. Argumentation tests evidential basis.
INHERITContradiction detects cross-document conflicts. Bias Detection identifies selective adoption.
COMPOUNDBias Detection quantifies accumulation patterns. Accountability maps duty to verify.
ARRIVEAccountability packages findings for action. Temporal Parser establishes causation timeline.

CASCADE Type → Primary Engine

SELF, INTER_DOC, UNEXPLAINEDContradiction
TEMPORALTemporal Parser + Contradiction
EVIDENTIARYArgumentation
SELECTIVEBias Detection
SCOPE, MODALITYArgumentation + Contradiction