Theory Metrics Overview & Taxonomy¶
This document provides a consolidated taxonomy, mathematical specification, and overlap analysis of all metascientific, structuralist, topological, coherence, and evolutionary metrics used to evaluate reconstructed theory graphs in Episteme.
The metrics are conceptually and mathematically organized into five foundational pillars, underpinned by a global connectionist relaxation engine.
Click on any metric title in the tables below for its dedicated formulation, algorithmic implementation, and metascientific diagnostic guidelines.
Central Algorithmic Engine¶
Some metrics are calculated explicitly via targeted graph algorithms, while others are resolved dynamically through a global relaxation model.
| Metric | Definition & Conceptual Goal | Measurement Approach | Specific Algorithm, Formula, or Mathematical Scope |
|---|---|---|---|
| The LPG-ECHO Algorithm [4, p. 435–467] | Central connectionist engine for parallel constraint satisfaction, resolving belief propagation, conflict resolution, systemic decay, and global coherence. | Executes synchronous / asynchronous relaxation loops across the multi-relational property graph until convergence. | Assigns continuous epistemic status \(a_i^* \in [-1, 1]\) to every node. Updates: \(a_j(t+1) = a_j(t)(1-\theta) + \text{net}_j(t)(a_{\max} - a_j(t))\) for \(\text{net}_j > 0\). Intrinsic decay half-life: \(t_{1/2} = \frac{\ln(0.5)}{\ln(1-\theta)} \approx 13.5\) iterations at \(\theta = 0.05\). |
Pillar 1: Structural and Topological Metrics¶
Evaluates the physical graph layout, hierarchy, community segmentation, bottlenecks, and topological robustness of the local intra-theory graph (theory-net) or global inter-theory graph (theory-holon).
Organized along a Four-Dimensional Taxonomy:
Macro-Topology (Global Cohesion & Economy)¶
| Metric | Definition & Conceptual Goal | Measurement Approach | Specific Algorithm, Formula, or Mathematical Scope |
|---|---|---|---|
| Structural Elegance Index [2, p. xviii], [5, sec. 2.2], [6, ch. 14] | Composite Index: Evaluates whether a theory achieves robust Quinean global cohesion while remaining economically formulated without redundant claims. | Combines spectral algebraic connectivity with penalized graph density. | \(E(G) = \lambda_2(G) - \alpha \cdot \rho(G)\), where \(\lambda_2(G)\) is the Fiedler value (algebraic connectivity) and \(\rho(G) = \frac{2\|E\|}{\|V\|(\|V\|-1)}\) is graph density. |
| Connectedness & Cohesion [2, p. 173] | Foundational Input: Ensures a theory-net maintains a unified logical lineage rather than fragmenting into anarchical disjoint components. | Evaluates reachability via Weakly Connected Components (WCC) and algebraic connectivity (\(\lambda_2\)). | Graph reachability check verifying that any two distinct nodes share a common ancestor or specialization path; Laplacian spectrum Fiedler eigenvalue \(\lambda_2 \ge 0\). |
| Modesty [5, sec. 5.1], [6, ch. 6] | Foundational Input: Evaluates how constrained or bold a theory is by penalizing over-connected, dense "hairball" assertions. | Computes the inverse edge density of the hypothesis subgraph. | Inverse edge density \(M(H) = \rho(H)^{-1} = \frac{\|V_H\|(\|V_H\|-1)}{2\|E_H\|}\). |
Micro-Topology (Vulnerability & Bottlenecks)¶
| Metric | Definition & Conceptual Goal | Measurement Approach | Specific Algorithm, Formula, or Mathematical Scope |
|---|---|---|---|
| Bottleneck Fragility Profile [1, sec. 5.1], [5, sec. 3.2] | Composite Diagnostic: Evaluates whether a theory graph is over-reliant on isolated central dogmas, and quantifies the structural collapse if those hubs fail. | Unifies degree distribution entropy, vertex connectivity threshold (\(\kappa\)), and betweenness-based shortest path drop. | Diagnostic profile combining \(H_{\text{node}}\) (degree entropy), \(\kappa(G)\) (absolute failure threshold), and \(R_k(G)\) (refutability impact). |
| Structural Homogeneity [1, sec. 5.1], [5, sec. 3.2] | Foundational Input: Assesses whether structural components and degrees are uniformly distributed or skewed into star-network bottlenecks. | Calculates the normalized Shannon entropy and Gini variance of the node degree sequence. | Normalized Degree Entropy \(H_{\text{deg}} = -\frac{1}{\ln \|V\|} \sum p(k) \ln p(k)\) and degree variance \(\sigma_{\text{deg}}^2\). |
| Structural Refutability [5, sec. 5.1] | Foundational Input: Quantifies topological fragility by measuring the destruction of shortest path volume when top betweenness hubs are removed. | Simulates removal of top-\(k\) betweenness centrality nodes and measures path drop against vertex connectivity. | Hard falsifiability bound (\(\kappa\)-connectivity) and Top-\(k\) Refutability Score: \(R_k(H) = \frac{\|\Pi(H)\|}{\|\Pi(H)\| + \sum_{i=1}^k \|\Pi(H \setminus \{v_i\})\|}\). |
Meso-Topology & Socio-Epistemic Alignment¶
| Metric | Definition & Conceptual Goal | Measurement Approach | Specific Algorithm, Formula, or Mathematical Scope |
|---|---|---|---|
| Community-Bridges (Inter-Theoretical Links) [2, pp. 224, 317, 324] | Evaluates whether cross-community links between sub-disciplines represent genuine theoretical translations (constraints) or ad-hoc associations. | Computes edge betweenness centrality over cross-partition boundaries generated by community detection. | Edge betweenness centrality on cross-community edges \(e \in E_{\text{cross}}\). Flagged if betweenness is high but formal constraints are absent. |
| Modularity & Clustering [5, sec. 5.1], [6, ch. 14] | Foundational Input: Identifies the emergence of dense sub-networks or specialized theoretical domains within the global graph. | Executes partition optimization over the multi-relational graph. | Newman-Girvan modularity optimization (\(Q\)) or Leiden/Louvain community detection algorithm. |
| Centrality & Argumentative Discrepancy [1, p. 119], [5, sec. 5.1] | Evaluates the systemic importance of hypotheses and flags discrepancies between structural network necessity and textual discourse emphasis. | Compares PageRank / Eigenvector centrality against NLP-extracted Argument-Core-Scores from source literature. | Eigenvector centrality \(x_v = \frac{1}{\lambda} \sum A_{uv} x_u\) vs. NLP discourse prominence; discrepancy \(\Delta_{\text{disc}} = \|C_{\text{structural}} - C_{\text{textual}}\|\). |
Logical Integrity Constraints (Prerequisites)¶
| Metric | Definition & Conceptual Goal | Measurement Approach | Specific Algorithm, Formula, or Mathematical Scope |
|---|---|---|---|
| Non-Cyclic Form (DAG Property) [2, pp. 172–173] | Enforces logical foundational stability and prevents circular reasoning in specialization and derivation hierarchies. | Identifies directed loops within the intra-theory derivation/specialization subgraphs. | Tarjan's Strongly Connected Components (SCC) or DFS topological sort. Requires all intra-theory derivation subgraphs to be Directed Acyclic Graphs (\(\|SCC_i\| = 1\)). |
| Tree / Hierarchical Conformity [2, p. 175] | Verifies that a theory-net strictly conforms to a single-rooted hierarchical branching structure (theory-tree). | Measures in-degree orientation and identifies the set of top-level minimal elements \(B(N)\). | Verifying that the set of minimal elements \(B(N)\) in the specialization poset is a singleton: \(B(N) = \{T_0\}\). |
Pillar 2: Epistemic Coherence & Parsimony Metrics¶
Evaluates internal logical consistency, parsimony, semantic clarity, and constraint-satisfaction harmony using connectionist coherence and model-theoretic principles.
| Metric | Definition & Conceptual Goal | Measurement Approach | Specific Algorithm, Formula, or Mathematical Scope |
|---|---|---|---|
| Simplicity (Einfachheit) [1, sec. 3.12, 5.11.2], [4, p. 437], [5, sec. 2.2] | Evaluates inferential parsimony on local explanation chains and global topic entropy, resolving Hempel's Conjunction Problem via Minimum Description Length (MDL). | Scales relation weights inversely by the intrinsic antecedent complexity \(c(a_i) = \|Ce(a_i)\|\) and computes universe simplification rate. | Local relation weight \(\phi(a_i, b) = \frac{\phi_{\text{base}}}{\sum c(a_j)}\), chain simplicity \(SR_{\text{MDL}} = \frac{1}{\sum c(a_j)}\), and universe topic entropy simplification \(S_{\text{global}}(H) = \frac{E(G \setminus H)}{E(G)}\). |
| Elegance and Economy [2, p. xviii], [1, sec. 5.2], [7] | Evaluates the macro-parsimony of formal theory-elements \(T = \langle K, I \rangle\), penalizing axiom and parameter bloat via practical two-part MDL. | Assesses axiomatic cardinality, parameter overhead, and residual data compression cost. | Axiom Parsimony Score \(APS(T) = \frac{1}{1 + \log_2(1 + \|Ax(T)\|) + \log_2(1 + \|\Theta(T)\|)}\), and Practical MDL: \(MDL(T) = L(M) + L(D \mid M)\), where \(L(M) = w_n\|A_{\text{core}}\| + w_e\|R_{\text{core}}\|\). |
| Tenability (Haltbarkeit) [3, pp. 120, 270], [1, sec. 2.3] | Evaluates whether empirical observations (\(I \subseteq M_{pp}\)) can be theoretically enriched to satisfy fundamental laws and cross-application constraints under admissible blurs. | Dual-enrichment architecture (\(\Phi_{\text{spec}} \circ \Phi_{\text{gen}}\)) optimizing the supremum over approximation blurs \(\mathcal{A}\). | Local tenability \(TS_{\text{local}}(y, M) = \sup \{1 - \delta \mid \exists x^* \in M : (\Phi(y), x^*) \in u_\delta\}\), edge tenability \(TS_{\text{edge}}\), and aggregated \(\text{Tenability}(T) = w_{\text{local}} \overline{TS}_{\text{local}} + w_{\text{edge}} \overline{TS}_{\text{edge}}\). |
| Consistency [2, p. xviii], [1, sec. 5.2] | Detects semantic drift, polysemy, and equivocation across documents and argumentative contexts. | Extracts contextual embeddings across entity mentions and computes cluster dispersion/silhouette scores. | Semantic embedding coherence \(S_{\text{embed}}(t) = \frac{1}{\binom{k}{2}} \sum \frac{\mathbf{v}_i \cdot \mathbf{v}_j}{\|\mathbf{v}_i\| \|\mathbf{v}_j\|}\) and Polysemy Silhouette Drift Score \(S_{\text{sil}}(t)\). |
| Conservatism & Epistemic Safety [5, sec. 5.2] | Evaluates conceptual detour (Conservatism) and evidential decay along reasoning chains from empirical sources to theoretical claims (Safety). | Computes Euclidean distance across weighted adjacency vectors and finds Maximum Reliability Paths. | Conservatism ratio \(C(H) = \frac{1}{\|\pi_s\|} \sum \frac{\delta(v_1, v_{\|p\|})}{\sum \delta(v_i, v_{i+1})}\), and Epistemic Safety: \(\text{Safety}(v_{\text{target}}) = \max_{p} \left(\prod_{e \in p} w(e)\right) \times c(v_{\text{source}})\). |
| System Coherence (Harmony, \(H\)) [4, p. 443], [1, sec. 5.3] | Measures the global explanatory stability and constraint-satisfaction state of the entire network. | Aggregates activations over symmetric adjacency weights during network relaxation. | Global Harmony Function: \(H(t) = \sum_{i=1}^N \sum_{j=1}^N w_{ij} a_i(t) a_j(t)\), and Normalized System Coherence Score \(SCS(G) = \frac{H(t^*)}{\sum \sum \|w_{ij}\|}\). |
| Node Activation (Acceptability) [4, p. 439], [1, sec. 5.3] | Represents the final equilibrium epistemic acceptance, rejection, or neutrality of an individual proposition. | Evaluates steady-state node activation after ECHO convergence (\(a_i^* = \lim_{t \to \infty} a_i(t)\)). | Continuous equilibrium value \(a_i^* \in [-1, 1]\) thresholded into Accepted (\(a_i^* \ge \tau_{\text{accept}}\)), Rejected (\(a_i^* \le -\tau_{\text{reject}}\)), or Neutral. |
Pillar 3: Empirical Power Metrics¶
Evaluates the falsifiability, empirical coverage, predictive fertility, and observational grounding of theory-elements.
| Metric | Definition & Conceptual Goal | Measurement Approach | Specific Algorithm, Formula, or Mathematical Scope |
|---|---|---|---|
| Effective Empirical Content (EEC v2.1) [1, sec. 1.2, 5.2], [2, p. 81] | Evaluates empirical content and non-triviality in an attributed hypergraph space, balancing restrictive power against theoretical degrees of freedom. | Evaluates forbidden empirical configurations scaled by syntactic class (\(S_v\)) and reliabilist robustness (\(R_v\)), regularized by theoretical mass. | \(\text{EEC}(K) = \Gamma_{\text{relative}}(K) \cdot \exp\left(-\lambda \frac{k_T}{M_w}\right)\), where \(\Gamma_{\text{relative}}\) aggregates forbidden hyperedges, \(w(v) = S_v \cdot e^{-\beta d_v}\), and \(M_w\) is independent empirical mass. |
| Empirical Adequacy & Grounding [2, pp. 93, 215], [3, pp. 40–45, 120] | Quantifies agreement between empirical observation data (\(I\)) and theoretical models (\(M\)) within admissible blurs (\(A\)), incorporating baseline empirical grounding. | Verifies non-empty paradigm applications (\(GR\)) and computes residual distances against declared error margins. | Grounding Ratio \(GR(T) = \frac{\|I_{\text{confirmed}}\|}{\|I_{\text{total}}\|}\), Adequacy Score \(AS = \max_X \left[1 - \frac{1}{\|I\|} \sum \min d(i, x)\right]\), and Blur Margin Compliance Ratio (\(BMCR\)). |
| Explanatory Breadth & Coverage [4, pp. 437, 442], [1, sec. 5.3] | Favors hypotheses explaining diverse evidence clusters while penalizing passive evidence omission (\(UER\)) and active empirical falsifications (\(ECR\)). | Computes consilience in-degree, union of covered evidence nodes, and topological community coverage. | Consilience In-Degree \(Cons(H)\), Relative Node Coverage \(REB_{\text{NODE}}(T) = \frac{\|V_E^{\text{exp}}\|}{\|V_E\|}\), Domain Coverage \(REB_{\text{CLUSTER}}\), Conflict Ratio \(ECR\), and Unexplained Evidence Ratio \(UER = 1 - REB_{\text{NODE}}\). |
| Theoretical Achievements [4, pp. 280–287], [1, sec. 5.2] | Justifies the cognitive introduction of theoretical terms via their ability to restrict observable states and generate cross-domain predictions. | Analyzes theoretical concept nodes bridging disconnected empirical observation subgraphs via constraints. | Combined Theoretical Achievement Score: \(TAS(\tau) = \frac{1}{2}(EFI(\tau) + PPM(\tau))\), where \(EFI\) is the Empirical Focus Index and \(PPM\) is the Prognostic Power Metric. |
Pillar 4: Theory Dynamics (Diachronic Evolution)¶
Evaluates the historical evolution, progressive shifts, degenerative immunizations, and reduction trajectories of a research programme over time (\(\mathcal{E} = \langle N_1, \dots, N_k \rangle\)).
| Metric | Definition & Conceptual Goal | Measurement Approach | Specific Algorithm, Formula, or Mathematical Scope |
|---|---|---|---|
| Progressiveness & Evolution Perfectness [2, pp. 222, 363], [3, pp. 180–195], [1, sec. 4] | Evaluates whether a sequence of theory-nets expands confirmed empirical content, sharpens precision, and converts open conjectures without uncompensated loss. | Decomposes propositions into content elements \(E_e(S)\) with cognitive weights \(w(P)\), tracking confirmed growth, precision tightening, and conjecture conversion. | Decoupled Progressiveness: \(\Pi_{\text{empirical}} = (RR_{Cn})^\gamma [w_1 \sigma(AGR_{Cn}) + w_2 PTM]\) and \(\Pi_{\text{theoretical}} = \sigma(TEI_{Cn})\); Evolution Perfectness Efficiency: \(EPR_{Cn}(\mathcal{E}) = \frac{\sum_{P \in A(I, t_0) \cap F(I, t_k)} w(P)}{\sum_{Q \in A(I, t_0)} w(Q)}\). |
| Degeneration Index & Immunization Tracking [1, pp. 51, 59, 265], [2, p. 333], [3, p. 184] | Diagnoses Lakatosian degeneration across versions and flags ad-hoc immunizing stratagems lacking excess empirical content. | Detects auxiliary modifications resolving anomalies without novel predictions (Typ-b failures) and aggregates version histories. | Node Immunization Index \(II = \frac{\Delta_{\text{anom}}}{\Delta_{\text{anom}} + \Delta_{\text{content}}}\), Progressive-to-Immunizing Ratio (\(RPI\)), Schurz's Degeneration Index \(D(V_j) = \frac{\|F_{\text{Typ b}}(V_j)\|}{\|S(V_j)\| + \|F_{\text{Typ a}}(V_j)\|}\), and Trend \(\Delta D\). |
| Kuhnian / Paradigm-Guidance [2, pp. 175–176], [3, pp. 169–171, 194] | Evaluates whether historical revisions represent coherent "normal science" descending from a root paradigm core (\(K_0\)) or erratic ad-hoc fragmentation. | Verifies specialization reachability \(T_0 \xrightarrow{\alpha, *} T\), measures incremental adherence, and validates domain conservation. | Composite Paradigm Guidance Index: \(PGI_t = \text{CoreStable} \cdot [w_1 PAR(G_t, T_0) + w_2 DCI(I(t), I_0)]\), where \(PAR\) is paradigm adherence and \(DCI\) is domain conservation. |
| Reduction Capacity [2, p. 372], [3, pp. 131, 277–278] | Formalizes the rational mechanism by which a successor theory-net (\(T'\)) supersedes and subsumes an older predecessor (\(T\)) while preserving empirical successes. | Identifies inter-theoretic :REDUCES_TO relations, translation morphisms \(\phi\), and limit condition parameters. |
Reduction Score: \(RS(T_1, T_2) = \frac{1}{2}(SPR(T_1, T_2) + ECI(T_1, T_2))\), combining the Structural Preservation Ratio (\(SPR\)) and Empirical Claim Invariance (\(ECI\)). |
Pillar 5: Metatheoretical Framework¶
Higher-order philosophy of science metrics for macro-paradigm analysis, logical unity, and semantic demarcation.
| Metric | Definition & Conceptual Goal | Measurement Approach | Specific Algorithm, Formula, or Mathematical Scope |
|---|---|---|---|
| Semantic & Axiomatic Homogeneity [1, pp. 11, 73], [2, p. 11] | Evaluates the logical unity of a theory, ensuring it cannot be factorized into disjoint, non-interacting sub-theories (resolving the Tacking Paradox). | Analyzes the logical inference graph between axioms to detect decomposable sub-axiomatizations. | Schurz's Factorization Criterion: verifies that \(E(Ax) = \operatorname{Cn}(E(Ax_1) \cup E(Ax_2))\) is false for all disjoint non-empty partitions; Axiom Interaction Density \(AID\) and Homogeneity Factor \(H_{\text{axiomatic}}(T) = \min AID(A_1, A_2)\). |
| T-Theoreticity [2, pp. 49–78, 391], [3, pp. 40–56], [1, pp. 251–252] | Formally determines whether a term or function can be measured independently of theory \(T\), or whether its measurement presupposes the validity of \(T\). | Traces incoming :DETERMINED_BY / :MEASURED_VIA dependency chains within the intra-theory core and global theory-holon. |
Sneed's Functional Criterion: \(TI(\tau, T) = 1\) if all measurement paths presuppose fundamental laws of \(T\) (\(\tau \in M_p \setminus M_{pp}\)); \(TI(\tau, T) = 0\) if an independent measurement path exists (\(\tau \in M_{pp}\)). |
| Higher-Level Warrant [4, p. 441], [1, sec. 5.3] | Models how hypotheses gain indirect credibility when they are deductively explained or justified by deeper, higher-level theories. | Identifies incoming directed :EXPLAINS / :JUSTIFIES edges from superordinate theory-element nodes. |
Higher-Level Warrant Score: \(HLW(H) = \sum_{T \in \text{Sup}(H)} w(T, H) \cdot a_T(t)\), and Depth of Justification: \(DoJ(H) = \max_{p} \text{length}(p)\). |
| Analogical Support [4, p. 437], [6, ch. 7] | Models lateral epistemic reinforcement between hypotheses in different domains that share isomorphic relational roles (bounded regular equivalence). | Computes Katz-damped path similarities across normalized adjacency matrices within a \(k\)-hop neighborhood. | Lateral Analogical Boost: \(LAB_{\text{norm}}(H_1) = \tanh\left(\sum_{H_2} w_{\text{analogy}}(H_1, H_2) a_{H_2}(t)\right)\), where \(w_{\text{analogy}}\) scales bounded regular equivalence \(\text{Sim}_{\text{reg}}^{(k)}\) with a cross-domain filter \(\phi_{\text{domain}}\). |
| Unification [1, pp. 6, 317–320], [4, p. 441], [8] | Favors theoretical frameworks that repeatedly use a compact set of basic axioms to derive a large volume of diverse consequences. | Evaluates the ratio of derived elements (\(D\)) to un-derived basic assumptions (\(B\)) on the deductive derivation DAG. | Schurz Unification Ratio: \(U_{\text{ratio}}(S) = \frac{\|D\|}{\|B\|}\), and Global Unification Index: \(GUI(S) = \frac{\|D\|}{\|B\| + \|D\|}\). (See detailed overlap analysis below). |
Metric Overlap, Redundancy, & Consolidation Analysis¶
Because formal philosophy of science, structuralism, and network theory approach theoretical virtues from multiple angles, several metrics exhibit mathematical or conceptual overlap. The following analysis clarifies these relationships and documents our architectural consolidation strategy.
In-Depth Analysis: What About Unification? Do We Need It?¶
The Unification metric (\(U_{\text{ratio}} = |D| / |B|\)) occupies a central place in classical philosophy of science (Kitcher, 1989; Schurz, 2024), formalizing the intuition that a good theory unifies diverse phenomena under a minimal axiomatic foundation.
Where Unification Overlaps:¶
- Overlap with Explanatory Breadth (\(REB\)):
- Both metrics measure explanatory yield. The derived elements set \(D\) in an empirical graph consists predominantly of explained observation nodes (\(V_E^{\text{exp}}\)).
- Key Distinction: Explanatory Breadth (\(REB = |V_E^{\text{exp}}| / |V_E|\)) evaluates domain coverage (what fraction of known evidence is explained), but is completely agnostic to how many axioms were needed to achieve that coverage. An ad-hoc system that introduces 100 axioms to explain 100 facts achieves a perfect \(REB = 1.0\), but its Unification Ratio collapses to \(100 / 100 = 1.0\) (severely un-unified). Unification penalizes axiomatic fragmentation.
- Overlap with Minimum Description Length (\(MDL\)) & Elegance & Economy:
- In algorithmic information theory, \(MDL(T) = L(M) + L(D \mid M)\) is the precise mathematical continuous counterpart to discrete Unification.
- \(L(M)\) represents the complexity of the axiom base (\(B\)), while \(L(D \mid M)\) represents the cost of uncompressed/unexplained empirical data (\(D\)). A high Unification Ratio directly corresponds to minimizing total description length: a compact model \(M\) compressing a large data mass \(D\).
- Implicit Enforcement in the LPG-ECHO Algorithm:
- As documented in adapted_echo_algorithm.md, the connectionist relaxation engine organically enforces unification. Clamped evidence nodes pump activation into the network, while systemic decay \(\theta\) bleeds activation each cycle. If a theory contains too many axioms (\(B\)) relative to verified empirical anchor points (\(D\)), activation is diluted and the excess axioms decay to zero ("ghost nodes").
Architectural Verdict on Unification:¶
Tip
Decision: Retain as a Macro-Diagnostic Ratio. We retain Unification as an explicit, high-level diagnostic ratio because:
- Computational Efficiency: Calculating \(U_{\text{ratio}} = |D| / |B|\) requires only simple \(O(1)\) node-degree queries on the deductive derivation DAG (
MATCH (b) WHERE NOT (b)<-[:DERIVED_FROM]-()), avoiding the overhead of full MDL estimation or iterative ECHO convergence. - Conceptual Interpretability: It provides domain scholars with an immediately legible quotient reflecting the classical Kitcher-Schurz unification standard.
- Role in Taxonomy: Rather than a primitive property, Unification is formalized as the macro quotient of Empirical Output (\(D \approx REB\)) over Axiomatic Base (\(B \approx 1/APS\)).
Comprehensive Overlap & Redundancy Map¶
The table below outlines all overlapping metric pairs, their formal relationships, and their consolidation status across the codebase and documentation:
| Metric Pair / Cluster | Overlapping Elements | Formal Relationship | Consolidation Status |
|---|---|---|---|
| True Partial Empirical Claim \(\leftrightarrow\) Empirical Adequacy | Both verify non-empty paradigm applications (\(I_0 \neq \emptyset, I_0 \in \operatorname{Cn}(K)\)). | True Partial Empirical Claim is identical to the Grounding Ratio (\(GR(T) = \|I_{\text{confirmed}}\| / \|I_{\text{total}}\|\)) in Empirical Adequacy. |
Absorbed: Fully integrated as the baseline condition (\(GR > 0\)) of Empirical Adequacy. |
| Unexplained Evidence Ratio (\(UER\)) \(\leftrightarrow\) Explanatory Breadth (\(REB\)) | Both track evidence coverage vs. omissions in empirical domain \(V_E\). | \(UER(T) = \frac{\|V_E^{\text{unexp}}\|}{\|V_E\|} = 1 - REB_{\text{NODE}}(T)\). It is the arithmetic complement of relative coverage. | Absorbed: Defined directly inside Explanatory Breadth (Sec. 3.4) as the passive omission gap. |
| Immunization Tracking (\(II, RPI\)) \(\leftrightarrow\) Degeneration Index (\(D, \Delta D\)) | Both detect ad-hoc auxiliary hypotheses resolving anomalies without excess empirical content (\(\Delta_{\text{content}} = 0\)). | Node-level immunization (\(II\)) aggregates directly into Schurz's version-level Degeneration Index (\(D(V_j)\)) as Typ-b failures. | Unified: Merged into Degeneration Index & Immunization Tracking. |
| Evolution Perfectness (\(EPR\)) \(\leftrightarrow\) Progressiveness (\(\Pi_{Cn}\)) | Both evaluate the transition of open conjectures \(A(I)\) into confirmed successes \(F(I)\) over time. | Perfectness is the asymptotic efficiency limit (\(EPR_{Cn}\)) of a diachronic research programme evolution \(\mathcal{E}\). | Integrated: Formalized as the asymptotic efficiency component (\(EPR_{Cn}\)) in Progressiveness. |
| System Tolerance & Skepticism (\(\theta\)) \(\leftrightarrow\) LPG-ECHO Algorithm | Both model belief decay \(a_i(t) = a_i(0)(1-\theta)^t\) and belief half-life \(t_{1/2}\). | Skepticism is an intrinsic control parameter of the connectionist relaxation loop, not an independent topological metric. | Integrated: Formalized as an intrinsic mechanic in The LPG-ECHO Algorithm. |
| Simplicity \(\leftrightarrow\) Elegance and Economy | Both operationalize Ockham's razor and Minimum Description Length (MDL). | Decoupled by structural scale: Simplicity governs micro-level explanation chains (\(\phi(a, b) \propto 1/\sum c(a_j)\)) and graph entropy; Elegance & Economy governs macro theory cores (\(APS\), \(L(M) + L(D \mid M)\)). | Paired (S&E): Retained as complementary multi-scale parsimony metrics. |
| Connectedness + Modesty \(\leftrightarrow\) Structural Elegance Index | Balances algebraic connectivity (\(\lambda_2\)) against inverse graph density (\(1/\rho\)). | Composite index: \(E(G) = \lambda_2(G) - \alpha \rho(G)\). | Hierarchical: Structural Elegance is the composite index; Connectedness and Modesty are foundational inputs. |
| Homogeneity + Refutability + Connectivity \(\leftrightarrow\) Bottleneck Fragility Profile | Unifies degree entropy (\(H_{\text{deg}}\)), vertex connectivity (\(\kappa\)), and betweenness shortest-path drop (\(R_k\)). | Composite diagnostic report evaluating single points of failure and collapse impact. | Hierarchical: Bottleneck Fragility Profile is the composite diagnostic; individual metrics serve as foundational inputs. |
| Epistemic Safety \(\leftrightarrow\) LPG-ECHO Node Activation | Both model confidence propagation from empirical grounding to theoretical conclusions. | Safety is the single maximum-reliability path product; ECHO is the multi-path connectionist relaxation across the full network. | Complementary: Safety provides deterministic path bounds; ECHO provides continuous network equilibrium. |
Consolidated Grounding Sources Directory¶
- [1] Gerhard Schurz (2014 / 2024 / 2026):
- Philosophy of Science: A Unified Approach. New York: Routledge, 2014/2024.
- Wissenschaftstheorie: Einführung. 5., aktualisierte und überarbeitete Auflage. Nomos, 2026.
- (Grounds relevant empirical content elements \(E_e\), dynamic truthlikeness, tacking-paradox resolution, axiomatic homogeneity, Degeneration Index \(D\), and unification ratios).
- [2] Wolfgang Balzer, C. Ulises Moulines, Joseph D. Sneed (1987):
- An Architectonic for Science: The Structuralist Program. Dordrecht: Reidel Publishing Company.
- (Grounds model-theoretic cores \(K = \langle M_p, M, M_{pp}, GC, GL \rangle\), theory-nets, theory-trees, structural connectedness, diachronic specialization, admissible blurs, and reduction type \(w_1\)).
- [3] Wolfgang Stegmüller (1976):
- The Structure and Dynamics of Theories. New York: Springer-Verlag.
- (Grounds the Sneedian model-theoretic triad (\(M_{pp}, M_p, M\)), functional T-theoreticity, strict reduction (\(\text{RED}\)), tenability, empirical adequacy, and the central Sneed-Ramsey theory claim).
- [4] Paul Thagard (1989):
- "Explanatory Coherence." Behavioral and Brain Sciences, 12 (3), pp. 435–467.
- (Grounds parallel constraint satisfaction, neural network relaxation, system harmony \(H(t)\), consilience, co-hypothesis simplicity scaling (\(1/n\)), and systemic skepticism/decay dynamics).
- [5] Vít Nováček et al. (2015):
- "Formalising Hypothesis Virtues in Knowledge Graphs: A General Theoretical Framework and its Validation in Literature-Based Discovery Experiments." arXiv:1503.09137v2.
- (Grounds graph-theoretic adaptations of epistemic virtues: Modesty via inverse density, Conservatism via characteristic context vectors, Structural Refutability via shortest-path drops, and universe simplification entropy).
- [6] Mark Newman (2018):
- Networks: An Introduction. 2nd Edition. Oxford: Oxford University Press.
- (Grounds algebraic connectivity \(\lambda_2\), Newman-Girvan modularity, edge betweenness centrality, degree entropy, and network density).
- [7] Peter Grünwald (2005):
- Advances in Minimum Description Length: Theory and Applications. Cambridge: MIT Press.
- (Grounds two-part practical Minimum Description Length \(MDL = L(M) + L(D \mid M)\) as learning via data compression).
- [8] Philip Kitcher (1989):
- "Explanatory Unification and the Causal Structure of the World." In P. Kitcher & W. Salmon (Eds.), Scientific Explanation, Minnesota Studies in the Philosophy of Science, Vol. XIII, pp. 410–505.
- (Grounds the explanatory unification model via deductive derivation schemas and argument patterns).