Glossary

This page is the single source of truth for terminology. The other concept pages link here. The monograph in papers/src/EPP/ is the formal authority; this page is the operational synopsis.

Core terms

Causaloid: A self-contained unit of causality. Wraps a causal function and metadata. Composes recursively into Singleton, Collection, and Graph forms that share the same type. See Causaloid.

Causaloid Graph: A directed graph whose nodes are Causaloids and whose edges express the order of evaluation. The result of evaluating the graph is the effect produced at its terminal node(s).

Causal Discovery Language (CDL): A typestate-builder pipeline that ingests observational data and produces a Causaloid graph. Lives in deep_causality_discovery. See CDL.

Causal Function: The pure function a Causaloid wraps. Two signatures: stateless (CausalFn<I, O>) takes the input alone; contextual (ContextualCausalFn<I, O, STATE, CTX>) also receives a Context.

Causal Monad: The monadic algebra that composes Causal Effect Propagation Processes. Exposes pure and bind. The axiom m₂ = m₁ >>= f is its bind operation. See Causal Monad.

Causal Reasoning: The act of running one or more Causaloids against a Context and consuming the resulting propagating effect.

Causal State Machine (CSM): A higher-level construct that links recognized causal states to deterministic actions. Used in the Effect Ethos for separating inference from action. The monograph introduces it in teleology.tex.

Context: An explicit hypergraph encoding the environment in which Causaloids operate. Nodes are Contextoids; edges are typed relations. See Context.

Contextoid: The atomic node of a Context. Carries a typed payload: Datoid, Spaceoid, Tempoid, SpaceTempoid, or Symboid. Non-recursive by design.

Contextual Fabric: The monograph’s name for the union of Contexts and Context Hypergraphs in which effects propagate. The Rust code exposes this through BaseContext and its generic relatives.

Dynamic Causality: The library’s framing of causality as a process whose structure (or context, or both) can evolve. The default operating mode. See Dynamic Causality.

Effect Ethos: The verification layer that checks every causal effect against a set of named Teloids. Implements defeasible deontic inference with Lex Posterior, Lex Specialis, and Lex Superior. Lives in deep_causality_ethos. See Effect Ethos.

Effect Log: An append-only audit log carried by every CausalEffectPropagationProcess. Every Causaloid invocation contributes one entry; bind merges logs across the chain.

Effect Propagation Process (EPP): Both a concept and a literal type in code. The concept: the directed flow of effects through a Causaloid chain. The type: CausalEffectPropagationProcess<Value, State, Context, Error, Log> in deep_causality_core. See Effect Propagation Process.

Effect Value: The enum that holds the payload of a propagating effect: None, Value(T), ContextualLink, RelayTo, or Map.

Evidence: A unit of factual data in the monograph’s ontology. In code, evidence enters the system as Contextoids of type Datoid.

Higher-Kinded Types (HKT): Type-level functions that take types as arguments and return types. The library encodes them via the witness pattern (HKT3, HKT5) defined in deep_causality_haft. See HKT.

Propagating Effect: The stateless alias PropagatingEffect<T> = CausalEffectPropagationProcess<T, (), (), CausalityError, EffectLog>. The everyday return type of a Causaloid’s function.

Teloid: The atomic deontic rule inside an Effect Ethos. Encodes a modality (obligatory, impermissible, optional), a condition, and a Context query. The monograph defines them in teleology.tex.

Former framing

The project was previously described in some materials as “hypergeometric computational causality”. That phrase is retired. The framework’s surface (Causaloid, Context, Effect Ethos) and its formal axiom (m₂ = m₁ >>= f) are about dynamic causality, and the public documentation uses that phrase consistently. The monograph itself never used the older phrase; the rebrand is a cleanup, not a pivot.

On the three modalities

The monograph distinguishes three modalities of causal model:

• Static: fixed Causaloid structure, fixed Context. Recovers classical causality (Pearl SCMs, Bayesian networks, Granger). Epistemology: correspondence theory.
• Dynamic: fixed Causaloid structure, mutable Context. The library’s default operating mode. Epistemology: coherence theory.
• Emergent: mutable Causaloid structure, mutable Context. The frontier case where the rules themselves evolve. Epistemology: pragmatic efficacy.

All three are first-class. The same crates support all three. Choose by configuration, not by library swap.