Explanation

These pages explain the ideas behind doxastica: the theory it implements, the architecture it follows, and the design decisions that make it correct and reusable. Read them when you want to understand why the library works the way it does. The tutorial and how-to guides show you how.

Theory and architecture

• What Is AGM Belief Revision? Revise, Expand, Contract: The epistemic problem doxastica solves, belief base versus belief set, the three operations, and what doxastica keeps and drops from the theory.
• The Kumiho Architecture: The graph-native, append-only design doxastica implements, and its two deliberate departures from the paper.

The core model

• Scopes and the World Scope: What a scope is, why multi-scope extends single-agent Kumiho, and why contracting the world scope is forbidden.
• The Superseded Chain: Append-Only, No Recovery: Why nothing is ever deleted, how the revision spine is structured, and how it replaces AGM recovery with an audit trail.
• Derived Current State and the UUID7 Ordering Contract: Why current state is computed rather than stored, and how the ordering contract handles intra-millisecond collisions.

Design boundaries

• The Two Seams: BeliefStore vs BackendPort: The public protocol consumers code against, the internal one storage backends implement, and why there are two.