2. Language Objectives
2.1 Problems CADL Solves
This chapter, like Chapter 1, describes the goals of the language; what the reference implementation provides today is listed in Chapter 10 and Appendix A, §A.12.
CADL aims to describe the three layers of institutions, protocols, and algorithms in a single language for SoS, and to verify consistency and safety through formal verification at the design stage. The goal is to establish an engineering cycle for institutions: "describe → verify → deploy → monitor."
CADL addresses the following specific problems:
Formalization of institutions: Convert natural language institutional descriptions into machine-readable and verifiable formats. Enable authority structures, information sharing scope, and incentive designs to be expressed as quantitative parameters.
Automatic consistency verification: Automatically detect contradictions between multiple contracts, authority conflicts, and deadlock possibilities at the design stage.
Dynamic institutional regime transitions: Describe dynamic institutional switching in response to environmental changes (e.g., transition from C-SoS to A-SoS) with safety guarantees.
Compositional verification support: When constituent systems join or leave, verify integrity using differential verification rather than full re-verification.
Runtime monitoring: Support runtime monitoring of deployed institutions and detection/response to contract violations.
2.2 Design Philosophy and Principles
Declarative description: Describe what institutions should achieve, while the toolchain determines how. Designers are freed from implementation details.
Hierarchical composition: Adopt a hierarchical description structure corresponding to the five-layer framework of SoS design [Shimoyama & Matsubara, 2026]: (1) Objective (policy), (2) Institutional and Governance, (3) Control and Coordination, (4) Execution and Operation, and (5) Environment and Disturbance. CADL's own three layers sit inside this framework as follows.
| Five-layer framework | Where it appears in CADL |
|---|---|
| 1. Objective (policy) | Not a layer of CADL. The evaluation criteria that follow from the objectives are written in metrics:. |
| 2. Institutional and Governance | Institution layer: actors: (the participants), contracts: (authority, information, responsibilities, incentives), and transitions:. |
| 3. Control and Coordination | Protocol layer (protocols:, the coordination procedure) and Algorithm layer (algorithms:, which names the central and local algorithms without describing their internals). |
| 4. Execution and Operation | Outside the language. It is reached through code generation and simulator configs (Appendix D). |
| 5. Environment and Disturbance | Not a layer of CADL. Assumptions about the environment are written in context:. |
Parametric institutional design: Make institutional parameters such as information sharing degree (α), decision centralization degree (β), and incentive intensity (λ) numerically manipulable.
Contract-based composition: Based on assume-guarantee contracts, clearly specify each actor's preconditions and guarantees, enabling compositional verification.
Incremental refinement: Enable gradual refinement from abstract institutional descriptions to detailed implementation specifications. At early design stages, allow natural language-like descriptions and progressively add information needed for verification.
Domain-independent: Provide domain-independent descriptive capability applicable to transportation, energy, IoT, daily life, etc. Domain-specific extensions are provided through libraries.
Universal Readability: Enable students, enterprise practitioners, municipal officials, citizens, and other stakeholders without specialized knowledge of programming or formal methods to understand the intent and structure of institutional descriptions. The means include YAML-based declarative syntax, vocabulary close to natural language, graduated description levels, and AI-assisted barrier reduction.
2.3 Target Users and Application Scenarios
Based on the principle of "universal readability," CADL envisions a broad range of users from experts to non-experts. Below are the main user archetypes and their levels of involvement.
SoS Architects and Designers: Experts who design and verify institutional structures for large-scale SoS. Primarily use verification-level descriptions and define formal assume-guarantee contracts.
Institutional Design Researchers: Researchers who experimentally verify theories such as game theory and mechanism design. Leverage parametric design-level descriptions and explore institutional parameter spaces.
Students and Educators: Students and instructors learning SoS and institutional design concepts in university and technical college courses. Start with overview-level descriptions and progress to design level. Educational templates and AI support accelerate learning.
Enterprise Practitioners and Project Managers: Practitioners designing inter-departmental cooperation rules and supply chain institutions. Use natural language-like overview-level descriptions, with AI supporting refinement to design level.
Municipal Officials and Policymakers: Government officials responsible for institutional design in regional public service coordination and smart city initiatives. Through visualization and AI support, participate in institutional discussion and consensus-building without focusing on technical details.
Citizens and Community Participants: Users who describe household rules, condominium management institutions, and local community institutions in CADL with AI support. Participate in everyday institutional design through overview-level descriptions and AI dialogue.
AI Agents: AI systems that perform automatic conversion from natural language to CADL descriptions via LLM and provide institutional recommendations based on CADL. Generate and process descriptions at all levels.
Below shows the correspondence of description levels and support mechanisms for each user type.
| User Type | Primary Description Level | Primary Support Means |
|---|---|---|
| SoS Architect | Verification Level (Formal Descriptions) | IDE Plugin, Type Checker |
| Researcher | Design Level (Parametric) | Parameter Exploration Tools |
| Student/Educator | Overview → Design Level | Educational Templates, Tutorials |
| Enterprise Practitioner | Overview Level → AI Refinement | AI Support, Visualizer |
| Municipal Official | Overview Level | Visualization, AI Dialogue |
| Citizen | Overview Level (AI-Supported) | NL→CADL Conversion, AI Assistant |
| AI Agent | All Levels | API, Parser Library |