Reading Order & Prerequisites

Reading Order & Prerequisites

This guide helps you navigate the 162-page documentation site efficiently by showing dependencies between sections and suggesting paths based on your goals.

Section Dependency Graph

flowchart TB
    subgraph Foundation["Foundation (Read First)"]
        GS[Getting Started]
        CC[Core Concepts]
    end

    subgraph Theory["Theory (Build Understanding)"]
        DR[Delegation Risk]
        PD[Power Dynamics]
        ENT[Entanglements]
    end

    subgraph Application["Application (Take Action)"]
        DP[Design Patterns]
        CS[Case Studies]
        XD[Cross-Domain Methods]
    end

    subgraph Advanced["Advanced (Deep Dives)"]
        RES[Research]
        EXP[Experimental]
    end

    GS --> CC
    CC --> DR
    CC --> PD
    CC --> ENT
    DR --> DP
    PD --> DP
    ENT --> DP
    DR --> CS
    DP --> CS
    CC --> XD
    DR --> RES
    RES --> EXP

    style GS fill:#e3f2fd
    style CC fill:#e3f2fd
    style DR fill:#fff3e0
    style PD fill:#fff3e0
    style ENT fill:#fff3e0
    style DP fill:#e8f5e9
    style CS fill:#e8f5e9
    style XD fill:#e8f5e9
    style RES fill:#f3e5f5
    style EXP fill:#f3e5f5

Prerequisites by Section

To Read…	First Understand…	Time Investment
Core Concepts	Nothing (start here)	20 min
Delegation Risk	Core Concepts	45 min + Core
Power Dynamics	Core Concepts	30 min + Core
Entanglements	Core Concepts	60 min + Core
Design Patterns	Core Concepts + at least one of (DR, PD, or ENT)	2+ hours
Case Studies	Core Concepts + Design Patterns	1-2 hours
Cross-Domain Methods	Core Concepts	30-60 min
Research	All theory sections	5+ hours

---

Paths by Goal

”I’m building an AI system from scratch”

1. Core Concepts — Understand the framework
2. Design Patterns Index — See what patterns exist
3. Least-X Principles — Core design philosophy
4. Quick Start — Step-by-step checklist
5. Entanglements — Avoid correlated failures

Total time: 2-3 hours

”I’m assessing risk in an existing system”

1. Core Concepts — Framework basics
2. Delegation Risk Overview — Quantification approach
3. Risk Decomposition — How to break down risk
4. Case Studies — See examples
5. Cost-Benefit Tool — Evaluate mitigations

Total time: 2-3 hours

”I’m skeptical this framework works”

1. FAQ — Common objections answered
2. Sydney Case Study — Real-world failure analysis
3. Nuclear Safety PRA — Similar methods that work
4. Lessons from Failures — Historical context
5. Research — Theoretical foundations

Total time: 2-3 hours

”I want to understand the math”

1. Core Concepts — Conceptual foundation
2. Delegation Risk Overview — Formulas
3. Delegation Walkthrough — Worked examples
4. Risk Decomposition — Formal treatment
5. Power Dynamics — Agency formalization
6. Experimental Estimates — Squiggle distributions

Total time: 4-6 hours

”I want to apply this to my organization”

1. Core Concepts — Basics
2. Quick Start — Practical checklist
3. Human Systems Case Studies — Organizational examples
4. Cost-Benefit Tool — ROI analysis
5. Entanglements: Mitigation — How to fix issues

Total time: 3-4 hours

”I’m a researcher”

1. Core Concepts — Framework overview
2. All Theory Sections — Full understanding
3. Research Index — Open problems
4. Potential Projects — Contribution ideas
5. Experimental — Probabilistic methods

Total time: 10+ hours

---

Section Overviews

Getting Started (Foundation)

What you’ll learn: The core problem (managing risk when delegating to AI systems), the basic formula, and key concepts like harm modes, defection risk, and risk budgets.

Key pages: Introduction, Core Concepts, FAQ

Delegation Risk (Theory)

What you’ll learn: How to quantify delegation risk mathematically, decompose it into components, and track risk through hierarchical systems.

Prerequisites: Core Concepts

Key pages: Overview, Walkthrough, Risk Decomposition

Power Dynamics (Theory)

What you’ll learn: How to formalize agent power, authority, and the “Strong Tools Hypothesis” about capability constraints.

Prerequisites: Core Concepts

Key pages: Agent Power Formalization, Strong Tools Hypothesis

Entanglements (Theory)

What you’ll learn: How correlated components undermine safety assumptions, how to detect entanglement, and how to mitigate it.

Prerequisites: Core Concepts

Key pages: Index (Independence Illusion), Detection, Mitigation

Design Patterns (Application)

What you’ll learn: 45 patterns for building safer delegation systems, organized by threat model.

Prerequisites: Core Concepts + at least one theory section

Key pages: Index (pattern matrix), Least-X Principles, Tools

Case Studies (Application)

What you’ll learn: How these concepts apply to real AI systems (Sydney, code review bots) and human systems (nuclear, finance).

Prerequisites: Core Concepts + Design Patterns recommended

Cross-Domain Methods (Application)

What you’ll learn: How mature risk management fields (nuclear, finance, carbon budgets) handle similar problems.

Prerequisites: Core Concepts

Research (Advanced)

What you’ll learn: Theoretical foundations, open problems, and connections to academic literature.

Prerequisites: All theory sections

Experimental (Advanced)

What you’ll learn: Probabilistic estimation tools, Squiggle distributions, and uncertainty quantification.

Prerequisites: Research section recommended

---

Quick Reference

If you have…	Read…
5 minutes	Five-Minute Intro
30 minutes	Introduction + Core Concepts
2 hours	Foundation + one theory section + Quick Start
Half a day	Foundation + all theory + Design Patterns index
Full day	Everything except Research and Experimental

---

See Also

• Site Map — Visual map of all content
• Glossary — Term definitions
• FAQ — Common questions