← Dynamic Systems Engineering Foundations

Example Library

A structured repository of case studies for systems reasoning. Examples aren't demonstrations — they're training environments where students and AI systems practice system identification, causal reasoning, state estimation, feedback analysis, knowledge engineering, validation, and design improvement. A strong example teaches many concepts repeatedly, and evolves in depth as the learner advances.

Standard Example Artifact Format

Every example in this library is documented the same way: Example Name / System / Domain / Complexity Level / Primary Concepts, System Description, System Boundary, Variables (Observable / Hidden / Essential), Causal Model, Mathematical Representation, Failure Modes, Validation Methods, Improvement Opportunities, and Knowledge Artifacts (reusable concepts, related examples, connected theories).

The Five Core Anchor Examples

Used repeatedly across Weeks 1–10 rather than as one-time illustrations — each concept lands differently the fourth time you see it applied to a familiar system than the first.

Thermostat / Feedback Control — Level 1 complexity. Covers systems, variables, feedback, control, state estimation, validation. Simple enough for beginners but contains the fundamental structure of intelligent regulation.

Human Learning System — Level 2. Covers knowledge, adaptation, feedback, state estimation, personalization, AI tutoring. Demonstrates that dynamic systems exist in biological and cognitive domains, not just engineered ones.

Artificial Intelligence Assistant — Level 2–3. Covers knowledge engineering, uncertainty, reasoning, artifacts, human-AI collaboration. The AI itself becomes both a tool and a system being analyzed.

Transportation System — Level 2–3. Covers complex interactions, optimization, emergence, control, prediction. Demonstrates how local decisions create system-level behavior.

Small Business System — Level 2. Covers resource allocation, feedback, optimization, risk, decision-making. Demonstrates DSE in practical human systems.

Advanced Research Examples

These are drawn from this site's own real, verified work — not generic placeholders. Where something isn't actually verified from direct experience, that's stated rather than filled in.

Non-Collapsible Four-Channel Architecture (NCFCA) — Level 3.
Concepts: Single-writer event-sourced state, formal transition tables, AI proposal-only patterns, replay verification.
Real, specific detail: The actual kernel (Serialized Event Kernel) separates state into reality/cognitive/observer/integrating channels, with a DeterministicValidator that re-checks every AI proposal at processing time rather than trusting a pre-packaged decision — closing the obvious forgery path. Verified: 112/112 unit tests, and a real directional-consistency check where shuffled event delivery produced results identical to true-order delivery (18/20 sign agreements either way), confirming the kernel's canonicalization claim empirically, not just architecturally. A related contamination benchmark showed 0% vs. 44.1% [95% CI 43.11%–45.06%] contamination rates between isolated and monolithic architectures across 10,000 trials — and a separate, less rigorous version of that comparison (built by another tool) was caught using a hardcoded-probability toy simulation dressed as empirical measurement, and excluded from the site for that reason. Both the real result and the caught error are part of what this example actually teaches.

AI Knowledge Architecture — Level 2–3.
Concepts: Externalized memory, artifact typing, revision history, human-approved persistence.
Real, specific detail: This curriculum's own AI Artifact Specification — five real artifact types (Concept, System, Research, Revision, Learning), each with a defined field structure, plus an explicit review cycle before anything gets finalized. Not a hypothetical design — it's the actual protocol used to build this course.

Educational Framework (Wayfinder) — Level 2–3.
Concepts: Adaptive mastery gating, prerequisite DAGs, misconception-targeted content, real confidence intervals instead of raw pass/fail.
Real, specific detail: The Wayfinder GED system gates node mastery on a Wilson score interval's lower bound, not the raw success rate — a 42/42 streak and a 9/10 streak get correctly distinguished by evidence strength rather than treated as equally confident. The real RLA Reading Comprehension content (9 nodes) ties each wrong multiple-choice answer to a specific, named, previously-documented misconception rather than a generic distractor. 148 real structural and functional checks verified before deployment, including the harder case: a node with three simultaneous prerequisites correctly staying locked until all three are actually met.

Sports Prediction Systems — Level 2.
Concepts: Statistical modeling, uncertainty intervals, backtesting.
Real, specific detail: The NFL Strategy Composer uses the same real Wilson interval formula as Wayfinder — previously implemented twice, independently, with different scales and defaults; consolidated into one shared, verified source (wilson.js) after confirming both implementations were numerically identical to floating-point precision first. Real backtest data is used; deeper claims about the backtest's own predictive track record haven't been independently reverified here and shouldn't be assumed from this entry alone.

Project Vigil — not included.
Named in an earlier draft of this document as an advanced example (human-machine authority management), but there's no real, verified detail available from direct work on it to draw from. Left out rather than filled in with a plausible-sounding description — if this should be a real example, the actual specifics need to come from you, not be inferred.

Example Teaching Method

Every example uses the same seven-phase cycle: Observe (what do we know?), Define (what is the system?), Model (what variables and relationships exist?), Predict (what should happen?), Test (did reality match the prediction?), Improve (how should the model change?), Preserve (what knowledge should become an artifact?).

AI Usage Guidelines

When AI analyzes an example, it should: identify assumptions, ask clarifying questions, separate observations from interpretations, generate alternative models, preserve uncertainty, suggest validation methods, and ask whether important discoveries should become artifacts.

Final Principle

The Example Library is the practical laboratory of Dynamic Systems Engineering. Concepts become understanding only when they can be repeatedly applied to real systems.