Module 3: Start Here

Overview

Module 2 left you with an ASR list, a Utility Tree and an NFR budget. Those are the inputs to a decision you have not yet made: what shape is this system?

This module answers:

How do I choose a structure, and how do I turn that choice into something a company with no spare capacity can actually follow?

This is the module closest to conventional system design, and that is exactly why it has to stay at strategy altitude. You already know how microservices work, what a queue is for, and where caching goes. None of that is re-taught here. What is taught is the part that is genuinely hard: picking a structure from requirements rather than from familiarity, and sequencing the work when every engineer is already busy.

How the module is built

Two sections. The first chooses a structure. The second turns a choice into a strategy — which is a different skill, and the one more architects are missing.

graph TD
  subgraph S1["Section 1 — Choosing a structure"]
    A["1 · Styles and<br/>what they cost"]
    B["2 · Coupling<br/>and cohesion"]
    C["3 · The quality<br/>tactics catalog"]
    D["4 · Tactics for<br/>probabilistic parts"]
    E["5 · Choosing under<br/>constraint"]
    A --> B --> C --> D --> E
  end
  subgraph S2["Section 2 — From choice to strategy"]
    F["6 · Reference<br/>architecture"]
    G["7 · Strategy as a<br/>sequence of bets"]
    H["8 · Sequencing with<br/>no spare capacity"]
    I["9 · The roadmap as<br/>a negotiation"]
    J["10 · A repeatable<br/>design process"]
    K["11 · Criteria catalogs<br/>and scale calibration"]
    F --> G --> H --> I --> J --> K
  end
  E ==> F
  K ==> L(["Workbook — Exercise 3"])

  classDef s1 fill:#5B8FB9,stroke:#4A7A9E,stroke-width:2px,color:#ffffff
  classDef s2 fill:#E8A33D,stroke:#C4832A,stroke-width:2px,color:#ffffff
  classDef cap fill:#D9695B,stroke:#B4483A,stroke-width:2px,color:#ffffff
  class A,B,C,D,E s1
  class F,G,H,I,J,K s2
  class L cap

The module’s path. Section 1 produces a decision; Section 2 produces something other people can execute.

Section 1 — Choosing a structure

What the options cost → what actually drives the choice → the moves available → the ones that are new → deciding inside constraints you did not pick.

1. Styles and what they costFive structures, stated in quality attributes rather than enthusiasm.

2. Coupling and cohesionThe two properties that decide the style choice, and how to see them before you build.

3. The quality tactics catalogAvailability, performance and scalability as deliberate selections with stated costs.

4. Tactics for probabilistic partsDegradation to deterministic fallback, and cost-aware routing.

5. Choosing under constraintSelecting a structure when a migration is already consuming the company.

Section 2 — From choice to strategy

A decision nobody can execute is not a decision.

6. Reference architectureConstraining future decisions without prescribing implementations.

7. Strategy as a sequence of betsEach move stated as a bet, with its reasoning and its reversal condition.

8. Sequencing with no spare capacityWhat to do first when everything is first, and what gets displaced.

9. The roadmap as a negotiationThree stakeholders want three different orders. One sequence ships.

10. A repeatable design processFrame, constrain, generate, evaluate, decide, record — written down so it is not you.

11. Criteria catalogs and scale calibrationWhy you carry selection criteria rather than recommendations, and why real numbers matter.

How to work this module

About six hours, split evenly:

1. Read the eleven topics — about 3 hours. Topics 1–3 will feel familiar; read them for the altitude rather than the content, because the workbook exercise asks you to justify a choice, not describe one.

2. Optional — Workbook exercise 3 — about 3 hours. Three artifacts, each scored against Module 2’s ASRs.

In the workbook

Three artifacts for Lintel:

• A style selection with an explicit trade-off table, scored against your Module 2 ASRs
• A reference architecture sketch (C4 Level 2) showing where the AI components sit relative to the existing estate
• A 12-month roadmap sequenced against the fact that engineering capacity is already fully consumed

Before you start

You need Module 2’s ASR list and NFR budget open. The rubric requires the style choice to cite specific ASRs, and a choice justified by general preference fails it. If your NFR budget has no ordering sentence, go back and add one — this module’s trade-off table is unfillable without it.

Nothing in the prerequisite material is re-taught. Distributed systems failure modes, consistency models, queue and cache selection, service mesh and gateway mechanics — all assumed. Where a topic needs one, it gets a line and a pointer.

Check your understanding

Answer now, revisit at the end.

1. Think of a style choice you have seen made. What was the stated reason, and what do you think the real one was?
2. Name a quality tactic you have applied without calling it one. What did it cost?
3. If every engineer in your organisation is already committed, what is the honest first step of any new initiative?

→ Start with Styles and what they cost.