Skip to content
tl-first-principles logo

First Principles of Software Design

tl-first-principles

Foundational software design principles traced to their intellectual origins. Covers information hiding, separation of concerns, abstraction, SSOT/DRY, conceptual integrity, and composition. Use when making architectural decisions, evaluating trade-offs, or understanding *why* best practices exist.

toddlevy/tl-agent-skills0installs0stars

SKILL.md

Full skill instructions

<!-- Copyright (c) 2026 Todd Levy. Licensed under MIT. SPDX-License-Identifier: MIT -->

First Principles of Software Design

The foundational axioms of quality software, traced to their intellectual origins.

"First Principles" carries a deliberate double meaning:

  1. Epistemological — Reasoning from irreducible truths rather than by analogy
  2. Historical — The first people who articulated these principles; the founders

This skill provides the why behind best practices by connecting modern conventions to their intellectual lineage.

When to Use

  • Making architectural decisions with trade-offs
  • Evaluating whether code violates design principles
  • Understanding why a pattern exists, not just how
  • Teaching or explaining software design
  • Code review with principled reasoning

Principles Index

PrincipleFounder(s)Key WorkLink
Information HidingParnas1972Decompose by secrets
Separation of ConcernsDijkstra1974One thing at a time
Abstraction & ContractsLiskov, Hoare1974, 1969Interfaces as promises
Single Source of TruthHunt & Thomas1999Every fact once
Conceptual IntegrityBrooks1975One coherent vision
Fail FastHamilton, Shore1960s, 2004Early detection
Composition Over InheritanceGoF1994Flexible assembly
Explicit Over ImplicitPeters1999Clarity always

Founders Index

NameEraPrimary ContributionLink
David Parnas1970sModularity, information hiding
Edsger Dijkstra1960s-70sStructured programming, SoC
Barbara Liskov1970s-90sData abstraction, substitutability
C.A.R. Hoare1960s-70sContracts, formal reasoning
Frederick Brooks1970sConceptual integrity, system design
Margaret Hamilton1960sSoftware engineering, reliability
Andy Hunt & Dave Thomas1990sDRY, pragmatic practice
Gang of Four1990sDesign patterns

How to Use This Skill

For Decision-Making

When evaluating a design choice, identify which principles are at stake:

"Should we duplicate this validation logic in both services?"

→ Principle: Single Source of Truth (Hunt & Thomas)
→ Risk: Drift when one copy changes but the other doesn't
→ Decision: Extract to shared module or single service

For Code Review

Cite the principle and its lineage when explaining why something matters:

"This component fetches data AND renders AND handles errors.
 That's three concerns in one place.

→ Principle: Separation of Concerns (Dijkstra, 1974)
→ Why it matters: Each concern changes for different reasons
→ Suggestion: Extract data fetching to a hook"

For Learning

Trace modern patterns back to their origins:

React hooks → Composition over Inheritance → GoF (1994)
TypeScript interfaces → Contracts → Hoare (1969), Liskov (1987)
Redux single store → SSOT → Hunt & Thomas (1999)
Microservices → Information Hiding → Parnas (1972)

Convergence Map

These principles emerged from different lineages but converge on the same goal: managing complexity in systems that must change over time.

                         ┌─────────────────────────────────────┐
                         │   MANAGING COMPLEXITY IN SYSTEMS    │
                         │       THAT MUST CHANGE OVER TIME    │
                         └───────────────┬─────────────────────┘
                                         │
          ┌──────────────┬───────────────┼───────────────┬──────────────┐
          ▼              ▼               ▼               ▼              ▼
   ┌─────────────┐ ┌───────────┐ ┌─────────────┐ ┌─────────────┐ ┌──────────┐
   │ MODULARITY  │ │ LEGIBILITY│ │ ABSTRACTION │ │  INTEGRITY  │ │ FEEDBACK │
   │   Parnas    │ │  Dijkstra │ │Liskov/​Hoare │ │   Brooks    │ │ Hamilton │
   └─────────────┘ └───────────┘ └─────────────┘ └─────────────┘ └──────────┘
          │              │               │               │              │
          ▼              ▼               ▼               ▼              ▼
   Information     Separation      Contracts &     Conceptual      Fail Fast
     Hiding        of Concerns    Substitution      Integrity
          │              │               │               │              │
          └──────────────┴───────────────┴───────────────┴──────────────┘
                                         │
                         ┌───────────────┼───────────────┐
                         ▼               ▼               ▼
                  ┌─────────────┐ ┌─────────────┐ ┌─────────────┐
                  │    SOLID    │ │     DRY     │ │   PATTERNS  │
                  │ Uncle Bob   │ │ Hunt/​Thomas │ │     GoF     │
                  └─────────────┘ └─────────────┘ └─────────────┘
                         │               │               │
                         └───────────────┼───────────────┘
                                         ▼
                         ┌─────────────────────────────────────┐
                         │        MODERN PRACTICE              │
                         │  Clean Architecture, Microservices, │
                         │  Functional Core, React Composition │
                         └─────────────────────────────────────┘

Primary Sources

See sources/​primary-sources.md for annotated bibliography with links to original papers and books.


Decision Heuristics

When principles appear to conflict, use these heuristics:

Information Hiding vs Explicitness

Choose Information HidingChoose Explicitness
Implementation will changeInterface is the value
Multiple consumersSingle-use utility
Encapsulation aids testingDebugging requires visibility

SRP vs DRY

Choose SRP (accept duplication)Choose DRY (accept coupling)
Code changes for different reasonsCode truly represents one concept
Teams own different copiesCentral ownership is clear
Coupling cost > duplication costDuplication cost > coupling cost

YAGNI vs Extensibility

Choose YAGNI (build less)Choose Extensibility (build hooks)
Uncertain requirementsKnown variation points
Prototype or MVPLibrary or framework
Single consumerMultiple consumers

Scale Considerations

ContextPrinciple Emphasis
Solo developerConceptual Integrity (one mind)
TeamSeparation of Concerns (parallel work)
PrototypeYAGNI, Fail Fast
ProductionInformation Hiding, Contracts
LibraryLiskov Substitution, Stable APIs
ApplicationComposition, Flexibility

Anti-Pattern Catalog

Anti-PatternViolated PrincipleConsequence
God ObjectSRPChanges ripple everywhere; untestable
Leaky AbstractionInformation HidingConsumers depend on implementation details
Shotgun SurgerySoCOne change requires editing many files
Primitive ObsessionAbstractionScattered validation; unclear semantics
Feature EnvyInformation HidingMethod uses another object's data more than its own
Divergent ChangeSRPOne class changes for multiple unrelated reasons
Parallel InheritanceCompositionAdding one class requires adding another
Dead CodeExplicit over ImplicitConfusion about what's active
Speculative GeneralityYAGNIComplexity for unused flexibility
Inappropriate IntimacyInformation HidingClasses know too much about each other

Diagnostic Questions

  • Does this change for multiple reasons? → SRP violation
  • Could I swap this implementation? → Information Hiding check
  • Is this fact represented once? → DRY/​SSOT check
  • Can I test this in isolation? → Coupling smell
  • Does the name explain the purpose? → Abstraction quality

Principle Interactions

Reinforcing Relationships

graph LR
    SoC[Separation of Concerns] --> SRP[Single Responsibility]
    SRP --> Testability
    InfoHide[Information Hiding] --> Abstraction
    Abstraction --> Substitutability
    Testability --> SRP
    DI[Dependency Injection] --> Testability

Tension Relationships

TensionResolution
DRY ↔ CouplingAccept duplication when coupling is worse
YAGNI ↔ ExtensibilityBuild for known requirements only
Explicitness ↔ HidingHide implementation, expose intent
Performance ↔ AbstractionOptimize measured bottlenecks only

Related Skills

This skill provides the theory. For practice, see:

NeedSkill
Database patternsdrizzle-patterns
Code quality setupcode-quality-setup
Codebase auditcodebase-audit
Complexity assessmenttl-complexity-assessment

References

Quilted Skills

First-Party Academic Sources

Modern Interpretation

Books

Skill References