Code Complete vs Refactoring

Status: reviewed Research basis: mini-only

Verdict: ✅ Complementary

Conflict: 12% Overlap: 38% Complementarity: 78%

Loading Decision

Use together when changing existing code: one rule set controls safe change sequencing while the other defines the target design, construction, architecture, data, or production quality.

Book A Pressure

• Code Complete should drive tasks where defect reduction, data clarity, defensive checks, evidence-based debugging, and reviewability dominate.
• Evidence: code-complete/code-complete.mini.md lines 3-5: applies to implementation, change, review, debugging, refactoring, and tuning of production code.

Book B Pressure

• Refactoring should drive tasks where behavior-preserving structural change and current-smell scope control dominate.
• Evidence: refactoring/refactoring.mini.md lines 3-5: applies when changing existing code, preparing a feature/bug fix, reviewing cleanup, or reducing structural friction without changing observable behavior.

Complementary Forces

• Claim: Code Complete contributes defect-reduction, data-clarity, defensive-check, evidence-based-debugging, and reviewability pressure; Refactoring contributes behavior-preserving, small-step, test-backed refactoring pressure. Together they are useful only where both scopes are active.
• Evidence:
  • code-complete/code-complete.mini.md lines 35-47: fires when solution-first coding, hard-to-name routines, hidden data meaning, trust boundaries, complex control flow, repeated mappings, god objects, weak tests, guess-based debugging, risky refactoring, performance work, stale comments, or local style drift appear.
  • refactoring/refactoring.mini.md lines 13-26: requires observable behavior preservation, small reversible steps, safety nets, preparatory/follow-up refactoring around feature work, current-smell focus, simplest named moves, intent-revealing names/functions, behavior and state with owners, explicit data/mutation/contracts, honest conditional simplification, evidence-based abstraction, preserved error semantics, reviewable patch intent, and stop conditions.

Overlap

• Claim: They overlap where both affect safe existing-code change, tests, behavior preservation, ownership, and stopping before speculative cleanup; the overlap score reflects how often an agent would receive similar pressure from both.
• Evidence:
  • code-complete/code-complete.mini.md lines 51-56: checks requirements, architecture fit, construction approach, readable code structure, deliberate inputs/errors/invariants, inspectable flow, evidence-based validation, and reviewable change size.
  • refactoring/refactoring.mini.md lines 43-49: checks behavior preservation, separated structural/behavior/test updates, safety net, real friction removed, clearer ownership/control/data/interfaces, reviewable runnable patch, and stopped cleanup.

Conflicts

• Claim: The tension is scope creep: design or architecture improvements must not override behavior preservation, characterization, or the current-smell stop condition.
• Evidence:
  • code-complete/code-complete.mini.md lines 7-9: corrects accidental construction by choosing lower defect risk and easier reasoning over clever idioms.
  • refactoring/refactoring.mini.md lines 7-9: corrects cleanup turning into rewrite, hidden feature change, or speculative architecture.

Use Together When

• Use together when existing code must be reshaped toward Code Complete goals without changing observable behavior or turning cleanup into redesign.

Prefer One When

• Prefer the refactoring rule set when observable behavior must stay unchanged; prefer the other book when designing new behavior rather than reshaping existing structure.

Source Basis

• code-complete/code-complete.mini.md lines 3-5: applies to implementation, change, review, debugging, refactoring, and tuning of production code.
• code-complete/code-complete.mini.md lines 7-9: corrects accidental construction by choosing lower defect risk and easier reasoning over clever idioms.
• code-complete/code-complete.mini.md lines 35-47: fires when solution-first coding, hard-to-name routines, hidden data meaning, trust boundaries, complex control flow, repeated mappings, god objects, weak tests, guess-based debugging, risky refactoring, performance work, stale comments, or local style drift appear.
• code-complete/code-complete.mini.md lines 51-56: checks requirements, architecture fit, construction approach, readable code structure, deliberate inputs/errors/invariants, inspectable flow, evidence-based validation, and reviewable change size.
• refactoring/refactoring.mini.md lines 3-5: applies when changing existing code, preparing a feature/bug fix, reviewing cleanup, or reducing structural friction without changing observable behavior.
• refactoring/refactoring.mini.md lines 7-9: corrects cleanup turning into rewrite, hidden feature change, or speculative architecture.
• refactoring/refactoring.mini.md lines 13-26: requires observable behavior preservation, small reversible steps, safety nets, preparatory/follow-up refactoring around feature work, current-smell focus, simplest named moves, intent-revealing names/functions, behavior and state with owners, explicit data/mutation/contracts, honest conditional simplification, evidence-based abstraction, preserved error semantics, reviewable patch intent, and stop conditions.
• refactoring/refactoring.mini.md lines 43-49: checks behavior preservation, separated structural/behavior/test updates, safety net, real friction removed, clearer ownership/control/data/interfaces, reviewable runnable patch, and stopped cleanup.

Review Notes

• External context was not used as decisive evidence for Code Complete vs Refactoring; the verdict is based on the cited local mini line ranges.