liveBook · Manning

Technical debt

As explained in Vibe Engineering

The accumulated cost and risk created when expedient software decisions or deferred maintenance make future development and operation more difficult.

A service can run reliably today while becoming harder and riskier to change. Technical debt is the accumulated cost and risk of expedient decisions or deferred maintenance: unnecessary code raises maintenance costs, while aging dependencies and missing tests make future changes more hazardous. section 7.3.2 section 10.6

AI-generated code can make a shortcut particularly tempting. Rapid generation helps with prototypes and boilerplate, but code whose implementation choices are poorly understood can be difficult to debug, maintain, and extend. Speed is useful; it is not evidence that the result is ready for production. section 1.3

A metrics framework that measures too much

An AI-generated tracing solution added custom collectors, services, controllers, analysis, and a separate load simulation rather than relying on established metrics libraries and reusing the existing benchmark. It produced useful measurements, but required more than 800 lines compared with roughly 80 for a standard-library integration. Ten times the code does not necessarily mean ten times the maintenance burden; it does create a larger surface for bugs and cognitive load. The tracing did not measurably affect production latency, so retaining useful instrumentation still matters. section 7.3.1 section 7.3.2

Passing tests can preserve a mistake

A snapshot records what code does, not necessarily what it should do. If a generated parser produces incorrect output when its first golden files are created, later tests can pass while preserving the error. Verify the initial output by other means before using snapshots to detect regressions. section 8.3.5

Deferred upgrades can also compound risk. A dependency with known vulnerabilities may remain pinned because an earlier upgrade broke production and nobody established why. Services without tests, documentation, or knowledgeable maintainers can keep serving customers until an incident or another external demand forces a risky change. section 10.6

Reduce debt by making the tradeoffs visible:

  • Review generated code for unnecessary custom machinery and opportunities to reuse established libraries or existing benchmarks. section 7.3.1
  • Keep instrumentation that helps detect future regressions, but account for its maintenance surface as well as its runtime impact. section 7.3.2
  • Verify expected behavior before accepting snapshots, and treat passing snapshots as change detection rather than proof of correctness. section 8.3.5

How Vibe Engineering covers it

How deep it goes
Examines technical debt through implementation comparisons and production risks, without prescribing a numerical debt calculation.
Examples
Compares custom and library-based tracing, examines unverified parser snapshots, and describes vulnerable dependencies left pinned after failed upgrades.
What you can do after
Evaluate technical debt by examining generated code's maintenance surface, validating test baselines, and recognizing risks in deferred upgrades.
How it connects
Connects technical debt to performance monitoring, regression testing, dependency management, and AI-generated code.

Also explained in

Compare how 16 books explain technical debt →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Vibe Engineering. The section links point to the parts of the book it draws on.