# 30. Maintenance, Failure, and Repair

## Keeping systems alive without worshipping optimization

**Status:** Outline for future development  
**Primary fields:** Phase, Arc, Form, Lattice, Repair  
**Primary scale:** organism, machine, household, organization, and infrastructure

### Purpose

Study degradation, maintenance, failure detection, repair, throttling, renewal, and graceful decline across living, technical, and social systems.

### Central question

> What must a system repeatedly do to remain viable, and how can it fail without taking everything around it with it?

## Proposed sections

1. **Maintenance is not failure prevention alone** — care, inspection, cleaning, updating, replenishment, and attention.
2. **Failure modes** — acute, chronic, cascading, hidden, intermittent, and socially normalized failure.
3. **Signals before breakdown** — drift, noise, latency, fatigue, conflict, and loss of redundancy.
4. **Repair versus replacement** — reversibility, continuity, salvage, and the politics of obsolescence.
5. **Throttling and graceful degradation** — reducing load before a system crosses a dangerous threshold.
6. **Redundancy and resilience** — buffers, diversity, modularity, mutual aid, and single points of failure.
7. **Postmortems without blame theater** — sequence, conditions, decisions, incentives, and learning.
8. **Care labor and invisible infrastructure** — who maintains what and whose work disappears from the model.
9. **Renewal and release** — when preserving a form blocks transformation or makes harm persistent.

## Five-Foci sketch

- **Phase:** duty cycle, aging, inspection interval, fatigue, recovery, and renewal.
- **Field:** load, resources, environment, dependencies, incentives, and maintenance culture.
- **Arc:** warning, threshold, fault, safe state, repair, redesign, or retirement.
- **Form:** routine, checklist, spare capacity, institution, habit, and remembered failure.
- **Lattice:** component, caretaker, sensor, supplier, user, environment, and downstream effect.

## Practice

Choose one system you depend on. Create a maintenance card with normal state, early warnings, hard stop, safe fallback, repair owner, recovery test, and retirement condition. Include the human labor the system requires.

## Evidence boundary

Resilience is not invulnerability, and maintenance language should not blame individuals for structural failure. A system that survives by exporting harm is not necessarily healthy. Document the boundary and the costs.

## Open questions

- Which early signals are ignored because responding to them is inconvenient?
- How can a system preserve identity while changing form?
- What is the minimum safe state during repair?
- When should Anthrocybernetics recommend release rather than optimization?

## Development sources

Reliability engineering, safety science, ecological resilience, organizational learning, public infrastructure, care ethics, fault-tree analysis, and the existing `11-repair-throttling-and-renewal.md` module.