Files
trust-issues/src/content/stories/repair-is-a-test.md
T
Compleet 0e7871f395
validate / validate (push) Successful in 22s
Restructure navigation around What's up / Opine and the reader's utilities
- Stories gain a required type (report | essay); all 28 classified per
  the boundary rule now stated in EDITORIAL.md
- Nav: What's up · Opine · Podcast · Book · Guides · Forum, each gated
  on real content (Forum waits on a forumUrl); logo is Home
- /whats-up (reports, newest first) and /opine (essays, editorial order)
  share a StoryArchive component with subject filters
- Static /tags/<tag> pages for subjects carried by 2+ stories; tags stay
  overlapping, never exclusive departments
- /corrections: generated from revision history (commits beginning
  'Correction:'), so the page cannot under-report
- About, RSS, Corrections, Weekly briefing move to the footer

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-14 14:37:44 +01:00

2.4 KiB

title, type, description, published, editorialOrder, reviewed, author, maintainers, status, featured, homepage, section, draft, tags, trustPattern
title type description published editorialOrder reviewed author maintainers status featured homepage section draft tags trustPattern
Repair is a test, not a moral duty essay Trying to fix a system can generate the evidence that the system itself—not one parameter—is wrong. 2026-07-14 25 2026-07-14 Ana
Ana
developing false false Repair false
repair
institutions
failure
change
claim trusted failure
A flawed institution should be repaired because replacement destroys accumulated trust and knowledge. Participants must trust that repair attempts are bounded, measured, reversible, and capable of questioning the institution's basic categories. Endless repair can force vulnerable people to subsidise a structure whose fixes contradict one another.

"Repair, do not replace" sounds mature. It respects accumulated knowledge, relationships and infrastructure. Replacement is expensive and often produces its own naïve failures.

"Keep trying" can also become a sentence imposed on the people already absorbing harm.

A useful repair has boundaries. Change one mechanism, observe what happens, and preserve evidence. Did errors decrease? Did access improve without moving the burden somewhere less visible? Did the fix survive outside the pilot?

When repairs conflict

Sometimes every local fix breaks another part of the system. Stronger identity checks reduce fraud but exclude legitimate users. Easier recovery restores access but increases coercion and account takeover. More transparency produces privacy loss. Centralisation improves consistency while removing local judgement.

Trade-offs are normal. But repeated, mutually incompatible repairs can reveal that the system is using the wrong categories. Perhaps the problem is not the accuracy threshold but making one identity check govern every service. Perhaps the global reputation score cannot be made contextual because globality is the defect.

Repair attempts then become probes. Their failure is not wasted effort if it shows why the current structure cannot hold all required values at once.

A finite obligation

Before asking people to remain inside a repair process, state:

  • what will be tried;
  • what evidence counts;
  • how long the attempt lasts;
  • which harms trigger an immediate stop;
  • what replacement or exit becomes available after failure.

Repair deserves a chance. It does not deserve infinity.