Read in this order
For when the problem is no longer naming the harm but asking whether the system has any real stopping points.
- Institutions Should Be Built to Die We know how to found an agency and have no way to end one. Shutting it down feels like a verdict, so it drifts on and traps the people inside.
- You Can Design Harm Out Most harm is not malice. It is the cheapest way for a system to stay stable, and cheap can be redesigned.
- Anarchists Want Institutions Without Masters An institution can be large and durable if it absorbs shocks instead of extracting obedience.
- How to Build Institutions That Can't Coerce Institutions claim care while quietly discarding the people who need the most help. Four principles for systems that structurally cannot coerce.
- The Right to Fuck Up We replaced repair with disposal, and an employer that can only remove the people who err never finds out what went wrong.
- You Can Hit Every Target and Lose the Point Hospitals, platforms, universities, and economies all improve the metrics that define success while quietly eroding the conditions that make those metrics meaningful.
- Guilty Until Verified When institutions treat every user as an adversary, compliance becomes evasion.
Questions this pattern asks
- If this goes wrong, who can undo it, and how fast?
- Where are the real stopping points, rollback mechanisms, or repair obligations?
- Does correction depend on harmed people carrying all the proof, labor, and timing costs themselves?
How this pattern runs
Procedural Erosion
Narrates itself as modernization, streamlining, alignment with best practices.
Did a decision move from a public meeting to an internal review?
Vocabulary
- Institutional apoptosis after Hannan and Freeman (1984)
- Evidentiary inversion
- Andon cord after Toyota Production System
- Recalibration bond
Where to go next
Part of the archive’s answer to the four defenses: give the people who carry a cost the power to make the arrangement change.