Institutions Should Be Built to Die

We are good at founding institutions, agencies, committees, and regulations, and we have no routine mechanism for retiring them. There’s no process that is normally expected to run, the way a sunset clause or a review mandate occasionally does. We treat governance as permanent accumulation. A problem shows up, we build a structure to solve it, the problem evolves, the structure stays. It accumulates staff, budget and procedural weight. None of this is a claim that no institution has ever been retired; it’s a claim about the default: retirement is exceptional, ad hoc, and usually political rather than procedural.

Layers accumulate: older solutions under newer ones, each still half in force, slowing everything above it.

Think of a housing crisis layered over 1970s zoning, 1990s environmental carve-outs, and 2008-era foreclosure relief, each one still partly in effect.

Biology solved this a long time ago.

In a healthy organism, death is a maintenance function. The process is apoptosis: programmed cell death.

When a cell is damaged or just obsolete, it doesn’t wait to be killed by external trauma. Instead it runs a specific sequence. It dismantles its own machinery, packages its resources, and signals its neighbors to recycle its matter. The exit is clean and the surrounding tissue stays stable.

Diagram

A failing homelessness office, wound down or left to collapse

View source

%% title: A failing homelessness office, wound down or left to collapse
%% caption: If the directorate’s charter carries an end date, five years of rising homelessness end its funding without a vote, and its assets, staff and data go somewhere useful. Without one, it defends its budget, its rules pile up under newer ones, and it ends in crisis.
flowchart TD
  classDef actor stroke-width:1.6px;
  classDef system stroke-width:1.2px;
  classDef gate stroke-width:1.6px;
  classDef binding stroke-width:3px,font-weight:bold;
  classDef status stroke-dasharray:4 3;
  classDef burden stroke-width:2px;
  classDef repair stroke-width:2.4px,font-weight:bold;
  classDef failure stroke-width:2.2px,stroke-dasharray:6 3;

  O["Homelessness directorate<br/>Five years of rising numbers"]:::system --> P{"Does its charter<br/>carry an end date?"}:::gate

  subgraph Apoptosis["Apoptosis: Wound Down on Schedule"]
    direction TB
    P == "end date written in" ==> A1["Missed metrics end<br/>the funding, no vote"]:::repair
    A1 --> A2["Assets to nonprofits,<br/>staff to a talent reserve"]:::repair
    A2 ==> A3(["Data archived, orderly exit"]):::binding
  end

  subgraph Necrosis["Necrosis: Kept Alive Until It Breaks"]
    direction TB
    P -. "no end date" .-> N1["Directorate defends<br/>its budget and staff"]:::status
    N1 --> N2["Its rules stay half in force<br/>under newer ones"]:::burden
    N2 ==> N3(["Crisis: lost jobs, vanished services"]):::failure
  end
  Apoptosis ~~~ Necrosis

The alternative is necrosis: traumatic, inflammatory rupture. Or worse, the cell refuses to cycle out at all, replicating without purpose until it destabilizes the host. In biology we call that uncontrolled growth. In politics we call it bureaucracy, but the mechanic is the same. The part puts its survival ahead of the health of the whole.

We’ve built a specific fragility into our society by designing institutions that can only end via necrosis (collapse, crisis) rather than apoptosis (resolution).

Coding for mortality

If apoptosis is how a body stays healthy, we need the equivalent in law. That means moving from statutory permanence to statutory provisionality: nothing runs forever by default.

Expiration by default. Every charter, agency, and program carries its own conclusion, written in at the start. Not a vague “review in five years.” A real sunset. The default state of any government program should be winding down unless it can affirmatively show it is still doing necessary work.

The burden of proof. Right now, closing a zombie agency takes enormous political capital. Inertia protects the status quo. Apoptotic design flips the friction: renewal is what takes effort. If a directorate created to reduce homelessness watches homelessness rise for five years, it shouldn’t take a special vote to defund it. The funding falls off because the metrics were missed.

The graceful exit. We fear institutional death because we picture a void: lost jobs, vanished services. That is necrosis. An apoptotic ending is planned. Assets get transferred or sold to nonprofits and local actors. Civil servants move into a talent reserve for new projects instead of being scattered. Data gets archived and made usable, and the institution winds down in an orderly way.

Governing with endings

The resistance is mostly cultural. We treat endings as moral verdicts. Shutting down an agency looks like admitting it was a mistake. Sunsetting a law looks like disowning the values that created it.

So we pick the comfortable drift of irrelevance over the clarity of a conclusion. We tell ourselves this is compassionate. It is not—not for the people trying to live and work inside these structures. There is nothing compassionate about a public forced to navigate systems that no longer fit the world. There is nothing kind about trapping talented workers inside organizations that can’t succeed.

The 20th century built institutions that could last forever. Institutions built now need their ending written in at founding. I don’t know of one designed that way from the start, and the ones I have watched end did it badly.1

Put in legal terms, an agency’s authority would run for a fixed term and be renewed only under conditions written down in advance. An institution that cannot retire its own failures ends up treating its power as permanent property. Oversight bodies can harden into bureaucracies too. The offices that review and correct other agencies would carry their own sunset clause, expiring unless they can show the people they govern that they are still needed.2