State as Operating System: Citizens as Tenants

Modern states resemble long-running more closely than machines or markets. Not because a parliament is a , but because both are arrangements that stay switched on for years, hand out resources, provide the floor everything else stands on, and quietly accumulate programs that then refuse to leave. You can judge the arrangement by the same questions you would ask of any machine: who can install, who can remove, and how much does it cost to sit idle?

The Long-Running Windows Install

The dominant pattern is a Windows install that shipped in 2009 and never stopped adding software. It arrives with a pile of defaults, runs processes nobody ordered, phones home, enforces its own update schedule, and makes removal expensive enough that most people give up. Idle cost, the price of merely being on, stays high. Most high-income administrative states match this shape. An agency is created, a reporting regime is layered on top, a compliance system is bolted to that, and each addition is justified as coordination or protection. The cost floor rises even when no new service reaches a citizen. Subtraction is rare enough to be a scandal when it happens.

I am a tenant, a person living inside a configuration I did not write, in a building I am not allowed to renovate.

Coherence Is Not Sovereignty

Polished closed systems are the other temptation. macOS and iOS feel finished because you are not permitted to open them. The boot chain is signed, the core services are not fully inspectable, and the vendor keeps unilateral power over what runs. In exchange you get coherence: fewer rough edges, less obvious junk, a machine that behaves like an appliance. A technocratic state can feel the same way. Reliable services, tidy dashboards, a competent class that would rather you did not argue with the architecture. It is pleasant right up until the moment you need to contest something and discover the appeal button does not exist.

Mobile platforms add the next layer: persistent identity, sensors that stay available, and one store that decides what you may install. That maps neatly onto states that have instrumented payments, movement, and identity so thoroughly that living outside the official stack stops being practical.

Deliberate Composition

Accretion piles on. Deliberate composition declares, rebuilds, and can roll back.

The alternative is composition on purpose. Arch starts with almost nothing and makes me choose every addition. goes further: the whole machine is described in one configuration, a rebuild produces exactly that state, and previous generations stay on disk, so yesterday is still there when today is wrong. Undeclared services do not run. gives curated defaults while leaving the system open enough to inspect and change. A state built this way would behave the same: every agency, rule, and data system present because it was deliberately added to a readable configuration, and removable from it. New functions would need explicit addition and visible funding, and the whole institutional set could be rebuilt from a known, smaller baseline.

Government-as-a-Platform has the right instinct about shared components but the wrong default about growth. It still assumes a permanent central platform that accumulates. Modular blocks beat silos, where offices cannot see each other, but they do not touch baseline accretion unless removal is normal.

Bloat Follows Bloat

Administrative bloat follows software bloat. Systems that add more readily than they remove will grow, no matter how elegant the org diagram looks. Only a change in the composition method, the rule for how a thing is allowed to exist in the first place, produces a lasting reduction. The question is never “how do we trim this?” It is “who is allowed to add, and what happens when someone wants to take it away?”

Further reading: Build Canada¹, public dashboards of government spending.

Build Canada — Projects. Their project list, including the public spending dashboards. ↩