State capacity matters.
Rights and services that cannot be implemented remain weak promises. Competent staff, fiscal resources, legal authority, data, procurement, operations, maintenance and institutional memory matter.
Summary judgment
Keep subsidiarity as a design heuristic, not an empirical law. The stronger rule asks whether an institution can perform the job and remain answerable while protecting rights and cross-boundary interests.
Human Scale does not want weak government. It wants government strong enough to perform legitimate tasks and constrained enough to remain answerable to human beings.
What survives
Administrative burden is real.
Learning what a program requires, gathering documents, traveling, waiting, logging in, proving eligibility and appealing errors consume time and psychological capacity. Some burden protects legitimate goals; some exists because systems are badly integrated.
Local knowledge can improve decisions.
Residents and frontline workers often know details distant institutions cannot see. Local knowledge can also be biased, incomplete or dominated by organized insiders.
Participation can add information and legitimacy—and be captured.
Meetings and consultations can overrepresent people with time, status, language fluency and strong existing interests. Participation should be evaluated for representativeness and actual influence.
Subsidiarity needs explicit reasons to scale up.
Pollution, epidemics, grids, civil rights, insurance, research, national defense and financial systems cross local boundaries. Local regardless of problem structure is not Human Scale.
Large institutions can purchase real professional capacity.
Scale can support specialized expertise, cybersecurity, procurement, infrastructure, standards and surge capacity while also creating remote responsibility and bureaucracy.
Transparency is not accountability.
Publishing data does little if citizens still cannot tell what happened, who decided, why, at what cost, and what they can do next.
Digital government can reduce friction and create a digital gate.
Online services can eliminate travel and office-hour limits while creating lockout, identity, device, language, accessibility and human-support problems.
Metrics can reveal performance and distort it.
Wait time, cost, safety and error rates help, but institutions can game metrics or optimize them instead of the mission.
Corruption and arbitrary enforcement undermine capability and legitimacy.
Anti-corruption controls should target actual risks without turning every public process into universal suspicion and paperwork.
Trust should be calibrated to trustworthiness.
High trust in a corrupt institution is bad; low trust in a competent one can damage cooperation. Build performance worthy of trust rather than merely public relations.
Emergency powers can be necessary and dangerous.
Fast coordination may require temporary concentration of authority. Defined triggers, scope, review, time limits, reporting and termination help a system return to normal governance.
AI can reduce bureaucracy and obscure responsibility.
Translation, search, triage and prefilling can help; hallucinated rules, inconsistent decisions and automated loops with no responsible human can make government less legible.
Recommended chapter changes
Core: Put power at the smallest scale that can solve the problem competently, protect rights, internalize important spillovers, and remain accountable.
Two-axis test: Capability — can it do the job? Accountability — can affected people understand, challenge, review or constrain it?
Participation: Treat public input as information and legitimacy, not guaranteed representativeness or technical correctness.
Burden budget: Estimate citizen time, forms, documents, travel, logins, waiting and appeals when designing a policy or service.
Digital infrastructure: Reuse identity, payments, notifications, status tracking and accessible data exchange where appropriate without creating one unchallengeable portal.
Transparency: Design around answerable questions: who decided, under what authority, what evidence, what cost, what happened, and how to appeal.
What Chapter 8 gets right
Power should be understandable. Local knowledge matters. Rights can require larger institutions. Government needs real capacity. Bureaucracy should be simpler for citizens. Digital tools should make institutions legible. Emergency powers need limits. Corruption and opaque responsibility matter.
The chapter is weakest wherever “smallest scale” can be mistaken for “best scale regardless of capacity, capture, rights or spillovers.”
What would change our minds?
We should revise favored governance rules if local participation systematically worsens outcomes through capture; centralized professional systems repeatedly outperform local discretion without major accountability losses; burden reduction materially increases fraud or safety failures; digital-first systems exclude vulnerable populations; or AI assistance reliably improves accuracy, accessibility and appeal without hiding responsibility.
Result of Audit 08
Put public power at the smallest scale that can solve the problem competently, protect rights, internalize important spillovers, and remain accountable.Build larger institutions when scale buys real capability, and stronger review when scale makes power harder to see.