A framework for better government
The Observable
State.
Five ideas for better government.
Begin readingMake the state
observable to citizens.
Introduction
Better institutions.
Better possibilities.
Distribute power.
Align incentives.
Make results visible.
Learn what works.
Build the capacity to deliver.
Modern politics spends enormous energy debating what governments should do. Far less attention is paid to how governments actually work.
Yet institutional design can be just as important as political ambition.
Who gets to make decisions? What incentives shape their behaviour? How do citizens know whether promises have been kept? Why can some governments build infrastructure in years while others take decades? And why do ineffective policies so often survive long after their shortcomings become apparent?
These are not simply questions of ideology. They are questions about how we design the institutions that govern our lives.
Across the world, different countries, cities and organisations have developed remarkably different answers. Switzerland has distributed political authority across cantons and municipalities. Estonia has built digital infrastructure that connects public institutions. Singapore has rethought the delivery of government services. Other places have found ways to encourage local economic development, evaluate policies scientifically, or deliver ambitious projects at unusual speed.
None of these systems is perfect. And their lessons do not need to be adopted wholesale.
But together, they offer inspiration for something better.
The Observable State is a proposal for rethinking how government works, built around five principles:
- 01Give local governments the freedom and incentives to succeed.
- 02Make power visible and accountable.
- 03Build the foundations others can build on.
- 04Test, measure, improve.
- 05Build a state that can deliver.
These principles do not prescribe which political goals a society should pursue. Those choices belong to its citizens, through democratic debate.
They offer a framework for designing institutions that can pursue those goals more effectively.
Distribute power. Align incentives. Make results visible. Learn what works. Build the capacity to deliver.
Above all, the goal is not to make citizens more observable to the state.
It is to make the state more observable to citizens.
Five principles
From principles to practice.
Give local governments the freedom and incentives to succeed.
Decentralise decisions. Align incentives. Let successful ideas spread.
Different places face different challenges. They should have meaningful freedom to develop different solutions.
Municipalities and regional governments should have real authority over the decisions that shape their communities, from planning and infrastructure to public services and economic development. They should also have incentives to improve outcomes and benefit from a reasonable share of the economic value they help create.
When a city attracts investment, builds housing or improves infrastructure, part of the resulting prosperity should help finance its future development.
Local autonomy creates room for experimentation. Constructive competition encourages ambition. Comparable, independently verified results allow citizens to see what works and other communities to learn from it.
The central state should establish shared rules, coordinate issues that cross local boundaries, protect fundamental rights and maintain solidarity between regions. Decentralisation should not mean fragmentation or a race to the bottom.
The goal is not to make every place operate identically. It is to give every place the opportunity, and the incentives, to improve.
Explore real-world inspiration6 examples
Real-world inspiration for principle 1
-
Chinese local governments have historically competed to attract businesses, develop industrial clusters and grow their regional economies. Hefei, for instance, invested municipal capital in BOE's display manufacturing and later backed other technology companies. Robin Rivaton and Ben Southwood explore this model in Works in Progress, describing the role of local autonomy, financial incentives and competition between cities. The mechanisms are interesting even if the model's political structure, debt risks and performance incentives cannot simply be transplanted into democracies.
-
Beginning in 1989, Porto Alegre introduced participatory budgeting, allowing residents to influence how parts of the city's investment budget were allocated. Community assemblies helped identify priorities for infrastructure and public services, including underserved neighbourhoods. Studies of its early years document the redistribution of investment toward poorer areas. The experience illustrates a form of decentralisation that gives citizens themselves a direct role in public decisions.
Research — Participatory Economic Democracy in Porto Alegre · Open-access paper — University of Utah (PDF)
-
United Kingdom
Making local performance comparable
The Local Government Association's LG Inform platform allows local authorities to compare service outcomes, finances and contextual indicators with those of similar councils. Rather than requiring every municipality to use the same approach, comparable information helps identify differences in performance and opportunities for learning. It illustrates how shared measurement can support decentralised government.
-
Hong Kong's MTR Corporation uses a Rail + Property model, combining railway development with property development rights around stations. Capturing part of the increase in land value generated by better transport connections helps finance infrastructure and supports integrated urban development. A 2024 Asian Development Bank Institute study by Ka Ying Wong and K. E. Seetharam explains the model and the particular land, legal and financial conditions that enable it.
Asian Development Bank Institute — The Rail-Plus-Property Model
-
Private Japanese railway companies such as Tokyu combine transportation with housing, retail, commercial property and other businesses along their lines. This allows them to benefit not only from ticket revenue but also from the economic activity made possible by their networks. Matthew Bornholt and Benedict Springbett examine in Works in Progress how this vertically integrated model, combined with permissive land-use rules, helped shape Japan's railway system and urban development.
Make power visible and accountable.
Citizens should understand what politicians stand for, what governments do and whether public institutions deliver.
Democracy requires more than periodic elections.
Citizens should be able to compare candidates' positions before voting, follow the decisions elected representatives make, understand how public money is spent and see whether political commitments translate into results.
Major public projects should have accessible records of budgets, responsibilities, milestones, delays and outcomes. Public contracts and conflicts of interest should be open to scrutiny. This information should be independently verifiable rather than controlled exclusively by the institutions being evaluated.
But greater government transparency must not come at the expense of individual privacy.
Citizens should be able to see how institutions use their personal information, while public digital systems should collect only what is necessary. Where possible, individuals should be able to prove an attribute or eligibility without unnecessarily revealing their complete identity.
Institutions should be transparent to citizens. Citizens should not have to be transparent to institutions.
Explore real-world inspiration6 examples
Real-world inspiration for principle 2
-
Switzerland
Understanding what electoral candidates believe
Developed by the nonprofit Politools, smartvote allows voters to answer policy questions and compare their positions with those of candidates. Its smartspider visualisations present political views across multiple dimensions instead of reducing them to a left-right spectrum. During Switzerland's 2023 federal elections, more than 84% of candidates participated and the platform generated over 2.1 million voting recommendations.
-
Switzerland
Making parliamentary behaviour observable
Also developed by Politools, smartmonitor maintains a database of Swiss parliamentary activity and voting records. It makes it possible to analyse how elected representatives behave during their mandates. Together, smartvote and smartmonitor illustrate two complementary aspects of accountability: understanding what candidates say before elections and scrutinising their legislative activity afterwards.
-
Starting in 2003, Brazil randomly selected municipalities for audits of federally transferred funds and published the findings. Researchers Claudio Ferraz and Frederico Finan compared municipalities whose audit results became public before and after the 2004 elections. They found that disclosure significantly affected incumbents' electoral performance, especially where local radio communicated the findings. The experiment provides empirical evidence that public information can change democratic accountability.
Ferraz & Finan — Quarterly Journal of Economics · Open full text — authors’ copy (PDF)
-
Estonia's Data Tracker allows citizens to review certain operations performed on their personal information by participating public-sector systems. Accessible through the national eesti.ee portal, the tool provides visibility into how connected databases use personal data. Rather than simply asking citizens to trust public institutions, it gives them a practical way to scrutinise aspects of their information being accessed.
-
Ukraine's Prozorro system makes public procurement information accessible through an electronic platform. Businesses can compete for contracts, while journalists, citizens and independent organisations can examine tenders, suppliers and awards. Its design combines a public central database with connected commercial platforms, demonstrating how transparency and competition can reinforce each other.
Prozorro — About the System · Ukraine Digital State — Prozorro
-
Launched in 2015, vTaiwan brings citizens, businesses, civil society and government officials into structured discussions on public-policy questions. It has used Pol.is to identify points of agreement between participants with otherwise divergent views. A 2015 consultation on regulating ride-hailing platforms, including Uber, helped inform subsequent policy discussions. Rather than merely counting votes or amplifying opposing positions, the process helps make possible areas of consensus visible.
Test, measure, improve.
Policies should be judged by their results, not merely by the intentions behind them.
Good intentions do not guarantee good outcomes.
Every major public programme should begin with a clearly defined problem, measurable objectives and a credible plan to evaluate its effects.
Where uncertainty is high and risks are manageable, governments should test different approaches before committing to large-scale implementation. Independent evaluation should help distinguish genuine improvements from changes that would have occurred anyway.
Results should then influence decisions. Successful interventions should be expanded. Partially effective ones should be refined. Ineffective policies should be reconsidered or discontinued.
The same logic should apply to regulation. New technologies may be better understood through supervised experimentation than through rules designed entirely around predictions made in advance.
And existing regulations should not survive indefinitely simply because they were adopted in the past.
Democratic debate determines which goals society wishes to pursue. Evidence helps determine whether the chosen interventions achieve them.
Public policy should be an ongoing process of learning, not a permanent defence of previous decisions.
Explore real-world inspiration6 examples
Real-world inspiration for principle 4
-
United Kingdom
Testing educational interventions before expanding them
The Education Endowment Foundation funds rigorous evaluations of interventions designed to improve educational outcomes. Its Maths Champions programme, which trains nursery staff to improve early mathematics teaching, has undergone two trials. The larger evaluation found an average of three additional months of progress in maths, and the programme subsequently received funding for wider delivery. It illustrates how scientific evaluation can inform the decision to expand a public intervention.
-
Mexico's Progresa programme, launched in 1997, provided conditional cash transfers intended to improve education, health and nutrition. Its phased rollout enabled researchers to compare outcomes between communities receiving support at different times. Follow-up studies found that some children who gained earlier access remained in education longer and earned more as adults. The experiment helped establish an influential model of rigorous evaluation in social policy.
-
United Kingdom
Improving tax collection by testing small changes
The Behavioural Insights Team worked with HM Revenue & Customs to conduct randomised experiments on overdue-tax reminder letters. Research involving more than 200,000 individuals found that messages referring to social norms and public services increased payment rates compared with standard reminders. An inexpensive administrative intervention was tested in real conditions, measured and improved rather than justified by intuition alone.
Hallsworth et al. — Journal of Public Economics · Open-access manuscript — University of Warwick (PDF)
-
United Kingdom
Testing innovation under regulatory supervision
Since 2016, the Financial Conduct Authority's Regulatory Sandbox has allowed eligible financial firms to test innovative products with real users under defined safeguards. Its early cohorts demonstrated how regulators could gather direct evidence about emerging business models while enabling controlled experimentation. The approach offers an alternative to treating every unfamiliar technology as something that must be fully regulated before it can be tested.
Financial Conduct Authority — Regulatory Sandbox Lessons Learned
-
South Korea introduced regulatory sunset mechanisms in 1998 and expanded them in 2009 to include scheduled reassessment. Certain regulations must be reviewed after a predefined period, often three years, to determine whether they remain necessary or effective. The OECD notes that actual expiration is less common than scheduled review, but the institutional principle is important: regulations should not automatically remain in force forever.
-
Washington State's Institute for Public Policy reviews academic research on interventions across areas such as education, criminal justice and public health. It combines estimates of effectiveness with implementation costs and projected social benefits, allowing policymakers to examine which programmes are likely to deliver value. Its public benefit-cost model makes trade-offs more explicit, including uncertainty about whether an intervention's benefits will exceed its costs.
Washington State Institute for Public Policy — Benefit-Cost Analysis
Build a state that can deliver.
Attract capable people. Give them responsibility. Make it possible to build.
A government should be judged not only by the quality of its decisions, but by its ability to turn them into results.
That begins with the people who run public institutions.
Government should be capable of attracting talented professionals, recognising competence and giving experienced leaders meaningful responsibility. Career progression should reflect performance, while strong rules on conflicts of interest, transparency and independent oversight protect institutional integrity.
But capable people also need capable systems.
Housing, energy infrastructure, transport networks and public facilities should not be obstructed by unnecessary layers of approval, fragmented responsibilities or unpredictable decisions.
Clear rules, coordinated reviews, firm deadlines and accountable project leadership can dramatically improve the ability to deliver.
Governments should also be willing to create new institutions, adopt different operating models and give ambitious programmes room to experiment when conventional structures prove inadequate.
Speed is not an end in itself. Legal protections, public participation, environmental safeguards and safety standards remain essential.
The objective is not simply a faster state. It is a state that can turn collective decisions into tangible results.
Explore real-world inspiration8 examples
Real-world inspiration for principle 5
-
Singapore's civil service uses performance-linked remuneration, merit-based progression and periodic salary reviews intended to keep public employment competitive. In February 2026, its Public Service Division announced adjustments for approximately 22,000 eligible officers, explicitly citing the need to attract and retain capable people. The approach illustrates an institutional commitment to the quality and motivation of public servants, although its precise contribution to government outcomes is difficult to isolate.
-
CORENET X restructures the building approval process by consolidating more than 20 regulatory touchpoints into three coordinated gateways. Agencies can review projects in parallel and share clearer feedback. In 2025, Singapore reported that 97% of participating pilot projects could achieve regulatory time savings of up to 20%, with some saving as much as two months. Since October 2026, CORENET X has been mandatory for new projects with at least 5,000 square metres of gross floor area.
Building and Construction Authority — CORENET X Results · BCA — Updated Implementation, July 2026
-
Between 1995 and 2003, Madrid carried out major expansions of its metro network at unusually low costs compared with many international projects. A detailed Eno Center for Transportation study identifies the contribution of experienced in-house engineering teams, dedicated delivery structures and political commitment. Favourable geology and lower labour costs also mattered. The experience illustrates the importance of retaining technical expertise and organisational learning inside institutions responsible for major projects.
-
Following the disruption of Russian gas supplies in 2022, Germany passed the LNG Acceleration Act to streamline approvals for specified energy infrastructure. A new floating LNG import facility at Wilhelmshaven opened in December 2022, following construction and coordination carried out over a matter of months. The case illustrates how clearly defined national priorities, streamlined procedures and coordinated delivery can accelerate strategically important projects. Its exceptional legal treatment and environmental trade-offs limit how directly the model can be applied elsewhere.
German Federal Ministry for Economic Affairs — Wilhelmshaven LNG Terminal
-
Japan's nationally defined zoning system establishes many land-use and development rules in advance. Projects that comply with applicable zoning, building and technical requirements generally have a defined route to approval, rather than requiring a fresh discretionary political decision on their desirability. This reduces uncertainty for developers and gives greater predictability to investment and construction. The Centre for Cities examines the system in comparison with more discretionary planning regimes.
-
United Kingdom
Creating a different kind of research institutionEmerging experiment
The Advanced Research and Invention Agency was created to fund ambitious, high-risk scientific research through an operating model designed around autonomy and empowered Programme Directors. In its 2025–26 annual report, ARIA reported commitments of over £500 million to programmes supporting more than 200 research and development teams. Its ultimate scientific impact remains uncertain, but the agency demonstrates how governments can experiment with new institutional structures rather than assuming that existing funding systems are the only available model.
-
South Korea
Making administrative inaction consequentialEmerging experiment
In 2026, South Korea enacted legislation designed to accelerate the development of AI data centres. It includes coordinated permit processing and a timeout mechanism under which certain authorisations may be deemed granted if the administration fails to respond within the required period. The promulgated law is scheduled to enter into force on 10 March 2027, so its results remain untested. The principle is to prevent administrative silence from becoming an indefinite veto while retaining legally defined safeguards.
South Korean Ministry of Science and ICT — AIDC Special Act · Official Legislation
-
Dubai
Automating permits when compliance can be verified by softwareEmerging experiment
In August 2026, Dubai Municipality announced a project to develop an AI-powered system capable of examining submitted architectural and engineering drawings, checking compliance with the Dubai Building Code and automatically issuing permits for certain categories of villas. The plan involves development, pilot testing and eventual rollout. Its promised time savings are not yet demonstrated at scale, but it illustrates the possibility of automating routine rule-checking while retaining human responsibility for exceptional cases.
An open-ended proposition
A state
that learns.
No country has perfected government.
And no institutional model can be copied without considering the democratic, legal, cultural and economic context in which it operates.
But useful ideas can be found almost everywhere.
The examples above come from democracies and other political systems, from large nations and small city-states, from contemporary reforms and long-established institutions. What matters is not adopting any of these systems in their entirety, but understanding the mechanisms that help explain their results.
Together, the five principles point towards a different approach to governance.
Give local institutions room to experiment and incentives to succeed. Make political power and public performance observable. Build common infrastructure that enables citizens and businesses to act. Test policies against reality, and develop institutions capable of turning decisions into results.
Above all, create a system in which success can be recognised, shortcomings can be corrected, and effective ideas can spread.
Democracy determines what societies want to achieve.
Better institutional design helps them achieve it.
The state is one of humanity's most important inventions. Like every important system, it should be capable of learning and improving.
Not a state that observes everything. A state that can be observed, understood, held accountable and improved.