Governance at the Right Scale – Decentralisation Must Serve People, Not Institutions

 

Modern governance relies heavily on indicators, targets and performance metrics. Quantification makes comparison, monitoring and coordination possible, but it also creates systematic blind spots. Social processes that are difficult to measure including care, trust, legitimacy, informal power, knowledge quality and long-term systemic risks, receive less institutional attention. Phenomena that are difficult to represent numerically tend to receive less regular attention, weaker budgetary ownership, and less influence in performance review, even when professionals and affected communities continue to experience them as important[1].

The central issue is not whether public institutions should measure, but how measurement is governed. Indicators select a purpose, define a category, privilege certain forms of evidence, and create incentives for those being assessed. A responsible system therefore treats every measure as a partial representation: useful for a stated decision, open to qualitative and experiential correction, traceable to its assumptions, and removable when it distorts the practice it was meant to improve.

Measurement is not neutral. Decisions about what to measure, which categories to use and how results should be interpreted determine what institutions perceive as relevant and governable. Indicators also shape behavior: once organizations are assessed through particular metrics, they begin optimizing their activities around those measures. When an indicator becomes a target, it may cease to represent the underlying reality accurately. Measurable performance can improve while the actual social problem remains unresolved or worsens.

This dynamic produces several forms of institutional blindness. Unpaid and relational care remains largely absent from conventional economic statistics[2]. Informal influence, lobbying and network power are difficult to quantify. The reliability of knowledge, the effects of disinformation and trust in institutions resist simple measurement. Slow-moving ecological, social and financial risks may remain invisible until they become acute. As a result, policy may prioritize easily measured outputs over justice, resilience and lived experience.

Indicators should therefore be broadened beyond production, productivity and cost. Health, well-being, social connection, mental welfare, care capacity, institutional trust and perceived justice can provide a more complete picture of social development. Measuring the quality and distribution of care can reveal burdens concealed by conventional statistics. Surveys, participatory evaluations and qualitative research can provide insight into legitimacy and social support[3].

Broader measurement nevertheless creates methodological and administrative challenges. Concepts such as trust and well-being have contested definitions, and excessive numbers of indicators can make governance unmanageable. A carefully selected dashboard is therefore preferable to reducing complex realities to one composite score[4]. Presenting several dimensions alongside one another preserves trade-offs and tensions that would otherwise disappear through aggregation.

Quantitative and qualitative knowledge must also be combined. Statistics reveal patterns, scale and comparison, while interviews, case studies, narratives and experiential data explain how institutional processes operate in practice. Citizens, professionals and affected communities can identify mechanisms and consequences that aggregate data overlook. Qualitative evidence may be harder to standardize, but its contextual depth is essential for interpreting numbers correctly[5].

Triangulation should be designed rather than merely encouraged[6]. A decision record should identify where statistical patterns, professional judgement, and lived experience converge; where they conflict; and what further inquiry or precaution follows from the disagreement. Qualitative evidence should not be treated as decorative testimony, but neither should an individual account be generalized without examining its representativeness and causal relevance.

Measurement systems themselves must remain correctable. Indicators should be reviewed periodically for their relevance, validity, distributional effects and unintended incentives[7]. Changes in society or knowledge may require modifying or abandoning established measures. The assumptions, categories and normative choices underlying metrics should be explicitly examined with input from researchers, professionals and affected groups. This creates “meta-correctability”: not only policy outcomes but also the criteria used to judge them remain open to revision.

Every consequential indicator set should therefore have an identified owner, a stated purpose, a theory linking the measure to the underlying phenomenon, a record of excluded dimensions, a disaggregation plan, and a scheduled review date. Review should examine validity, gaming, distributional effects, data quality, administrative burden, and whether the indicator still informs a decision that an institution is competent to make.

Transparency is central to this process[8]. Institutions should disclose the sources of their data, methodologies, uncertainty margins and interpretive limitations[9]. Numbers should not be presented as complete or objective representations when they depend on contestable choices. Transparency enables researchers, citizens and oversight bodies to detect errors and biases, although information must be communicated accessibly rather than released in overwhelming or unusable quantities. For high-impact decisions, disclosure should distinguish sampling uncertainty, measurement error, missing data, model assumptions, and value judgements. Institutions should present ranges or scenarios where point estimates would create false precision and should explain which conclusions remain robust under plausible alternative assumptions.

Knowledge production should also include experiential and local perspectives[10]. People directly affected by policy understand practical barriers and unintended effects that may be absent from official data. Local knowledge can reveal cultural, geographical and institutional conditions overlooked by uniform national measures. Incorporating these perspectives should provide genuine influence rather than symbolic participation. Conflicting interpretations should be openly examined through methodologically careful and plural evaluation.

Some relevant phenomena cannot be measured adequately, but they must still be governed. Expert panels, deliberative processes, qualitative monitoring and structured professional judgement can supplement incomplete indicators. Early-warning systems can track weak signals concerning social cohesion, institutional trust or epistemic instability before they become visible in conventional statistics. Independent journalism, science and civil society can identify blind spots in official evaluation systems and should therefore receive a formal role in scrutiny.

Artificial intelligence may help classify text, detect anomalies, and compare large mixed datasets, but it also introduces errors of provenance, representation, opacity, and automation bias. Any consequential use should preserve the underlying evidence, document model and human choices, test performance across relevant groups, permit independent replication where lawful, and identify the official responsible for interpretation. AI may extend institutional perception; it cannot determine which outcomes are just or which uncertainties are acceptable[11].

Several permanent tensions remain. Governance needs simplification, but simplification loses nuance. Data can support objectivity, yet their production and interpretation always involve assumptions. Policymakers require direction and predictability, while social reality remains uncertain and dynamic. Rapid decisions need concise information, whereas responsible judgement sometimes requires deeper investigation. Technology can expand analytical capacity but cannot provide contextual meaning or settle normative questions.

These tensions require layered and adaptive decision-making. Immediate action may rely on concise indicators while continuing analysis examines deeper causes and overlooked effects. Measurements should provide orientation without pretending to offer complete control. Human judgement, public reasoning and institutional responsibility must remain central.

Measurement is both a condition and a limitation of democratic correctability. Institutions cannot correct problems they cannot perceive, but a narrow focus on quantification can make essential realities invisible. Effective governance must therefore move from a culture of measuring in order to control towards a culture of learning in order to correct. Its quality depends not on the apparent precision of its numbers, but on its ability to recognize uncertainty, revise its knowledge systems and take seriously what can only be partially measured.

Would you like to explore how a relational understanding of humanity can transform democracy, law, the economy, and public institutions? Click the link below to read the full book, Becoming Human Together, and discover a new framework for building a more humane, democratic, and correctable society within ecological limits.

https://www.researchgate.net/publication/413458836_Becoming_Human_Together_Rethinking_Democracy_Power_and_Institutions_for_a_Complex_and_Changing_World

---

[1] Michael Power, The Audit Society: Rituals of Verification (Oxford: Oxford University Press, 1997); Wendy Nelson Espeland and Michael Sauder, ‘Rankings and Reactivity,’ American Journal of Sociology 113, no. 1 (2007): 1–40, https://doi.org/10.1086/517897.

[2] Marilyn Waring, If Women Counted (San Francisco: Harper & Row, 1988); International Labor Organization, Care Work and Care Jobs for the Future of Decent Work (Geneva: ILO, 2018), https://www.ilo.org/publications/major-publications/care-work-and-care-jobs-future-decent-work.

[3] Joseph E. Stiglitz, Amartya Sen, and Jean-Paul Fitoussi, Report by the Commission on the Measurement of Economic Performance and Social Progress (2009); OECD, How’s Life? 2024: Well-being and Resilience in Times of Crisis (Paris: OECD Publishing, 2024), https://doi.org/10.1787/90ba854a-en.

[4] Sabina Alkire et al., Multidimensional Poverty Measurement and Analysis (Oxford: Oxford University Press, 2015); OECD, Handbook on Constructing Composite Indicators (Paris: OECD Publishing, 2008), https://doi.org/10.1787/9789264043466-en.

[5]J John W. Creswell and Vicki L. Plano Clark, Designing and Conducting Mixed Methods Research, 3rd ed. (Los Angeles: Sage, 2018); Nancy Cartwright and Jeremy Hardie, Evidence-Based Policy: A Practical Guide to Doing It Better (Oxford: Oxford University Press, 2012).

[6] Norman K. Denzin, The Research Act, 3rd ed. (Englewood Cliffs, NJ: Prentice Hall, 1989); OECD, OECD Guidelines on Measuring Subjective Well-being, 2023 Update (Paris: OECD Publishing, 2023), https://doi.org/10.1787/920eb5f0-en.

[7] United Nations General Assembly, Fundamental Principles of Official Statistics, A/RES/68/261 (2014), https://unstats.un.org/unsd/dnss/gp/fundprinciples.aspx; Sally Engle Merry, The Seductions of Quantification (Chicago: University of Chicago Press, 2016).

[8] Sheila Jasanoff, The Fifth Branch: Science Advisers as Policymakers (Cambridge, MA: Harvard University Press, 1990); Andrea Saltelli et al., ‘Five Ways to Ensure That Models Serve Society,’ Nature 582 (2020): 482–484, https://doi.org/10.1038/d41586-020-01812-9.

[9] David Spiegelhalter, The Art of Statistics (London: Pelican, 2019); National Academies of Sciences, Engineering, and Medicine, Communicating Science Effectively (Washington, DC: National Academies Press, 2017), https://doi.org/10.17226/23674.

[10] Frank Fischer, Citizens, Experts, and the Environment (Durham, NC: Duke University Press, 2000); Sheila Jasanoff, ed., States of Knowledge (London: Routledge, 2004).

[11] National Institute of Standards and Technology, Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1 (2023), https://doi.org/10.6028/NIST.AI.100-1; European Union, Regulation (EU) 2024/1689 (Artificial Intelligence Act). Date-stamp any account of applicable duties.