Reading the Receipts: How Public Data Exposes the Flattery of Global Rankings

In the first essay, we asked who builds the index. In the second, we built one and watched the numbers embarrass the people who usually do the grading.

This essay does something simpler. It demonstrates that the entire industry of survey-based global rankings, the ones that produce the same hierarchy every year, Nordic countries on top, Global South at the bottom, designers scoring well on their own instruments, can be replaced in an afternoon by anyone with a laptop and access to public databases.

No surveys. No self-reported attitudes. No proxies that flatten complex social realities into a binary checkbox. Just raw administrative data: court registries, government transparency logs, emissions ledgers, parliamentary records. Data that institutions produce as a byproduct of operating. Data that leaves fingerprints.

Six indices. Each one uses only public, auditable sources. Each one is harder to game than anything the World Bank or Transparency International publishes, because the underlying records are administrative byproducts, not answers to questionnaires. These indices measure what institutions do, not what they say. You can coach a survey response. You cannot quietly erase a court registry.

Each fingerprint index was set against the survey-based ranking that claims to measure the same thing. The gaps are not marginal. France ranks “satisfactory” on press freedom while leading the sample on censorship enforcement. Germany is a “full democracy” whose parliament does not resemble its population. The Netherlands is a top climate performer that offshores half its emissions. In every case, the survey flatters and the fingerprint exposes.

The formulas, data sources, and full methodology for each index are in the exhibit at the end.

1. Censorship Enforcement Index (CEI)

What it measures: How much coercive pressure a state actually applies to speech. Not what its constitution promises, but what its institutions do. Removal requests filed by governments to platforms. Network interference events. Criminal convictions for speech-related offenses.

What it reveals: The distance between a country’s self-image and its enforcement reality. A state that files thousands of removal requests and prosecutes incitement broadly is not a free-speech society with exceptions. It is a speech-regulation society with a good reputation.

The survey says: The RSF World Press Freedom Index (2025) ranks France 25th out of 180, score 76.62/100, classified as “satisfactory.” Germany ranks 11th at 83.85. The UK sits at 20th. All three are considered countries where the press operates freely.

The fingerprint says: France, 0.71, the highest censorship enforcement in the sample. Driven by aggressive use of the EU’s Digital Services Act notification system and a broad incitement-law regime that produces more speech convictions per capita than most countries Europeans would consider authoritarian. Germany is close behind.

The gap: RSF asks experts whether they perceive the press as free. The CEI counts how many times the state acted to restrict speech. France scores 25th on perception and first on enforcement. One measures reputation. The other measures behaviour.

2. Demographic-Political Divergence Index (DPDI)

What it measures: Whether a country’s elected legislature reflects who actually lives there. Census figures on population composition cross-referenced with parliamentary biographical data.

What it reveals: Elite closure. A society can have diversity in its streets, its schools, its labor force, and none of it in its parliament. The index does not ask whether people feel represented. It counts seats.

The survey says: The Economist Intelligence Unit’s Democracy Index (2024) gives Germany 8.80/10, a “full democracy,” ranked 12th in the world. The methodology includes 60 indicators across five categories, one of which is called “political participation.”

The fingerprint says: Germany, 0.71, the highest divergence in the sample. 29.7% of the population has a migration background. 11.6% of Bundestag members do. Europe’s starkest representation gap.

The gap: The EIU asks whether elections are free and whether citizens participate. The DPDI asks who ends up in the seats. Germany’s elections are free. Whether that freedom produces a parliament resembling the population is a different question. The numbers say no. France does not collect ethnic statistics, which means it cannot be scored. That absence is itself a datum.

3. Elite Kinship Network Density (EKND)

What it measures: Dynastic persistence in political power. The degree to which elected office circulates within families rather than across the population. Parliamentary biographical records analyzed as a network graph.

What it reveals: Whether meritocracy is a description or a brand. Survey-based governance indices reward countries that believe in meritocracy. This index checks whether they practice it.

The survey says: The World Bank’s Worldwide Governance Indicators place the United States in the 88th percentile for Government Effectiveness. The EIU Democracy Index scores it 7.85/10. Neither index has a category for dynastic power.

The fingerprint says: The United States, 0.64. Bush, Clinton, Kennedy, Cheney, Pelosi, Cuomo, Paul, Murkowski, Dingell. Dynastic clusters that would be called oligarchic if they appeared in a country the State Department monitored. The networks are not hidden. They are celebrated. Legacy is rebranded as “public service families.”

The gap: A country can score in the 88th percentile on effectiveness while running a political system where last names predict careers. The survey never notices because it was not designed to look.

4. Diversity Extraction Ratio (DER)

What it measures: Whether international diversity is distributed across all levels of power, or concentrated on the revenue side and absent from decision-making. Student enrollment and workforce figures on one side, faculty and board composition on the other.

What it reveals: The difference between diversity as a value and diversity as a revenue stream. A university that enrolls 30% international students, charges them three times the domestic fee, and employs 5% international faculty is not diverse. It is running an export business with a campus attached.

The survey says: The Multiculturalism Policy Index (Queen’s University, 2020) ranks Australia joint first with Canada, the highest score in the world for inclusive immigration policy. Eight indicators measuring policy intent.

The fingerprint says: Australia, 1.00, the maximum extraction ratio in the sample. 28.5% of university students are international, paying fees that subsidize the entire tertiary system. 18.2% of academic staff are international, dropping sharply at the professorial level. Fees flow up. Power does not flow down.

The gap: No major index measures whether diversity generates shared power or merely shared revenue. Australia’s multiculturalism scores well because the question is whether diverse populations coexist. The DER asks whether they coexist as equals. The answer, by Australia’s own institutional data, is no.

5. Carbon Loophole Index (CLI)

What it measures: The gap between a country’s territorial emissions (what it produces) and its consumption emissions (what it consumes, including imports). The difference is offshored pollution: emissions that serve one economy but appear in another’s inventory.

What it reveals: Whether green credentials are real or accounting. A nation that closes its factories, imports the same goods from countries with dirtier grids, and celebrates its falling emissions has not decarbonized. It has relocated the chimney.

The survey says: The Climate Change Performance Index (CCPI 2025) ranks the Netherlands 5th in the world, a “high performer.” Denmark is 4th. The UK is 6th.

The fingerprint says: The Netherlands, 1.00, the extreme case. Consumption emissions exceed territorial emissions by roughly 50%. Goods flow through Rotterdam, emissions are booked at the country of origin, and the Netherlands claims one of Europe’s more favorable carbon profiles. Switzerland, Sweden, and the UK all score high.

The gap: The widest in the set. The Netherlands ranks 5th in the world on climate performance and worst in this sample on carbon offshoring. Both numbers are correct. One counts the chimney. The other counts the smoke.

6. Urban Hostility Index (UHI)

What it measures: How aggressively a city engineers its public spaces to repel its poorest residents. Hostile architecture installations, anti-loitering ordinances, vagrancy enforcement, public amenity availability.

What it reveals: The distance between stated compassion and built environment. A city that bolts armrests onto every bench, removes seating from transit stations, and arrests people for the crime of sitting down has not solved poverty. It has made poverty invisible. The spikes are the budget line.

This index also asks a question that rarely surfaces in development discourse: how can a government claim goodwill abroad while treating its own poor with measurable hostility? The fingerprints are in the concrete.

The survey says: The EIU’s Global Liveability Index (2025) ranks Copenhagen 1st, with Western cities dominating the top tier. The Oxford Economics Global Cities Index places New York 1st and London 2nd. These rankings assess stability, healthcare, culture, and infrastructure. None has a category for how a city treats people who cannot afford to be there.

The fingerprint says: London and Paris compete for the top. London’s Camden bench, engineered so that sleeping, skating, and sitting too long are physically impossible, is an icon of hostile design. Paris removed benches from metro stations entirely. New York conducts more encampment sweeps per year than most European countries combined.

The gap: The world’s highest-ranked cities on liveability are among the most aggressive in engineering public space against their poorest residents. A liveability index that does not account for this is not measuring liveability. It is measuring liveability for those who can pay.

What the fingerprints show

Six indices. Public data. No surveys. In every case, the survey and the fingerprint claim to measure approximately the same thing. In every case, they disagree.

The pattern does not require commentary. The countries that score highest on survey-based indices score worst on the fingerprint indices that measure the same claimed domain, using their own administrative data.

The data sits in transparency reports, census bureaus, emissions inventories, parliamentary records, and procurement budgets that the countries themselves publish. The indices did not need a survey. They read the receipts.

Beyond six

There are more than six. We computed others, on knowledge extraction, debt conditionality, arms-to-aid incoherence, productive sovereignty, linguistic sovereignty, and could compute dozens more. The data is there. It has always been there. That is the point.

The point is not the indices. The point is that they are easy. Any country with a statistics office, a university, and an internet connection can build instruments like these in weeks. The question has never been technical capacity. It has been permission: the assumption that measurement is something done to certain countries by others, and that the results should be accepted as the starting point of every conversation about development or reform.

A critique without numbers is dismissed as emotional, envious, or political. A critique with numbers becomes an analysis. That asymmetry has kept a particular hierarchy comfortable for decades. The numbers in this essay are not perfect. Some are preliminary. All are improvable. But they exist, they are replicable, and they point in a direction no survey-based index has been willing to face.

The invitation is not to accept these scores as final. It is to stop accepting anyone else’s scores as final either. Count what matters to you. Measure what your institutions actually do. Publish the methodology. Let the numbers sit next to the rankings that claim to tell the same story, and see which one survives the comparison.

The tools are not the bottleneck. They never were.

Exhibit: Methodology

The indices below are our construction. The formulas, variable selection, and weighting are open for scrutiny, replication, and use by anyone.

Censorship Enforcement Index (CEI)

Data sources: Google Transparency Report (government removal requests), Meta Transparency Center, OONI (Open Observatory of Network Interference) probe data, Eurostat and national justice ministry statistics (speech-related convictions: incitement, hate speech, blasphemy, defamation where criminalized).

Formula:

CEI = 0.4 × R + 0.3 × O + 0.3 × C

Where:

• R = government content-removal requests per million internet users
• O = OONI network anomaly rate (proportion of tested endpoints showing interference)
• C = criminal convictions for speech-related offenses per 100,000 population

All three components min-max normalized across the sample before weighting.

Weighting rationale: Removal requests receive the highest weight because they represent direct, documented state action against specific content. OONI anomalies and convictions are weighted equally as complementary measures: one captures infrastructure-level interference, the other judicial enforcement.

Limitations: Countries that repress speech through informal pressure, self-censorship incentives, or media ownership concentration will score lower than their actual coercion level. The index captures recorded enforcement, not total repression.

Demographic-Political Divergence Index (DPDI)

Data sources: National census or statistical office (population by migration background, ethnic or linguistic group), parliamentary biographical databases (elected representatives’ backgrounds).

Formula:

DPDI = Σ (w_g × |ln(P_g / S_g)|)

Where:

• P_g = population share of group g
• S_g = seat share of group g in parliament
• w_g = population weight of group g (so larger groups contribute more to the index)

The natural log ratio means a group with 30% of the population and 10% of seats contributes the same magnitude as a group with 10% and 30%. Divergence in either direction is captured.

Limitations: Depends on how countries define and count groups. France’s refusal to collect ethnic statistics makes it unscorable, which is itself informative.

Elite Kinship Network Density (EKND)

Data sources: Parliamentary biographical records, publicly available genealogical databases, government disclosure records (family relations of officeholders).

Formula:

EKND = 0.5 × G + 0.5 × D

Where:

• G = mean betweenness centrality of dynastic clusters in the officeholder network graph (nodes = officeholders, edges = first- or second-degree kinship ties)
• D = percentage of current legislators with a first- or second-degree relative who previously held elected office

Both components normalized to 0-1 across the sample.

Weighting rationale: Graph centrality captures the structural density of family networks. Dynastic percentage captures prevalence. Equal weighting reflects that both dimensions matter.

Limitations: Requires comprehensive biographical data. Spousal and in-law connections may be underreported. The index measures the pattern, not the cause.

Diversity Extraction Ratio (DER)

Data sources: National higher education statistics (international student enrollment), institutional annual reports (international academic staff by rank), labor force surveys (international workforce participation), corporate governance disclosures (board composition).

Formula:

DER = (international share, revenue side) / (international share, power side)

Where:

• Revenue side = mean of (international student share, international workforce share in sectors that charge premium rates)
• Power side = mean of (international share of senior faculty, international share of institutional boards and executive leadership)

A DER of 1.0 indicates parity. Above 1.0 indicates extraction: economic value is drawn from international presence without proportional access to governance.

Limitations: “International” is defined by nationality or country of birth, which may not capture second-generation immigrants. Power-side data is less standardized than revenue-side data across countries.

Carbon Loophole Index (CLI)

Data sources: Global Carbon Project (territorial vs. consumption-based CO₂ accounts), OECD Inter-Country Input-Output tables (trade-embodied emissions).

Formula:

CLI = (E_consumption − E_territorial) / E_territorial

Where:

• E_consumption = CO₂ attributed to final consumption within the country (including imports, excluding exports)
• E_territorial = CO₂ emitted within the country’s borders

A CLI of 0 means the country consumes what it emits. A positive score means it imports more emissions than it produces.

Limitations: Consumption-based accounting relies on trade models and input-output tables that carry uncertainty, particularly for complex supply chains. The index captures direction and approximate magnitude, not precise tonnage.

Urban Hostility Index (UHI)

Data sources: Municipal procurement and infrastructure records (hostile architecture installations), municipal code databases (anti-loitering and anti-camping ordinances), police and court records (vagrancy citations, encampment sweeps), public amenity audits (restrooms and drinking fountains per capita). Sourced through FOI requests and published city budgets.

Formula:

UHI = 0.25 × H + 0.25 × L + 0.25 × E + 0.25 × (1 − A)

Where:

• H = hostile-design installations per km² of public space
• L = anti-loitering and anti-camping ordinances per municipality
• E = vagrancy citations + documented sweep operations per 100,000 population
• A = public restrooms and drinking fountains per 10,000 residents (inverted: fewer amenities = higher score)

All components min-max normalized across the sample before weighting.

Weighting rationale: Equal weighting across four dimensions reflects that hostility operates through multiple channels simultaneously: physical design, legal prohibition, enforcement, and amenity withdrawal.

Limitations: Hostile architecture is not always catalogued in procurement records. Ordinance counts do not capture enforcement intensity. The index measures formal, recorded hostility.

All data sources are publicly accessible. All formulas are replicable. The indices are offered for use, critique, and improvement by anyone.