Democracy & Governance Futures

— where self-government is heading

This page is about measurement, not partisan politics — who is right or wrong about any party or leader is never the subject. It is about how researchers measure democracy worldwide, and what real experiments in citizen participation have actually delivered.

What’s actually happening

Three respected organizations that measure democracy worldwide currently tell noticeably different stories, and the gap between them is itself worth understanding. The V-Dem Institute’s Democracy Report 2026 (published March 2026) states that “democracy is back to 1978 levels for the average global citizen” — a population-weighted measure showing roughly a 30% decline from the index’s 2003 peak. On V-Dem’s count, 44 countries were “autocratizing” in 2025 against only 18 “democratizing.” Freedom House’s Freedom in the World 2026 report found global freedom declining for a 20th consecutive year in 2025 (54 countries deteriorated, 35 improved) — but also found a genuinely hopeful figure inside the same report: “more than 85 percent of the countries rated Free as of 2005 remained Free 20 years later” — Freedom House’s own wording, meaning established democracies have proven more durable than the headline decline suggests. The same report also records, under V-Dem’s own published methodology, that the United States lost its liberal-democracy classification for the first time in over 50 years, moving to the electoral-democracy category — V-Dem’s own press release calls the U.S. decline “unprecedented.” That is V-Dem’s classification, in V-Dem’s words; the measurement caveats described below apply to it in full, and this project takes no position on any country’s partisan politics.

The Economist Intelligence Unit’s Democracy Index 2025, released the same season, tells an importantly different top-line story: after eight straight years of global decline, its global average score edged up slightly, and nearly three-quarters of the 167 countries it covers were either improving or holding steady. None of these three organizations is wrong — they build their measurements differently. A 2019 peer-reviewed comparison found real weaknesses in how competing democracy indices combine their underlying data, and Canada’s own Library of Parliament noted plainly (September 2024) that in 2023, no country ranked in the top three across all the major indices at once.

Separately, real experiments in giving ordinary citizens a direct role in government decisions have grown substantially. The OECD’s own tracking database recorded 733 “deliberative” processes worldwide from 1979 through 2023, involving at least 80,622 randomly selected citizens as of its September 2023 count, across 34 countries — with a February 2026 update adding roughly 170 more processes. The clearest trend inside that data: standing, permanent citizen bodies (not one-off events) grew from 22 in 2020 to 41 by 2023.

Where credible people think it’s heading

A small number of cities and regions now run permanent citizens’ assemblies as ongoing institutions rather than one-time experiments. Belgium’s German-speaking Community (Ostbelgien) built the world’s first genuinely permanent regional assembly in 2019 and has run roughly one topic-specific assembly a year since. Paris approved a permanent citizens’ assembly in October 2021; its third annual cohort ran from September 2024 to July 2025. Brussels runs two separate permanent bodies under different branches of government — one tied to its parliament, one to its regional government — a distinction often blurred in secondary reporting.

But whether these bodies actually change policy is a genuine, unresolved argument among the researchers who study them. A 2023 piece from the European Democracy Hub states plainly that assemblies’ “impact on policy and practical change is patchy at best,” even with a formal political mandate: France’s Convention Citoyenne pour le Climat received a public promise from President Macron not to “filter” its recommendations, yet by the piece’s own account only “one in ten” were fully adopted as proposed.

On AI and elections specifically, the fear that dominated coverage of 2024’s “year of elections” (more than 60 countries held national votes) did not match what researchers found afterward. Princeton researchers analyzed all 78 documented cases of AI use in elections worldwide tracked through 2024 and found half were non-deceptive; among deceptive cases, low-tech “cheap fakes” (not AI-generated) were used roughly seven times more often than AI content in the U.S. election specifically, and more than 20 times more often in Bangladesh’s. Their conclusion: the constraint on political disinformation was never really the cost of producing it.

What’s honestly uncertain

Young people’s own attitudes toward democracy show a real generational gap that two different surveys measure two different ways. The Open Society Barometer (fielded May–July 2023 across 30 countries) found 86% of people globally prefer living in a democracy — but among 18–35-year-olds specifically, 35% said they’d favour “a strong leader who does away with legislatures and elections,” against only 26% among people 56 and older. A narrower, more recent U.S.-specific study (Tufts University’s CIRCLE, fielded November 2024, published April 2025) sorted 18–29-year-old Americans into three groups instead of one number: 62% showed “passive appreciation” of democracy, 31% showed “dismissive detachment,” and 7% showed “hostile dissatisfaction” — and these groupings cut across party lines within the same generation, meaning this is a disposition split, not simply a partisan one.

Whether AI tools can responsibly help run large-scale citizen deliberation — synthesizing thousands of submissions into something a government can actually use — is an active, current question with real pilots underway (including a 2025 wildfire-recovery process in California and a 2023 UK government consultation that processed more than 100,000 responses). The honest risk, named directly by researchers studying these tools: they “tend to sand down disagreements” and can lose minority viewpoints in the process of summarizing.

Why it matters for Toronto

Toronto has real, if currently idle, capacity here. MASS LBP, founded in Toronto in 2007, pioneered the “Civic Lottery” model used across Canada and has completed more than 200 policy projects and 50 citizen panels over 19 years. Its Canadian Youth Climate Assembly produced a final report with 23 recommendations in November 2025. Toronto’s own Planning Review Panel ran two full sortition-based cohorts (2015–2017 and 2017–2019) before being put on hold in 2019; as of a 2026-07-23 check, the City’s current public-engagement portal listed no active deliberative or sortition-based process. Canada also carries two honest cautionary precedents: British Columbia’s 2004 Citizens’ Assembly recommended a new voting system that won 57.7% support in a 2005 referendum but fell short of the required 60% threshold; Ontario’s own 2006–2007 assembly lost its referendum decisively, 63% to 37%. A well-run citizens’ assembly recommending change does not guarantee the public will vote to adopt it.