Democracies do not die in a single dramatic event. They are dismantled piece by piece, institution by institution, in a sequence so consistent that it can be mapped, measured, and — crucially — detected before it is too late.
Analysis of 38 backsliding episodes across 91 countries reveals that autocratisation follows an eight-step progression. The ordering is consistent in 84% of observed cases. Each step disables a specific democratic function. Each has measurable leading and lagging indicators. And each has a quantifiable probability of reversal.
The sequence matters because it defines the window for intervention. By Step 3, the judiciary has been captured, and the last institutional veto point is gone. By Step 5, the event horizon has been crossed. After that, recovery from within drops to 3%.
This report describes each step, provides empirical calibration from the Political Topology dataset, and illustrates the pattern with country case studies spanning Hungary, India, El Salvador, Poland, and the United States.
Democratic forbearance and mutual tolerance collapse. Political opponents are treated as enemies rather than rivals. Norms of restraint — not exploiting procedural loopholes, respecting the spirit as well as the letter of the law — erode. The formal institutions are still intact, but the informal guardrails that make them work are crumbling.
Independent media comes under pressure. Ownership is concentrated in regime-friendly hands. Critical journalists face legal harassment, tax investigations, or physical threats. State media expands. Social media platforms are co-opted or manipulated. The goal is not total censorship but creating an information environment where the regime's narrative dominates.
Courts are packed, expanded, or politically captured. The judiciary — the last institutional veto point — ceases to function as an independent check on executive power. This is the most consequential step, because once the courts are neutralized, there is no domestic institution that can enforce constitutional constraints.
Parliament becomes a rubber stamp. The legislature either loses its independent deliberative function or becomes dominated by the ruling party/coalition to the point where opposition is decorative. Legislative oversight of the executive — budgets, appointments, investigations — ceases to be meaningful.
Independent agencies — the central bank, electoral commission, anti-corruption bureau, civil service, statistical agencies — are politicized. The technocratic infrastructure that provides impartial governance is colonized by regime loyalists. Data itself becomes unreliable.
NGOs, trade unions, universities, religious oorganisations, and civic groups are restricted, defunded, or co-opted. The space for oorganised opposition outside the political system shrinks. Foreign-funded oorganisations are particularly targeted (the "foreign agent" playbook pioneered by Russia and now adopted globally).
Elections continue to be held — this is the signature of modern authoritarianism — but they are no longer free or fair. Gerrymandering, voter suppression, opposition candidate disqualification, media blackouts, and outright fraud ensure that incumbents cannot lose. The appearance of democracy is maintained whilst its substance is eliminated.
The regime rewrites the constitutional order to make power transfer impossible. Term limits are abolished. Emergency powers become permanent. The security apparatus is fully loyal to the leader rather than the state. The distinction between party, state, and leader collapses. The transition from democracy to consolidated autocracy is complete.
Viktor Orbán's deconstruction of Hungarian democracy is the canonical case of incremental erosion. After winning a two-thirds parliamentary supermajority in 2010, Orbán moved through the steps in order: captured media (2010-12), packed the Constitutional Court (2012-13), subordinated Parliament through supermajority rules (2013-14), colonized regulatory agencies and universities (2014-17), restricted NGOs via the "Lex CEU" and "Stop Soros" laws (2017-18), and gerrymandered elections (2018-22). Each step was individually defensible as a legitimate exercise of democratic power. Collectively, they built what Orbán himself calls an "illiberal democracy." The National Cooperation System (NER) now controls approximately 90% of Hungarian media.
India's erosion is distinctive because it proceeds under the cover of the world's largest electoral democracy. V-Dem downgraded India to "electoral autocracy" in 2017. The sequence: Hindu-nationalist media ecosystem (Step 2 from 2014), judicial pressure through appointment delays and "favourable" bench assignments (Step 3), parliamentary bulldozing via voice votes and limited debate (Step 4). Steps 5-6 are underway: the Election Commission's independence is questioned, the FCRA (foreign contribution) act is used to shut down NGOs, and academic freedom is restricted. India demonstrates that erosion can be compatible with large voter turnout and apparent electoral competition.
Poland proves that the sequence can be reversed — if intervention comes before the event horizon. After 8 years of PiS-led erosion (judicial capture, media politicisation, electoral manipulation), the 2023 election brought the Tusk-led opposition to power. The recovery is real but incomplete: restoring judicial independence requires navigating constitutional constraints, and President Nawrocki (PiS-aligned) creates institutional friction. Key lesson: reversal required (a) unified opposition, (b) high voter turnout (74.4%), (c) EU external pressure, and (d) the erosion not having reached Stage 5.
Nayib Bukele demonstrates that democratic erosion can be enormously popular. His "war on gangs" achieved 91% approval while suspending due process, imprisoning 83,000+ under a state-of-exception, and abolishing term limits (July 2025). El Salvador skipped steps — moving directly from Step 2 to Step 6 — because the population actively endorsed the trade. This is the hardest case for democratic theory: what happens when the people choose unfreedom?
| Country | Liberty | Stage | Velocity | Reversal Prob. | Key Risk |
|---|---|---|---|---|---|
| Finland | 100 | S1: Stable | 0.0 | n/a | None identified |
| France | 78 | S2: Early Warning | −0.3 | 71% | Far-right nnormalisation |
| Poland | 82 | S2: Recovering | +0.8 | 71% | Presidential veto, institutional friction |
| Israel | 60 | S3–4 | −0.7 | 45% | Judicial overhaul + war cover |
| India | 62 | S4 | −1.5 | 28% | Scale, media capture, Hindu nationalism |
| United States | 48 | S5 | −3.1 | 12% | Regulatory capture, judicial realignment |
| Hungary | 52 | S5–6 | −1.8 | 12% | NER system fully embedded |
| Turkey | 18 | S7 | −2.3 | 4% | Opposition prosecuted, media fully controlled |
| El Salvador | 35 | S6–7 | −4.5 | 4% | Popular mandate for autocracy |
| Russia | 13 | S8 | −0.5 | 2% | Fully consolidated |
The eight-step model's most important implication is about timing. The reversal probabilities decline steeply and non-linearly:
| Step | Reversal Probability |
|---|---|
| 1. Norm Erosion | 82% |
| 2. Information Capture | 71% |
| 3. Judicial Capture | 45% |
| 4. Legislative Subordination | 28% |
| Step | Reversal Probability |
|---|---|
| 5. Regulatory Capture | 12% |
| 6. Civil Society Suppression | 8% |
| 7. Electoral Manipulation | 4% |
| 8. Constitutional Consolidation | 2% |
The eight-step model was assessed as "Partially Valid" by the independent audit. The sequential ordering is consistent in 84% of cases (strong). The stage-specific reversal probabilities are directionally correct but add only +5 percentage points over a simple persistence baseline. The structural break at 2006 (pre-2006 recovery: +38%, post-2006: −23.3%) is a genuine and important finding. The model's primary value is as a diagnostic and early-warning tool rather than a precise predictive instrument.