On October 7, 2020, after a five-year trial, a Greek court ruled that Golden Dawn — a neo-Nazi movement that had won seats in parliament and, at its peak, pulled 7 to 8 percent of the national vote — was not really a political party at all. It was, the court found, a criminal organization “operating under the cover of a political party,” complete with hierarchical command structures and, in the prosecutor’s words, “battalion troops and hit squads”. Prosecutor Thanassis Kampagiannis called it “the biggest trial of fascist criminality since Nuremberg”. What made the case remarkable wasn’t just that a sitting parliamentary party got dismantled — it’s that Greece has no constitutional clause allowing parties to be banned for their ideology. Judges had to reach for the ordinary penal code instead, because, as one account of the case put it, Greek democracy “prosecutes deeds, not beliefs.”

Compare that with Turkey, where courts operate under an explicit constitutional license to dissolve any party that threatens “secularism” or “territorial integrity,” and have used it to shut down more than 27 parties since the 1960s — mostly Islamist and Kurdish opposition movements, rarely anything resembling Golden Dawn. Or with Germany, where the constitution contains one of the most powerful party-ban clauses in the democratic world, and the Federal Constitutional Court has used it exactly twice in the past two decades to actually ban anything — rejecting attempts to outlaw the far-right NPD in 2003 and again in 2017, even while a resurgent AfD now pulls over a fifth of the national vote.

Same legal tool, three wildly different outcomes. Political scientist Christina Armoni set out to explain why, building a new cross-national dataset of party bans — merging existing academic data with Wikipedia’s global list of banned parties and cross-referencing both against the Varieties of Democracy project’s judicial indicators, for 202 countries between 1945 and 2023 — and pairing it with deep case studies of Greece, Germany, and Turkey. Her answer, in short: having the legal power to ban a party barely matters. What matters is whether the court wielding that power is afraid of being wrong.

Formally banning a party turns out to be a genuinely rare event in world politics — showing up in fewer than half of one percent of all country-years in the dataset. But rarity cuts differently depending on regime type. Democracies, despite (or because of) their courts’ caution, actually rack up a higher rate of documented bans than autocracies: about 1.4 percent of democratic country-years involve a ban, versus roughly a third of a percent in non-democracies. That’s not because democracies are more repressive — it’s almost the opposite. Autocracies have easier, quieter tools available: gerrymandered electoral thresholds, harassment of opposition leaders, selective prosecution of individual politicians, media blackouts. None of that shows up in a dataset of formal “bans,” even though it accomplishes the same goal. Democracies, lacking those informal levers — or at least lacking the ability to use them without consequence — are more likely to reach for the blunt, visible, legally documented instrument instead.

The statistical patterns bear this out. Across a series of models isolating within-country change over time, judicial capacity — a composite measure of how independent, competent, and free of corruption a country’s courts are — consistently predicts a higher likelihood of a formal party ban. But that effect is dramatically stronger in democracies than in autocracies: strong courts in democratic systems are far more willing to actually use their power to dissolve a party than equally strong courts operating inside an authoritarian regime, where executives often prefer to handle opposition through other means. Judicial accountability tells a similar story — courts that are more transparent, reviewable, and answerable to the public are more likely to ban parties in democracies, but that same accountability has little effect, or even a mildly negative one, in non-democracies, where “accountability” mechanisms are frequently just another tool of political control rather than a genuine constraint.

What ties all of this together, in Armoni’s framework, is backlash — the very real risk that banning a party will blow up in a court’s face. Ban a popular movement without an airtight case, and you risk turning its members into martyrs, discrediting the judiciary, and handing the extremists exactly the persecution narrative they wanted. That risk is enormous in a democracy with a free press, competitive elections, and an opposition that can hold a court accountable at the ballot box or in public opinion. It is close to nonexistent in a state where the media is controlled, the opposition is weak, and elections aren’t really contests. So democratic courts — even ones equipped with sweeping legal powers — tend to wait, gather overwhelming evidence, and ground their decisions in the narrowest possible legal reasoning. Authoritarian courts, freed from that fear, can move fast and loose.

Greece is the clearest illustration of caution winning out. Lacking any constitutional ban clause, prosecutors spent years — while Golden Dawn’s members committed hundreds of assaults, culminating in the 2013 murder of the rapper Pavlos Fyssas — assembling more than 400 documented incidents before finally convicting the party’s leadership under ordinary criminal statutes for organized crime, grounded in a legal standard that Greece’s Supreme Court described as requiring “specific and reasoned justification”. The trial survived multiple changes in government, including a left-populist coalition, without political interference — evidence, in Armoni’s reading, of exactly the kind of high-capacity, high-accountability judiciary that can absorb the political risk of dismantling a sitting parliamentary party. The court later used the same precedent to bar Golden Dawn’s rebranded successor parties from subsequent elections — again, without ever invoking ideology as the legal basis.

Germany shows the same caution operating inside a system explicitly built for militancy. Article 21 of the Basic Law lets the Federal Constitutional Court dissolve any party that actively seeks to “impair or abolish the free democratic basic order” — a far more direct tool than Greece’s. And yet the court has used that power so sparingly that its two modern rulings on the NPD were both refusals: first in 2003, after judges discovered the case had been built partly on infiltrated state informants, and again in 2017, when the court concluded the party was simply too marginal to pose a real threat. That same reluctance now shadows the live debate over banning the AfD, a party the domestic intelligence service has formally classified as right-wing extremist even as it becomes more electorally dominant, not less. German courts, in other words, treat their own militant-democracy powers less as a tool to reach for than as a last resort to justify almost never using.

Turkey inverts the pattern entirely. Its constitution grants the Constitutional Court explicit authority to dissolve parties that threaten secularism or territorial unity, and the court has used it aggressively and asymmetrically — closing more than two dozen parties, disproportionately Islamist and Kurdish opposition movements, while a party with its own Islamist roots that happened to hold executive power survived a 2008 closure attempt. In 2009 the court unanimously dissolved the pro-Kurdish Democratic Society Party and banned 37 of its politicians; the same playbook resurfaced against its successor, the HDP, which the government has accused of being a front for a banned militant group. Following a 2016 coup attempt, more than 2,000 judges were purged and the body overseeing judicial appointments was placed under executive control — the kind of move that, per one recent analysis, has turned Turkish courts into engines that “retain formal independence but substantively operate in the service of executive consolidation.” Because Turkish elections aren’t fully competitive and the media isn’t fully free, there’s simply no backlash constraint left to restrain the court — party closures barely register outside the communities directly affected. And what doesn’t happen through formal bans happens anyway: Turkey has maintained one of the highest parliamentary entry thresholds in the democratic world, and opposition leaders face arrest and detention through channels that never touch Article 69 at all.

The uncomfortable implication of all this is that democracies don’t get to simply avoid the question of banning parties by staying pure. Cordon sanitaire — the informal refusal by mainstream parties to cooperate with extremists — is the preferred first line of defense in places like Germany and the Nordic countries, but the research on it is not encouraging: isolation just as often lets extremist parties cultivate a victimhood narrative and radicalize further. When soft containment fails, democracies are left with the same blunt legal instrument autocracies use to crush opposition — the only real difference being whether the court wielding it has the institutional strength to build an airtight case and the public accountability that forces it to. Judicial capacity gives a court the ability to act. It’s accountability, and the fear of getting it wrong, that decides whether that power defends democracy or just becomes another way to destroy it.

Angelos Tzortzinis/Agence France-Presse — Getty Images