Scored on the CIA's Instability Model, the United States Is in the Highest-Risk Category for Civil War, Coup, or Collapse
Pangram verdict · v3.3
We believe that this text is a mix of AI and human-written content.
AI likelihood · overall
MixedArticle text · 1,551 words · 4 segments analyzed
Photo-illustration by Alex Cochran. Source: Getty.Scored on the CIA’s Instability Model, the United States Is in the Highest-Risk Category for Civil War or CoupNote from the author: This article is long; it needs to be. There are plenty of great sources for quick hot takes, but here at the Existentialist Republic we aim to present factual information, effective solutions, and meaningfully orient the world towards justice. That just doesn’t happen in the length of a tweet.In 1994, at the request of Vice President Al Gore, the Central Intelligence Agency assembled a task force of academics and gave them one job, to work out which governments around the world were headed for civil war or collapse and whether it could be predicted years in advance. Ted Robert Gurr, a political scientist at the University of Maryland, built the first version. Jack Goldstone of George Mason University led the modeling that followed, and in 2010 he and seven colleagues published the result in the American Journal of Political Science.Tested against every country of more than half a million people from 1955 to 2003, the model was right more than eighty percent of the time, two years in advance, about which states would break down and which would stabilize. By law, the CIA is barred from studying the United States itself, so Barbara Walter, a political scientist who served on the task force, scored the country on her own. I did the same, working from her research, the CIA’s methodology, and using data covering the time since Walter’s exploration of the subject.Four things accurately predicted national and sub-governmental breakdown, out of dozens of categories the researchers tested: the type of political regime, the infant mortality rate, whether four or more neighboring countries were at war, and whether the government practiced discrimination against a defined group.Regime type predicted breakdown far more accurately than the other three. Of the regime types the model scores, the one with the highest odds of breakdown, a partial democracy with deep factionalism, had more than thirty times the odds of instability compared to a plain dictatorship.The odds run that way because the regimes at both ends of the scale last and the ones in the middle break.Håvard Hegre and three colleagues at the Peace Research Institute Oslo found, across 152 countries from 1816 to 1992, that “coherent democracies and harshly authoritarian states have few civil wars” and that the regimes in between have the most, even after they have had time to settle from a regime change. A full dictatorship can repress and a full democracy can accommodate, and a regime in the middle does neither well enough to be anything but fragile and tenuous. This can be good or bad news depending on which side of the autocratic takeover you happen to be on, though I take it as a sign for optimism that the GOP regime may be lacking in traits defining a more robust capture of the United States.Regime type in the model comes from the Polity data, which rates every country from positive ten for a full democracy to negative ten for a full dictatorship. The middle range, positive five down to negative five, is anocracy, a mix of democratic and authoritarian rule. Polity’s last reading, in 2020, scored the United States a positive five, below the cutoff for a democracy, after the executive refused congressional oversight and worked to discredit the vote count. Polity stopped scoring that year, and the V-Dem Institute’s 2026 report, which now does the same work, downgraded the United States from a liberal democracy to an electoral one.
Both readings fall short of a full democracy.The major independent measures of democracy all show the United States declining, and one of them shows the fastest decline it has ever recorded. On V-Dem’s index, maintained by the University of Gothenburg, the United States fell from twentieth to fifty-first in the world in a single year, its sharpest one-year fall in a record reaching back more than two centuries. On Freedom House’s hundred-point scale the United States fell from 93 to 81 over the past two decades, the steepest decline of any country it still rates as free. The Economist’s index rates the country a flawed democracy at its lowest score since the measure began in 2006. The Century Foundation’s new democracy meter scored it 57 out of 100 for 2025, down from 79 the year before.Polity had already recoded American political competition from competitive to factional in 2016, the condition the task force found to be the strongest single predictor of national split, collapse, coup, or balkanization. That recoding cut the score from positive ten to positive eight. Polity dated the switch to the presidential election and named the two factions as Trump’s supporters and the opposition. The Center for Systemic Peace, which maintains Polity, rated the United States at high risk of political instability when it recorded the 2020 drop, and it called the executive’s effort to overturn that election an attempted presidential coup. The people responsible for that coup were pardoned and placed back in power, they’ve been systematically removing Democratic safeguards for nearly two years since returning to power.Factionalism in the model is politics split into hostile camps that treat power as winner-take-all instead of a contest between inherently collaborative stakeholders. V-Dem’s 2026 report describes the Republican Party as far-right and anti-pluralist. V-Dem’s party index had scored it that way years earlier, rating parties from zero to one on four behaviors: weak commitment to pluralism, demonizing opponents, contempt for minority rights, and encouraging political violence.By the 2016 election the Republican Party had crossed the 0.43 line for an anti-pluralist party, and in V-Dem’s last coding, before the 2018 election, only 15 percent of governing parties in democracies this millennium scored more illiberal than it did, closer to Turkey’s ruling AKP at 1.0 and Hungary’s ruling Fidesz at 0.88 than to the British Conservatives at 0.35 or the German Christian Democrats at 0.05. The Democratic Party stayed in the mainstream range. That 2018 reading is the last one V-Party took.
The index has scored nothing since: the January 6 attack, the pardons of the people who carried it out, the mass internment camps, the ICE agents screaming “Make America Great Again, bitch” as they beat and abduct Americans, or the Republican midterm convention in Dallas on September 10, 2026, where an attendee shouted “He should be shot!” each time Senator Ted Cruz named a Democratic candidate, and Cruz did not react and later posted the clip of his speech with the shouts audible.
One of the country’s two governing parties behaved like the ruling parties of the world’s elected autocracies in 2018, and the eight years since are the ones no index has scored.Combine the two and the United States falls into the model’s most dangerous category, a partial democracy with factionalism. Polity’s last coding is six years old, and the test of whether it still applies is whether that party holds enough of the state to act on its anti-pluralism, because an anti-pluralist party in opposition is a problem and an anti-pluralist party in power is a regime. A party that controls the White House and the courts and no longer accepts the elections it loses, as it showed after 2020 and on January 6, 2021, when its supporters stormed the Capitol to stop the count, passes that test.It should be noted that the eighty percent is the CIA model’s track record, and its accuracy has fallen in the last decade. Still, what it predicts is instability, meaning civil war and the violent or unconstitutional replacement of a government. That is all I take from the model, a factional partial democracy with high odds of breakdown, and everything after this point is my argument about what that breakdown makes possible, resting on the evidence I give for it rather than on the model itself.The model predicts instability, and the words people reach for are civil war and coup. Secession is the one that many treat as unthinkable or shocking. Support for it exists, and it has flipped by party. About a fifth of Americans told YouGov in 2026 they would back their own state leaving. The national figure is down from a quarter in 2024, but only because Republican support collapsed once the party took power, from 29 percent to 14. Democratic support stayed flat while it climbed in blue states, and Democrats are now likelier than Republicans to want out of the union. The flat Democratic number hides something. When a Democrat thinks about secession, the word that comes to mind first is Confederacy. Many read the word secession through the lens of slavery, hate, oppression, and the Lost Cause, the postwar myth that recast the Confederacy as noble, so the instinct is revulsion before analysis. The Confederacy captured the idea of leaving in 1861, and that association has lasted 165 years since.It is odd that we associate moral good with the binding of a tumultuous union rather than good with good. The founding American document does not. The Declaration of Independence grounds government in the consent of the governed and says that when a government becomes destructive of rights, the people may alter or abolish it. The United States was born from a secession against an abusive authority.