A 2028 election simulation game organised by a Washington-based political strategy group is challenging participants to confront a question most would rather avoid: how, precisely, does a democratic transition of power fail?
The exercise is the creation of Dmitri Mehlhorn, a Democratic political strategist and co-founder of the Atoll Society. The group has now run versions of the game 30 times across major cities in the United States and Europe, drawing participants from academia, politics, technology, finance, and media.
Mehlhorn’s stated goal is to move discussion ‘toward more longer-term scenario planning so that the rule of law can be more resilient in the face of possible future threats.’ The Atoll Society’s website puts the stakes more baldly, declaring: ‘We don’t know whether democratic institutions will survive the next decade.’ A January 2026 profile in The Atlantic reported that Mehlhorn has gone further in his personal preparations, securing dual citizenship for his family on the Caribbean island of St. Kitts and using artificial intelligence to create a set of fictional Substack authors who write about aspects of his political vision.
Inside the 2028 Election Simulation Game
The scenario opens in the run-up to the 2026 midterms in a lightly fictionalised United States, where a nationalist president is struggling through a second term amid an unpopular war, high inflation, and an immigration crackdown. Participants are divided into three factions: Institutionalists, Nationalists, and Capitalists. Each round, teams decide their moves, which are then fed into an AI engine that determines outcomes and resets the board.
In the session attended by a journalist from Reason, the Nationalists showed considerable ingenuity. They conditioned 401(k) eligibility on political contributions to allied causes and effectively revoked tax-exempt retirement accounts from those who backed the Institutionalists. They also built a federal portal for preemptive presidential pardons. Mehlhorn called the retirement-account manoeuvre one of the most original moves he had seen. ‘No money spent, pure administrative law, and it worked,’ he wrote in a follow-up email. His broader observation was bleak: ‘The most effective moves looked like routine paperwork.’
The Capitalist faction, managing a notional $430 billion with a target of a 9 percent annual return, spent much of the game hedging. Funding centrist candidates and public campaigns about the value of free elections was considered safer than overt opposition. One participant proposed buying a private army; the idea was shelved. In a separate game, Mehlhorn noted, Capitalists did build a private army, promised it to the Nationalists, then flipped it at the last minute to bring down the regime.
The game resolved around a fictionalised January 6, with the Capitalists funding bonuses for National Guard troops who defected. A +4 bonus for the Institutionalists, accrued through earlier moves, proved decisive. But Mehlhorn’s assessment was unsettling: ‘Run it again with all the same moves and the Nationalists hold power roughly one time in six.’
Mehlhorn’s Broader Political Activity
Mehlhorn is not a disinterested observer. According to an SEC proxy filing, he co-founded an organisation called Investing in US alongside LinkedIn co-founder Reid Hoffman and serves as Hoffman’s chief political adviser. He told CNBC: ‘Our political philanthropy is focused on weakening the political power of the anti-American Trump-MAGA movement.’
Hoffman and Mehlhorn also backed the legal effort against Donald Trump, funding the nonprofit American Future Republic, which covered costs in writer E. Jean Carroll’s sexual abuse and defamation case against Trump. As Forbes reported from court disclosures, Mehlhorn stated that the grant was made before Carroll filed suit and that he had ‘no prior knowledge that our funding would go to support her in particular.’ A jury later awarded Carroll more than $5 million. The pair also funded an effort to remove Trump from the 2024 ballot under the 14th Amendment’s insurrection clause, which succeeded at the Colorado Supreme Court but ultimately failed.
The Atoll Society describes its technical mission as developing ‘verification systems and governance protocols for a world where shared truth can no longer be assumed,’ with members including technologists, legal scholars, and investors who write under both their own names and constructed fictional pseudonyms.
How Seriously Should the Simulation Be Taken?
Mehlhorn estimates a roughly 50/50 chance that Trump will seek the Republican nomination in 2028, despite the 22nd Amendment’s bar on a third term. Polymarket‘s 2028 presidential market currently places JD Vance as the leading candidate at approximately 23 percent, with the market having logged around $685 million in volume and $64 million in liquidity. Trump’s own odds stand at roughly 2 percent on the same market. Mehlhorn puts the real probability at least ten times higher.
Role-playing games are not reliable predictors of history. A similar exercise conducted in June 2020, involving more than 100 former officials and campaign staffers, warned of chaos on the day Congress certified the Electoral College vote if the result was close and contested. That warning proved more prescient than most participants expected.
The simulation’s most durable lesson may be structural rather than predictive. Each Nationalist power grab in the game made the next one easier, because the Overton window had already shifted. If the Nationalists had previously tried seizing assets, Mehlhorn explained, the final authoritarian move would have carried a better dice roll. Whether or not the game captures the future, that dynamic has clear historical precedent.
