The Democratic Party is about to make the most predictable mistake in American politics

Democrats will win in 2026 and 2028. The problem is winning in a system with a two-year half-life.

Consider the following scenario.

It’s 2031, and Republicans now hold 29 state government trifectas, up from 23 in 2026. After Democrats won back a governing trifecta on an anti-Trump wave in 2029, they thought the nightmare was over. Then came the predictable backlash against the incumbent party, and the 2030 elections were a wipeout.

Now, with the 2030 census completed, Republicans are redrawing congressional and state legislative maps for the next decade. The maps are more aggressive than anything in the modern era, no longer restrained by Voting Rights Act protections against racial gerrymandering. And still nursing fraud claims of a stolen 2028 election, a new generation of MAGA Republicans is passing new election administration laws, trying to lock in their power for at least a decade by making it even harder for Democratic constituencies to vote and have their votes counted.

Some warned about this back in early 2026. But then Democrats did well in the 2026 midterms by running against an unpopular Trump administration, and again in 2028, for the same reason. Democratic strategists convinced themselves that a vote against Trump was a mandate for Democrats, and that finally, the country would return to normal if only Democrats governed well and moderately.

Once again, they were mistaken. They ignored the most predictable dynamic in democratic politics. The governing party always loses popularity, and public opinion always moves against it.

That’s the most likely version of 2031. I’m writing it now because there’s still time to make it wrong. But making it wrong will require an escape from the magical thinking that some Democratic alchemist will finally discover the perfect policy and messaging formula for defying gravity.

The doom loop doesn’t care about your post-election narrative

In September 2017, I wrote an essay for Vox in which I first described what I called a “doom loop” of partisan escalation. I laid out a scenario in which Donald Trump would refuse to concede the 2020 election and encourage violence. I called it “frighteningly plausible.” Three years later, a shameful mob proved me distressingly prophetic.

The delusion that powers the doom loop, that made something like January 6 predictable, is the belief that one side must win a permanent victory in order to save the country, and if the other side wins, it is permanent devastation. This is the antithesis of democracy, which is a system in which parties can lose elections. Democracy depends on the losing side accepting the winner as legitimate. But each turn of the doom loop makes that harder. And each contested transfer of power raises the stakes of the next one.

Since that essay, I’ve been screaming that there is an exit. That staying in the doom loop is a choice to keep the default settings. The exit remains in plain sight. We can change the electoral rules to allow more parties to compete. The doom loop runs on the binary choice between just two parties. Break the binary choice, and the doom loop unravels. Mistake your turn in power for a chance at permanent rule, and the doom loop darkens.

Let’s review the history since 2017. We know the doom loop has tightened. But the illusion of building a permanent majority maintains its compelling hold. It is that illusion I come here to shatter.

In 2018, Democrats swept the midterms. The lesson Democratic strategists took: the electorate was rejecting Trump. Yes. It was. But the pattern was predictable, and perhaps had less to do with Trump than they assumed. The party in the White House, the party with unified control of government in Washington, lost support in the midterms. That’s the most predictable result in American politics. Every president since FDR who has held unified government has lost seats in the subsequent midterm.

In 2020, Democrats won the presidency. The lesson the party’s leaders took: democracy held, the system self-corrected, the voters chose wisely. What actually happened: backlash against the incumbent party delivered the White House. Voters were unhappy with Trump, yes. But perhaps more simply, a pivotal swath of disaffected voters just wanted something different, which is the same thing they always want. Not a vote for Democrats as much as a vote against Republicans. Time for a change. It’s always time for a change. Democrats passed $1.9 trillion in covid relief, then a bipartisan infrastructure law. Neither registered with voters.

In 2022, Democrats avoided the worst. The Dobbs decision overturning Roe v. Wade mobilized enough voters to hold the Senate and limit House losses to single digits. The lesson party officials took: good candidates and strong issues can beat the pattern. But the real lesson of 2022 was the opposite. It took the most galvanizing judicial decision in a generation to merely reduce the cost of ruling, not eliminate it. Democrats still lost the House.

In 2024, the anti-incumbent effect was still in force. Biden had tried a different approach from Obama: bipartisan infrastructure deals, ambitious climate and industrial policy, strong job numbers. None of it mattered. The cost of ruling collected its debt. Biden’s approval slipped and fell. Yes, it didn’t help that he insisted he was capable of being president until he accepted he wasn’t. Yes, inflation was not great. But the US economy outperformed every G7 country, and Democrats still lost. As I’ve argued previously, in the US, the economy is now delinked from electoral outcomes, largely because of the strength of partisanship as an information processing shortcut.

Governing is the price of winning

The “cost of ruling” is what it sounds like. Governing parties pay an electoral price for being in charge. They lose support for governing. On average, about two to three percentage points of vote share from election to election. This finding has been tested and confirmed across democracies, over decades.

Of course, conditions matter. A good economy is, on balance, better than a bad economy. But the basic pattern is remarkably durable: public opinion moves against the party in power. The one constant in politics is that people always think they want change. They want things to be better. And usually there’s a majority that can be convinced that the reason things aren’t better is because the party in power is doing a bad job.

So why is this so consistent? The deepest explanation is probably our basic negativity bias. We keep a running tally of disappointments, and the tally only grows. Even enthusiastic partisans lose enthusiasm over time.

Then there’s the media. Governing parties get more coverage. Cameras follow problems, and problems follow whoever is in charge. The opposition stays in the background, where scrutiny is lighter. The spotlight wilts even the prettiest flowers, eventually.

A third reason is that political coalitions fracture over time. It’s easy to unify at the start of an administration. But not everybody’s priorities can be first. Compromises leave some feeling compromised. The opposition picks up disappointed dissenters. The coalition shifts. Until the next election, when the electoral map shifts slightly, enthusiastic winners call it a realignment, and the pattern repeats.

And honestly, the pattern is mostly healthy for democracy. Rotation in power is good, so nobody gets too entrenched. Citizens should be critical and demand more. The press should hold power accountable. The danger is when the oscillation just becomes a doom loop, instead of a force for innovation and new leadership.

Political scientists often call this pattern thermostatic politics. Public opinion readjusts against whoever is in power, like a thermostat correcting the temperature. Public opinion on any issue is always a moving target.

When the system works, the thermostat keeps things from overheating, and pushes back against extremes. The problem is that America’s thermostat only has two settings, neither of which is temperate. And one of which is decidedly overheated today.

The table below shows what happens in the first midterm election after a president takes office with unified control. The pattern is almost mechanical: the president’s party loses House seats, often loses Senate seats, and loses state government trifectas while the opposition gains them. The sole exception is 2002, when the aftermath of September 11 temporarily overrode the pattern.

The average first-midterm cost since 1992 (excluding the 2002 anomaly): 42 House seats, 3 Senate seats, and a net swing of about 4 state government trifectas from the president’s party to the opposition. Biden’s 2022 was the mildest penalty in the dataset, and it still cost Democrats the House.

Going into the 2026 midterms, Trump’s unpopularity is real, and it is a genuine asset for Democrats in opposition. It fuels turnout, fundraising, and coalition unity. And “cost of living” and “affordability” are compelling slogans for a party out of power.

But Trump’s unpopularity and “affordability” are both traps. They are siren songs that lead Democrats to believe their electoral victories are a vote of confidence in them instead of a vote of protest against the Republicans. In a two-party system, the two are indistinguishable on Election Day.

A vote against Republicans only works when Republicans are in power. A pivotal swath of disaffected voters once again wants something different, which is the same thing they always want, no matter what happens in between elections.

Yes, Trump is uniquely unpopular. Yes, candidate quality matters at the margins. But Nikki Haley would be underwater at this point too, if she had somehow become president. So would Kamala Harris, had she won. If you can imagine a president or a governing agenda that would be at 55 percent approval in 2026, I’ve got a Herbalife business to sell you.

The negativity engine of modern media, the calcified electorate, the steady hum of brokenness and perpetual desire for something different: these are structural forces that grind down support for governing parties, no matter how they govern.

Policy matters, sure. It’s the point of politics. But transcending the cost of ruling with the right policy is like pushing a car uphill with no engine. You recruit help. Everyone pushes. Eventually the helpers get tired, and the car rolls back to where it started. The problem was never effort. The problem was the road.

The backlash is here. The backlash to the backlash is coming soon.

Democrats are going to win back the House in 2026, possibly even the Senate, and probably win the presidency in 2028. I’d put the probability at about 70 percent. Maybe this is too optimistic, but Trump’s approval rating is falling much faster in his second term than in his first, suggesting an even bigger wipeout ahead. Even without a trifecta, the cost of ruling grinds down the president’s party, and the 2030 midterms deliver the same state-level wipeout, giving Republicans back power. And don’t forget. Population projections for 2030 show representation shifting from blue to red states. Based on current trends, Texas is projected to gain as many as four House seats, Florida three. California could lose as many as four, New York two to three.

Presumably, Democratic strategists read the same polls that I do. Presumably they know that Democratic Party favorability has hit historic lows: as low as 34% in Gallup’s measure, the worst the party has ever recorded, with no recovery in early 2026 tracking. Fewer than three in four Democrats view their own party favorably, down from 87% before the 2024 election. Confidence in Democratic congressional leadership is at 25%, the lowest Gallup has measured since it began asking in 2001. A record 45% of Americans now identify as independents. These are all long-term declines.

But the risk is that Democratic strategists will believe that unpopularity is a policy-delivery problem. Fix costs, fix immigration, show results, and the numbers will follow. Obama believed this. Biden’s whole governing theory was built on it.

No modern American president has figured out how to enact and deliver a policy agenda so overwhelmingly popular that the cost of ruling ceased to function. And while the cost of ruling has been consistent, the small-d democratic costs of Democrats losing have increased. The Republican Party will only grow more radicalized out of power. The doom loop does not self-correct. A thermostat with only two settings doesn’t correct. It oscillates with increasing amplitude.

You can’t win a two-party doom loop. You can only take turns losing.

So here is my pitch to Democrats: The Democratic Party brand is at historic lows. Fewer than three in four Democrats can muster a favorable view of their own party. A record share of voters identify as independent. The coalition that will sweep Democrats back into power in 2028 is not a Democratic coalition. It’s an anti-Republican coalition wearing Democratic clothes. That coalition will not survive the cost of ruling under the current label. Americans don’t just dislike the two parties. They want more of them. The pitch writes itself: vote for us and we’ll give you more choices.

Nearly eighty percent say the country is in a political crisis. Just 4% say the political system is working well. Roughly eight in ten want major changes or complete reform. Clinging to what feels safe is how the system keeps breaking. The party system in America is long past its sell-by date.

I’ve written elsewhere about why the conditions for a major reform era are converging in this moment. The demand is already there. It’s looking for a vehicle. An electoral reform agenda that opens up the party system should be the driving engine of that vehicle.

The Democratic Party could run more effectively as a coalition of distinct parties addressing different parts of the electorate, rather than trying to unify behind a single message under an old brand that voters increasingly don’t trust. None of the internal Democratic tensions have easy resolutions, because they are real tensions. A coalition structure lets those tensions negotiate openly rather than detonate internally.

But this isn’t just a Democratic argument. Center-right Republicans who want to distance themselves from MAGA currently have nowhere to go without leaving the party entirely. Under proportional representation, a governing-oriented conservative party could exist alongside a populist-right party, competing for different voters instead of cannibalizing each other in primaries. The doom loop radicalizes the Republican Party with each cycle, because the primary system rewards whoever channels the most outrage. PR breaks that incentive. Republicans who care about democratic competition should want a system where losing an election doesn’t feel like an existential threat. The doom loop breaks for everyone, or it breaks for no one.

You don’t fix a structural problem with better messaging. You fix a structural problem with a structural solution.

There are two very clear ways forward: fusion voting and proportional representation. Both deserve the full attention of everyone reading this. I’ll take them in turn.

Fusion voting gives the thermostat more settings. Under fusion, multiple parties can nominate the same candidate, and voters choose which ballot line to support. You can vote for the Democratic candidate on a Working Families line, or a new Common Sense line. No wasted votes, no spoiling. The major and minor parties bargain with each other in full view, with the minor party getting more policy concessions as it produces more votes. The coalition shifts its internal balance without collapsing.

Today, fusion voting is legal in only two states, New York and Connecticut. But lawsuits challenging anti-fusion laws are active in Kansas and Wisconsin, and new parties have formed in several states in anticipation. The infrastructure exists. What’s missing is legislation in the states where Democrats already have the power to pass it. A candidate running on two ballot lines gets something no poll can provide: a precise, election-night accounting of which parts of the coalition showed up and why. And voters get a way to express frustration with the major parties without risking spoiling an election.

Proportional representation redesigns the thermostat entirely. Here’s how it works. Instead of one representative per district, several are elected from larger districts. If a party wins 40 percent of the vote, it gets roughly 40 percent of the seats. The winning threshold drops, which means more parties can compete and win. Today, Massachusetts sends nine Democrats to Congress despite a third of its voters voting Republican. Under proportional representation, those voters would have representation too.

Ideally, we’ll end up with both. Proportional representation for legislative bodies. Fusion voting for single-winner offices like President, Governor, US Senate, and Mayor, and for small states where proportional representation is less practical without expanding the size of Congress.

The comparative evidence finds that in multiparty democracies, the cost of ruling is significantly lower than in two-party systems, for two reinforcing reasons. First, the media’s negativity burden falls more heavily on single-party governments; in coalitions, scrutiny is distributed. Second, and more important, voters can readjust the coalition without tossing the whole government. They can shift support from one governing partner to another, punishing the faction they’re frustrated with while keeping a broadly similar coalition in power. The binary choice between “endorse everything” and “burn it down” disappears. Electoral losers in proportional systems maintain greater trust in government, because they’re not permanently locked out. The doom loop depends on a binary. Remove the binary, and the loop breaks.

There is a legislative path to fusion and proportional representation. It runs through 2029 and then it runs into the abyss.

The timeline for this is not a single bill in 2029. It’s a strategy that unfolds in stages, each creating the conditions for the next. The Democratic Party could begin by passing fusion voting wherever it holds state trifectas after the 2026 elections. Fusion is legislatively simple, and organizations supporting multiparty democracy already have infrastructure in several states.

In 2028, proportional representation could reach state ballots through citizen initiatives, and perhaps a few state legislatures that have already seen fusion in action. The proof-of-concept matters: voters who have experienced multiparty ballot lines will be more receptive to expanding the model.

By 2029, a Democratic Congress could pass PR nationally. This almost certainly requires eliminating the filibuster. Democrats will probably have done this already to pass the rest of their agenda. If they haven’t, PR should be reason enough. If Congress acts in 2029, PR goes into effect for the 2032 elections, for all states with at least two districts, the first elections after the 2030 census.

The reform also builds its own defense. Once fusion lines exist and candidates are running on them, new political constituencies form around the new system. Repealing fusion would mean killing ballot lines that real candidates are using, eliminating organizations with staff, donors, and volunteers, telling elected officials their path to office no longer exists. That’s a much harder vote than undoing a law nobody has used yet. Federal legislation is stronger still. Overriding a presidential veto takes two-thirds of Congress.

We certainly need a political realignment. But realignment works better when there is space for new political coalitions to build. The reform creates that space. The space protects the reform.

And the moment for reform is now.

The doom loop doesn’t just repeat. It spirals. Each cycle, the maps get more aggressive, the guardrails get thinner, the electorate gets more calcified. Redesign is possible, if we start now. After 2031, eulogy may displace strategy.

---

For more information on fusion voting and parties: Center for Ballot Freedom. For more information on proportional representation: Fix Our House.

---