Happy Monday. We are officially one day away from Election Day 2026.
Polls continue to suggest that tomorrow night will be a blowout for Democrats, though many in the party are panicked after a few late-breaking developments, from the New York Times reporting on Mary Peltola’s role getting a second Dan Sullivan on the ballot in Alaska to Democratic Rep. Haley Stevens’ last-minute endorsement of Republican Mike Rogers in the Michigan Senate race.
The biggest question I keep getting is: Can the polls be trusted? In my latest YouTube video, I share the biggest thing holding me back from fully trusting the numbers that we’re seeing. I hope you’ll give it a watch:
If you have appreciated my coverage this midterms season, please consider becoming a paid subscriber to support my work. I’ll be in your inbox Wednesday with full analysis of the results — and then will hit the ground running covering whatever the new Congress has in store for us.
But first, let’s take a step back and look at the 2026 midterms from a historical lens…
On Inauguration Day 2025, I quoted the political scientist Adam Przeworski, who has pithily defined democracy as a system in which “incumbent governments risk office in elections and yield it when they lose.”
By this definition, he has argued, the United States did not become a democracy in 1789, when George Washington was sworn in as the first president, nor in 1797, when his vice president and fellow Federalist John Adams succeeded him. It was only in 1801, when Adams gave way to the Democratic-Republican Thomas Jefferson — peacefully trading the White House across party lines — that the U.S. attained democratic status.
However, in the American system, there is not just the presidency to consider, but also the Congress; incumbent parties are not only expected to risk power, but also to share it, which might be an even taller ask. Perhaps, then, the U.S. did not reach the full extent of what its particular democracy demands for almost 40 years, until 1826, the first time voters elected both a House and Senate controlled by one party while the presidency was controlled by the other.
These was the first backlash midterms, and America’s first brush with divided government.
At the time, President John Quincy Adams was two years into his tenure. In his diary, even he acknowledged how unusual the situation was: “There is a decided majority of both Houses of Congress in opposition to the Administration— A State of things which has never before occurred under the Government of the United States,” Adams wrote on the day the new Congress organized.
No one quite knew what to do. “This, you know, would in England produce a resignation of the ministers,” future President John Tyler, then an opposition member of the Senate, noted in a letter. Indeed, in the parliamentary systems of Europe, there is no such thing as divided government: if the opposition defeats the prime minister’s party in an election, there is no thought that the two power centers will co-exist. The sitting government simply collapses. The prime minister comes from within the parliament, so it would be oxymoronic to think of the two branches of government ever working in conflict. It simply doesn’t happen. The executive and the legislature can only row in concert.
Divided government is an almost uniquely American phenomenon. In a 2004 study, Przeworski and two colleagues took every country that they classified as a democracy and, for each one, recorded every time the partisan distribution of seats in the legislature shifted between 1946 and 1999. So, for example, in the American context, each Congress would be one data point: Republican House, Democratic Senate in the 118th Congress; Republican House, Republican Senate in the 119th Congress; and so on.
In total, they counted 716 different partisan alignments across 133 different countries. Only 23 of these alignments yielded divided government (where the majority party of the legislature was different than the party of the chief executive), and 16 of those 23 took place in the United States. There were also two each in Colombia and the Philippines, and one each in Costa Rica, the Dominican Republic, and El Salvador. But that was it: in very few countries is it normal for voters to be given or take the opportunity to elect a legislature of the opposite party as the president.
But in America, it happens all the time.
Personally, I find this sort of beautiful. Every four years, we elect a president. Then, almost invariably, they act in a way that frustrates, or disappoints, or angers us. So, we get to rein them in. Midterm elections provide the ultimate form of accountability, ensuring that presidents either hew to the electorate’s wishes — or pay the price.
In Democracy in America, Alexis de Tocqueville noted that “several Presidents of the United States have been known to lose the majority in the legislative body without being obliged to abandon the supreme power.” Tocqueville said that he had heard some observers point to this as proof of “the power of the executive government in America” (the president can endure past an electoral drubbing!), but really, he wrote, “it is a proof of its extreme weakness” (even the president has to share).
Tocqueville was right.
Midterm elections weaken a president, allowing voters to tell him in the most democratic way possible — at the ballot box — exactly how they think his term is going, and to cut his knees out from under him if they think he deserves it. Presidents go too far; midterms keep them in check.
Political scientists often refer to the “thermostatic model” of American politics: how public opinion is constantly moving in the opposite direction of where the government is pushing public policy. When more liberal policy is pursued, the electorate becomes more conservative, and vice versa. The electorate functions like a thermostat, making sure the temperature in Washington never swings too far in either direction.
This is sometimes portrayed as indecision, but it can also be regarded as a rational response to presidents overreading their mandates and doing more than the voters wanted them to. Donald Trump was elected in 2016 to shake up the status quo, and then proceeded to shake it more than voters wanted, so they elected a Democratic House in 2018 and then a Democratic trifecta in 2020. But then Joe Biden, who campaigned as a centrist, governed as a progressive, so the thermostat swung in the opposite direction, electing a Republican House in 2022 and then a Republican trifecta in 2024.
And then.
Over the last two years, Donald Trump hasn’t overread his mandate so much as he’s ignored it.
Trump’s 2024 campaign could be boiled down to four key slogans: “No new wars,” “Drain the swamp,” “Make America affordable again,” and “Mass deportations now.”
Three of those he flip-flopped on to almost comical proportions; the fourth, he followed, but many Americans soon felt that he went too far. By launching the Iran war, engaging in almost cartoonish levels of self-dealing, and imposing a slew of tariffs that raised prices (as did the war), Trump violated campaign promise after campaign promise and gave voters a dramatically different presidency than the one that they expected.
As a result, the thermostat that he so keenly took advantage of in 2024 is now poised to turn against him.
There is no poll more trusted in the political world than the New York Times/Siena survey. Its final release last night produced mouth-watering results for Senate Democrats: an 11-point lead in Michigan, a seven-point lead in Texas, a six-point lead in Maine, five-point leads in both Kansas and Ohio, a two-point lead in Alaska, and a one-point edge in Iowa. Not just that, but the poll also found Democrats ahead in the governor’s races in each of these states as well, a full sweep of competitive elections.
If the Times/Siena poll is right, Democrats would be going from 47 Senate seats to 54 and from 24 governorships to at least 28 — both benchmarks they have not reached since the Obama wave of 2008. That was also the last time any party saw a lead in national House polling on the eve of an election greater than the nine-point edge Democrats now hold in the generic ballot polling average.
I don’t know whether that polling is correct. But regardless of whether a wave of such epic proportions washes over the country tomorrow, Trump’s party is expected to lose at least the House (as the president’s party almost always does in a midterm year), regardless of whether the Senate follows. He will soon be sharing custody of the American government.
As of this year, Americans have been proudly kneecapping their presidents in this way for two whole centuries. I don’t know if presidents talked about having “mandates” in the 19th century, but the 1826 midterms were similarly a response to John Quincy Adams overreading his.
In 1824, four Democratic-Republicans competed for the presidency. Andrew Jackson won a plurality (but not majority) of the popular and electoral votes, throwing the election to the House. Adams only won after a third candidate, Henry Clay, gave him his support; Clay was then appointed secretary of state, a move widely derided as a “corrupt bargain.”
Adams responded to this narrow (or, perhaps, non-existent) mandate by proposing an ambitious expansion of the federal government, including a major boost in infrastructure spending. Voters, in turn, responded by elevating Jackson allies to majorities in the House and Senate.
This first example of divided government didn’t go too well. Everything became a proxy battle for the 1828 rematch between Adams and Jackson (which Jackson ultimately won). Investigations of the executive branch flourished, even then. A Jackson ally physically assaulted Adams’ son in the Capitol Rotunda.
In the two centuries since, however, we’ve seen more promising examples of divided government: some of our country’s great legislative achievements have been produced via compromises between a president and Congress of dueling parties.
A Republican-led Congress approved Democrat Harry Truman’s Marshall Plan. (Why wasn’t it the “Truman Plan”? Because Truman knew that naming it after George Marshall, the former general respected on both sides of the aisle, instead of himself would increase its odds of bipartisan support.) A Democratic-led Congress worked with Republican Dwight Eisenhower on the Civil Rights Act of 1957, the first since Reconstruction. Divided government produced the Americans with Disabilities Act under a Democratic Congress and President George H.W. Bush, and the only budget surplus in modern history under a Republican Congress and President Bill Clinton, who were forced to compromise on fiscal matters.
Some political scientists have argued that Americans prefer divided government, at the state and national levels. It is the ultimate manifestation of the separation of powers, of ambition counteracting ambition. It is the voters telling two schoolchildren who are fighting: Share, or we’ll take your toys away.
The last time Donald Trump was in the White House without complete control of Congress, there were some notable achievements, as well. Yes, the Democratic-led House voted to impeach Trump in December 2019 — but the very next day, the chamber also approved the USMCA, a major North American trade deal, by a bipartisan vote. When Covid struck, Trump and Democratic lawmakers collaborated on the $2.2 trillion CARES Act, the largest economic stimulus bill in U.S. history.
In our forthcoming spin with divided government, there are surely issues for Trump and Democrats to work on, from reforming our immigration system to preventing Social Security from becoming insolvent in 2032.
It will be up to the president and Democratic leaders whether to work together on these critical issues — or whether to continue flouting the thermostat in their own ways: Trump, by continuing to accrete executive power against the voters’ will; or Democrats, by overreading their 2026 mandate and moving ever leftward, thus repeating the cycle over again.
Maybe the lesson of the thermostat is that Americans will never be satisfied: no matter what happens in a presidency, voters will always rebel against it.
But, also, it’s been some time since we’ve had a president who has at least tried to be respectful of public opinion and the will of the electorate.
Imagine a Trump administration that secured the border, but was careful in how it carried out its deportation campaign; that imposed some tariffs, but only targeted to certain countries and sectors; and held itself back from launching an unpopular war. That would not be a Trump administration at risk of losing the Senate tomorrow.
One day, perhaps, a new president will enter office conscious of the thermostat, trying to enact what he promised — and only what he promised — without moving so far in one direction that the electorate swings equally far in response. That president will likely be handsomely rewarded (or, at least, not so harshly punished).
Until then, the American thermostat will keep cutting over-aggressive presidents down to size, just as it has for 200 years and counting.