Brief us

WP-31: Terrain

The tide went out on the incumbent

What moved in the 2026 midterms, what did not, and why ballot measures register a wave before candidate races do.

By Brushworks research deskNovember 2026Working paper

Abstract. Large electoral swings are usually read as realignments and usually turn out to be tides: the same electorate punishing whoever governs. We apply that distinction to 2026 using the exit poll, national measures of trust and the November ballot measures, set out what a tide predicts for the next two years, and describe how a program reads the electorate weekly, separates switchers from voters who stayed home, and keeps a control.

In brief

The finding
The 2026 result reads as a tide, not a realignment: the winners share no program, the grievance crosses party lines, and trust has not recovered. Mandates won this way are borrowed.
For a campaign
Win the term, then govern as though the vote were on loan. The persuasion window for 2028 opens with the first budget.
For a company, an association or a public body
Public patience with large institutions will be tested by direct vote. Know the terrain before the question is written.

The 2026 result was large and it will be over-read. The evidence points to a tide: voters withdrawing consent from whoever governs, on the cost of living, with trust near its floor. A tide lifts candidates of many kinds and then turns on them. This paper sets out how to tell a tide from a realignment, what follows for the next two years, and where a program should be looking.

What happened, stated carefully

Seats and statehouses changed hands in states that had not been competitive for a generation, and coverage reached for the word wave.1

The campaign that preceded it was fought on prices and on trust. In the national exit poll, 89 percent of voters called the cost of living a crucial or major factor in their vote for Congress, and 90 percent said the same of government corruption.2 On the one new issue of the cycle, voters in ten states sided against data center development in all but one of some forty local measures, with majorities of both parties telling pollsters they favored limits.1

What voters said decided their vote for Congress, share calling each a crucial or major factor. Basis: official, national exit poll of November 3, 2026, as reported (note 2).

Tide or realignment

A realignment is a durable change in who votes with whom. A tide is the same electorate punishing the party in power. The literature on retrospective voting is unsentimental about which is more common: voters reward and punish on recent conditions, sometimes on conditions no officeholder controls, and the judgment is renewed at every election.3, 4

Telling a tide from a realignment: what each predicts, and what the 2026 record shows. Basis: the distinction as drawn in the retrospective-voting literature (notes 3 and 4); the right-hand column summarizes this paper.

Three things in the 2026 record read as tide.

The winners do not share a program. The same night rewarded candidates who ran as insurgents and candidates who ran as managers. One post-election analysis finds moderates still outran their party’s baseline, by up to about three points, and that the gap to the party’s left was the narrowest it has measured.5 When candidates of every type are lifted, the water is doing the lifting.

The grievance crossed party lines. Eighty percent of adults, including three quarters of Republicans and nine in ten Democrats, say political institutions are controlled by wealthy and powerful interests.6 A sentiment held at that level on both sides is not one side’s coalition. It is the weather.

Trust did not recover. Average confidence in the institutions Gallup tracks was 27 percent in June, a point above the record low.7 In September, the share trusting the American people as a whole to make judgments under the democratic system was 51 percent, the lowest Gallup has measured.8 A realigning electorate believes in something. This one is withholding.

Trust in 2026, three readings. Basis: official, Gallup and Economist/YouGov as published or as reported, field dates in notes 6 to 8.

What a tide does next

It turns. The party that was carried in now holds the offices, the budgets and the blame. The voter who decided 2026 on the power bill will open another one in 2027.

Mandates are borrowed. A candidate elected against the incumbent has been given a term, not agreement. Programs written as though the electorate had endorsed a platform will meet the same voter in a different mood.

Primaries are where the argument moves. In a wave year the general election settles little inside the winning party. The contest over what the result meant is held in low-turnout primaries, in safe seats, among small and attentive electorates. WP-30 takes up what a candidate with a record does there.

The ballot measure is where the tide reaches shore first. A measure has no incumbent to absorb the anger and no party label to sort the vote. It asks a direct question and gets a direct answer, which is why forty local votes this November said more about the public’s view of one industry than any candidate race did. Expect more of them, on siting, rates, taxes and the terms on which large institutions operate in small places. Expect them earlier, and expect them to be drafted by the side that organized first.

What a program does about it

Read the tide weekly, not by cycle. A map of 2024 was a poor guide to 2026, and a map of 2026 will be a poor guide to 2028. The quantities that hold are behavioral: who votes in which elections, who switched, who stayed home, what a household’s costs did. A program reads them on a standing schedule and revises the plan on the reading.

Find the switchers and the stayers-home. A tide is made of two groups: people who changed sides and people who did not vote. They need different things from a campaign, and they are found by what they did, not by a demographic guess at who they are.

Plan for the turn. For an officeholder elected this year, the persuasion window for 2028 opens with the first budget. For an institution, the window opens before the petition is filed. In both cases the cheapest month to be heard is the one in which nobody else is speaking.

Keep the control. In a tide, almost any program appears to work, because everything on the winning side moved. A tenth of every cohort held out is the only way to know what the program did and what the water did.

The ballot measure as a reading of trust

A ballot measure is the closest thing an electorate has to a direct answer, and in a tide it is the most legible one. There is no incumbent to absorb the mood and no party label to sort the vote, so what a measure records is the electorate's view of the question and of the people asking it.

The research on direct democracy has two findings that matter here. Voters facing a measure they know little about lean on cues: who is for it, who is against it, and whether the side they distrust is paying for it.9 And when the cues are unclear, the default is no. Measures are easier to defeat than to pass, because a voter who is unsure and suspicious has a safe answer available.10

Put those findings next to the trust numbers in this paper and the forecast for the next two years writes itself. Measures that ask the public to approve something a large institution wants will start behind, because the cue that matters most (who benefits) points at the institution. Measures that let the public restrain a large institution will start ahead for the same reason. November's local votes on data centers were the first large instance. They will not be the last, and the next ones will be on rates, on taxes, on land and on the terms of public money.

The next two years as a calendar for a program, from the 2026 result to the 2028 general election. Basis: illustrative; dates are seasons, not a jurisdiction’s schedule.

What this means for the side that wants a yes. The burden is to replace the cue. A yes campaign that runs on the institution's standing is asking a suspicious voter to trust the very thing she distrusts. One that runs on a checkable household result, said early, by a disclosed messenger the voter already knows, gives her a reason that does not depend on trust. The cases in WP-03 and WP-09 are the record for that approach.

What it means for the side that wants a no. The default is on its side, which makes it complacent. A no campaign still has to be true and specific, because the voter who is suspicious of institutions is also suspicious of campaigns. Every line sourced; nothing that cannot be shown.

What it means for anyone reading the result. A measure won or lost in a tide says something about the question and something about the weather. A program that held out a control can tell the two apart. One that did not will learn the wrong lesson from its own success.

Tides before this one

American political history offers many elections that were read as realignments when they happened and look, in retrospect, more like tides.

The pattern is familiar to historians of midterm elections. A party wins power, governs, and loses seats at the next midterm, often heavily, as voters register their judgment of the people in office. Commentators describe the loss as a repudiation of the governing party's program and the start of a new era. Two years later, the electorate turns again.

Not every large swing is a tide. Some elections do mark durable changes in who votes with whom, and the literature on realignment has argued for decades about which ones. The point of this paper is not that realignments never happen. It is that they are rarer than the coverage of any given election suggests, and that the evidence available a few days after a vote rarely distinguishes the two.

What distinguishes them, eventually. Whether the new coalition holds when conditions change. A realignment survives the next bad year for the party that benefited. A tide does not. That test cannot be run in November 2026; it will be run in 2027 and 2028.

What a program does in the meantime. It treats the electorate as provisional. It reads behavior on a standing schedule rather than drawing conclusions from one result, it holds out controls so that it can tell what its own work did from what the water did, and it plans for the possibility that the voters who decided this election will decide the next one differently.

Common mistakes

Reading a mandate into a tide. An electorate that punished the incumbent has not endorsed the challenger’s program.

Drawing the next map from this one. A map of 2026 will be a poor guide to 2028. Behavior read weekly is a better one.

Ignoring the stayers-home. A tide is made of switchers and of people who did not vote. They need different things.

Crediting the program for the water. In a tide, nearly everything on the winning side appears to work. Only a control can tell what did.

Implications

For a campaign. Win the term, then govern as though the vote were on loan. Start the next program at the first budget.

For a company, an association or a public body. The public’s patience with large institutions is the scarce resource of the next two years, and it will be tested by direct vote. Know the terrain before the question is written.

For both. A measure is won or lost on one sentence a household can check, said early. WP-03 sets out the case; WP-29 applies it to a siting.

Further reading

For readers who want to go further, these are the works this paper draws on most, or that argue the case at greater length. Each is listed with a line on why it is worth the time.

  • Achen, C. H., and Bartels, L. M. (2016). Democracy for Realists. Princeton University Press. The argument that voters judge incumbents on conditions, not programs.
  • Hetherington, M. J. (2005). Why Trust Matters. Princeton University Press. How political trust shapes what governments can do.
  • Sides, J., and Vavreck, L. (2013). The Gamble: Choice and Chance in the 2012 Presidential Election. Princeton University Press. How fundamentals and campaigns combine in a national election.

Notes and references

  1. Al Jazeera, November 7, 2026; FOX 4, November 2026. Counts of local measures differ by source (41 and 44).
  2. National exit poll conducted by SSRS for a consortium of news organizations, November 3, 2026, as reported in election-week coverage; see also Data Center Dynamics.
  3. Achen, C. H., & Bartels, L. M. (2016). Democracy for Realists: Why Elections Do Not Produce Responsive Government. Princeton University Press.
  4. Healy, A., & Malhotra, N. (2013). Retrospective voting reconsidered. Annual Review of Political Science, 16, 285–306.
  5. The Argument, November 12, 2026; preliminary, with ballots outstanding.
  6. YouGov, Economist/YouGov poll, December 26 to 29, 2025.
  7. Gallup, June 1 to 15, 2026, as reported by Axios, July 14, 2026.
  8. Gallup, Governance survey, September 1 to 17, 2026.
  9. Lupia, A. (1994). Shortcuts versus encyclopedias: information and voting behavior in California insurance reform elections. American Political Science Review, 88(1), 63–76.
  10. Bowler, S., and Donovan, T. (1998). Demanding Choices: Opinion, Voting, and Direct Democracy. University of Michigan Press.
  11. Healy, A., and Malhotra, N. (2013). Retrospective voting reconsidered. Annual Review of Political Science, 16, 285–306.

Basis of figures. Election results: news reporting from November 4 to 12, 2026, not certified. Polls: as published or as reported, with field dates in the notes. No Brushworks program figures appear in this paper.

Next paperThe Review

Start a conversation

Brief us.

Tell us the race, the calendar and the margin. Within a week we will tell you whether we can move it and what it would take. Candidate work and independent-expenditure work are taken by separate teams; say which you are. Governments and institutions: tell us the mandate and the jurisdiction, and we will tell you what registration it requires before we say anything else.

Send a briefbrief@brwrks.com

  • Candidate practice
  • Independent-expenditure practice
  • International practice

Brushworks Strategies Inc., a Wyoming corporation · Washington, New York, Ottawa, London, Munich, Dubai, Tokyo