ProPublica released the key report in the contest between the two candidates for the Maine Senate seat this morning.

It wasn’t a leak or opposition research; it was a properly documented account of how the Hawaii-based defense contractor Martin Kao set up a system in Washington for receiving cash in return for contracts—according to Kao’s own version of events given to the FBI, Susan Collins was the most important supporter of this arrangement.

I’m going to stick to what’s actually on the record, not the campaign’s talking points.

If you have only two minutes, start here: a federal contractor unlawfully transferred $150,000 from a fictitious “women scientists” LLC into a Collins super PAC. Several weeks later, the CEO informed his own executives that Collins had carried out $32 million worth of work for the Navy. The boats the Navy paid to have designed were never delivered. The CEO is now in federal prison. The anti-corruption unit, which was still raising questions in 2024, was abolished after Donald Trump returned to office. The case is now over.

That concludes the story. What follows is the path that led up to it.

What is actually documented

Start with something that is not based on Martin Kao’s word.

In 2018, Kao asked for a meeting with Collins and was told he had to donate first; he did. During the same cycle, Collins arranged for an $8 million Navy research appropriation to be included in the defense bill for Navatek’s work on the hull with the University of Maine — the 3D-printed boat project based in Portland.

Navatek held a publicity event for a model boat at NavaTek Portland Office in August 2019.

In Washington, Scott Reed—who then was, and is now, a key figure in Collins’ outside-money operation—met with Navatek executives at a Corner Bakery. He requested that $500,000 be paid to the 1820 PAC, the super PAC supporting Collins in the fierce 2020 campaign. Afterward, when the FBI interviewed Navatek executives, they said their request in return was a guarantee of tens of millions more in federal contracts and a promise that Collins would know who had written the check.

Soon after the meeting, $150,000 was deposited at 1820 PAC from a recently established Hawaii LLC named the Society of Young Women Scientist and Engineers, although they had misspelled “Scientist.”

The funds did not originate with a women’s STEM charity; instead, they came from Navatek’s account. Federal contractors are not allowed to make such a donation. This is not a grey area; Kao later pleaded guilty to violating the law.

A few weeks after the check cleared, Kao and his team received a “very warm reception” at Collins’ office. In an internal email examined by ProPublica, Kao told his executives a version of the sentence that should appear on each of the flyers in this race:

“We were very warmly received. Excellent meeting. Total of $32M will be supported.”

Collins, in a meeting Kao later described to agents, put it more cleanly:

“You’ve seen me deliver.”

Then her staff requested the CEO’s phone number so that they could thank him personally.

That sequence—access, money, use of appropriation language, and the thank-you call—is what is known as the Washington business model. ProPublica discovered checks being sent within 24 hours of key hurdles in the defense bill. According to the outlet, Collins was ‘its most important patron’ among all the members Navatek had approached.

What the Navy actually got

The section that should anger anyone who has ever completed a bid sheet or waited for a federal reimbursement.

According to the company itself, the government-sponsored projects it carried out in Maine produced almost nothing useful to the fleet. An ex-employee told ProPublica that the boats were never delivered because “the Navy didn’t want them at all.” The prototype eventually made its way into PowerPoint presentations and a white paper. One executive described the company’s approach as learning how to “thrive in the valley of death”—the area where research funds go and die without ever becoming a product.

Maine got the press conference.

Navatek got the contracts.

Collins got the PAC money.

The Navy got a slide deck.

If a shop in Brewer were to do that in a municipal tender, there would be a select-board meeting, and the county attorney would be involved. That procedure is known as appropriating in Washington.

Who Martin Kao is — and why that matters

He should not be turned into a choirboy. WONKSTACK isn’t involved in creating saints.

When Kao acquired Navatek, he used it as a political cash machine. With his network, he transferred almost $900,000 to politicians of both parties—yes, to Collins, but also to some names that will make certain of my readers recoil: Jim Clyburn, Dick Durbin, Jeanne Shaheen, Joe Biden’s operation, and Lindsey Graham. He unlawfully paid back relatives who had written checks. He was later convicted of a $13 million PPP fraud and, in September 2024, received a sentence of 87 months.

He went to the FBI after he had been caught; that’s what cooperators do. They also combine a desire for revenge with factual information. This is why you don’t draw up an indictment based solely on a cooperator’s testimony — and why the emails, the check through the shell company, the FEC complaint, the appropriations language, and the contemporaneous internal memos matter more than any single interview.

Kao’s line to the agents was crude and, if you have ever watched this town work, not hard to believe:

“It takes two to tangle.”

He spent several days going over the method with them, giving them a 50-page map and naming the staffers, the lobbyists, and the members. By the end of 2024, the agents in the Bureau’s public-corruption unit, CR-15, believed they had enough evidence to launch a wider bribery investigation—not a press release, but an investigation.

Here’s where the path divides.

What happened to the investigation

Two different things are being flattened into one sentence online today. Separate them.

First investigation, 2020–2021. The FBI raided Navatek’s Honolulu offices in September 2020. Prosecutors charged Kao with campaign-finance crimes and, later, PPP fraud. The Collins campaign says it cooperated, produced documents, and in 2021 disgorged the illegal Kao money it says it did not know was illegal — some of it, per reporting today, sent to rebuild a school in Frenchville that burned down. Deputy chief of staff Annie Clark calls Kao’s later accusations “outlandish” and “categorically false.” Ben Williamson, a spokesman for the FBI, said to ProPublica that the Bureau “ultimately found nothing that implicates Senator Collins or Senator Collins’ campaign. Any suggestion to that effect is completely false.”

The New York Times has now added a second official layer by having current and former officials familiar with the inquiry state that the agents found little evidence to support Kao’s specific allegations against Collins and Reed.

That’s nothing; it’s the defense, and it’s on the record.

Second inquiry, 2022 to 2024. Once Kao pleaded and started talking, CR-15 responded. It involved multiple-day interviews. The 50-page document contained questions regarding the Corner Bakery, questions about Collins, questions about Graham and a steak dinner in South Carolina, and questions about a list of members who had taken Navatek money while Navatek had taken public money.

ProPublica has shown that this second phase existed, that agents wanted to carry it out, and that it died down after January 2025.

The jury didn’t cause this; a personnel decision did. When Trump had Kash Patel as FBI director, he got rid of the unit that dealt with public corruption. The agent who had taken Kao’s confession — Michelle Ball — was dismissed in October 2025. According to the new organization chart, her offense was having previously worked on the 2020 election file. Her partner in the Navatek case, Kevin Gounaud, was removed in 2026. At the same time, the Justice Department’s Public Integrity Section was weakened. Prosecutors were instructed to halt work on any new corruption cases.

ProPublica’s sentence is straightforward: the investigation is now over. The government had stopped communicating with the informant who had provided them with a map showing how a mid-size contractor acquired a. You can regard Collins as having done nothing wrong and still consider it a scandal that the unit that could have tested that assertion was disestablished for reasons unrelated to Maine. Those two ideas can coexist in the same mind. That is what this publication is about. That is the entire point of this publication.

The part that is still sitting in Augusta

Scott Reed conducted the bakery meeting in 2019.

Amy Abbott ran the finance operation that handled the Kao network.

Both of them are mentioned in the news published today.

In 2026, both of them are still handling Collins’s money.

It isn’t an accusation of a new crime; it’s a personnel decision. The same company that cashed the check from the shell firm is now asking Maine voters for another six years.

Collins is positioning herself as the adult in the room, as the independent and as a buffer against Trump. Fox and the Portland Press Herald have already given the figure which makes that argument difficult to sustain: 55 percent of Maine voters are concerned about her connections with him. CNN’s recent portrayal of the race is even more straightforward in that it says the campaign is being waged on negative views of both Collins and Trump.

A senator who needs the FBI under Trump to close the file isn’t entitled to act as a check on Trump; that isn’t a matter of ideology, it’s a matter of arithmetic.

What this is not

Susan Collins was not indicted.

A jury has not found evidence of a bribe. In fact, no jury was appointed.

That doesn’t mean every dollar Navatek received was valueless; some of the UMaine composite work involves genuine research, even if the fleet never purchased the hull.

This is also not a claim that only Republicans do this. Kao’s donor list is a bipartisan embarrassment. If the next ProPublica chapter lands on a Democrat with an appropriations gavel, WONKSTACK will write that piece too. The method does not belong to one party. It belongs to the permanent class that treats the defense bill as a customer-relationship management tool.

It is a documented pipeline going from a contractor who had been banned, via a dummy LLC with a typo in its name, to a super PAC, alongside appropriations legislation and a thank-you call, referred to as pay-to-play by the person who managed the company, before being suppressed when the only team member left to investigate the claim was dismissed as a result of political revenge.

If that sentence leaves you unaffected, then you are not moderate; you are merely a spectator.

The Maine test

I have a simple test for this kind of thing.

If a lobsterman in Stonington had arranged for a no-bid contract to go to his cousin and had taken $150,000 via a dummy LLC, then we would know exactly what to call it. In that case, we wouldn’t have to carry out a 6,000-word investigation or use terms such as “patron,” “earmark,” and “disgorgement.” We would name it properly and expect a sheriff.

Washington has a different dictionary because Washington writes the dictionary.

Maine isn’t a company town for Navatek; it’s a state that has already seen its young people move away, whose harbors are competing for their share of the quota, and whose shipyard in Bath is actually doing the tough work—building the kind of ships the Navy wants. To accept the view that we should ignore a contractor who admitted the product had ‘no real value’ simply because the press conference took place in Portland is exactly how a state talks itself into losing its self-respect.

Collins has been a member of the Senate since 1997, a period longer than the time several voters in this election have been alive. Eventually, the argument that ‘she’s effective’ must meet up with the actual evidence of her effectiveness.

Effective at what?

For whom.

At what price.

They are not partisan questions; they are the only questions.

Read it yourself

Primary source, published this morning:

ProPublica — “The FBI Anti-Corruption Squad Was Circling Susan Collins — Until Trump Got in the Way”

— Maine

🫎🇺🇸🦞
X
Instagram
Facebook
TikTok
BlueSky
wonkstack.com

The Republicans have the money while the Democrats are out of it. But that does not end the midterms; it only tilts the money race. That is the setup, not the story.

That’s not analysis; it’s just a press release with a masthead.

Right: the committees and the super PACs are filled. MAGA Inc. entered September with around $416 million. Republican-linked groups intend to spend about $221 million more on advertising than the Democrats from this week until Election Day. The amounts set aside are approximately $682 million for the GOP and $491 million for the Democrats. Total advertising spending in Senate, House, and state races: $888 million compared with $666 million.

That section is genuine; treat it as such, and then read what follows as the next step.

Now consider the effect of the money. A war chest is not the same as a majority, and a TV reservation is not the same as a coalition that still exists. This piece answers that question.

Two piles of money

Don’t regard “money” as if it were a single figure.

There is so-called candidate money—that is, the cash belonging to a candidate as listed on the ballot, which enables him to secure lower advertising rates, and which must be raised from real people.

Funding also comes from outside sources—such as party committees, super PACs, MAGA Inc., and the twenty richest people in the country.

That pile isn’t the same one, and the distinction matters.

In August, Democratic and Republican candidates raised about $1.5 billion and $1.1 billion, respectively. Nine out of the ten largest individual fundraisers this cycle are Democrats. Jon Ossoff raised $77 million. James Talarico raised about $69 million. Before the flow of funds was increased, Ken Paxton’s campaign was in the single digits.

Read it again.

The party which is said to be ‘broke’ is winning the only money race that still resembles a campaign. The party with the ‘advantage’ is winning the money race that resembles a rescue, not a victory. That is the difference that matters.

Candidate adverts are less expensive than those run by super PACs. That is why ‘who has reserved more television time’ and ‘who can actually talk to voters at scale’ are not the same question: one is a campaign, the other is a florist’s budget at a funeral.

If you print just the second number, you’re doing PR. If you want the story, keep going.

Watch where they spend it.

September exposed them.

For the first eighteen days of the month, advertisers backing the Republicans in ten important Senate races spent over $204 million, which is $83 million more than the amount spent by the Democrats. This marked a change from August, when the Republican lead in those races had been around $9 million.

The biggest deficit was in Texas, where the amount was $62 million less than that of Paxton this month, compared with $17 million behind Talarico. CNN says the GOP is on course to spend 8 times as much as the other side in that race by Election Day.

You wouldn’t burn that much cash in a state Trump won by thirteen points if you felt at ease.

This action follows the 2024 ‘realignment’ and now shows it plainly.

I have already set out the Texas version of this story: they did not start a rescue operation worth nine figures because they were ahead of the situation, but because the internal situation had left them in a state of great anxiety. The present article is the national version of that same story.

The money isn’t paying for new states; it’s merely renting oxygen for seats that were previously automatic. That is what the spending is for.

The defense map

Here’s what they now have to pay for.

Texas: Trump leads by 13.6 percent in 2024, and the state is now a Cook toss-up. A Democratic state representative has outraised the Texas attorney general by a margin that would have been ridiculed two years ago.

It’s Ohio. Trump leads by 11. Sherrod Brown isn’t meant to be in this race, yet he is.

It’s Alaska. Trump leads by 13. Mary Peltola isn’t meant to be in this race, yet she is.

Iowa. Trump leads by 13. Open seat. Toss-up.

It’s Maine, Susan Collins, the founding member of the ‘she always survives’ franchise, and Harris has won that state. The race is a toss-up. Collins had a personal brand strong enough to overcome a national collapse; that brand is now just a line item.

Kansas has not had a Democrat appointed to the Senate since 1932; JD Vance had to fly in, too.

It’s Nebraska, and Trump is leading by 20 points. They are treating an independent candidate as if one were possible and are spending money as if that were the case.

Cook’s list of Senate races considered toss-ups now includes Texas, Ohio, Alaska, Iowa, Maine, Michigan, and New Hampshire. That isn’t the party being on the attack; rather, it’s the party realizing that a ‘Trump state’ has never actually been the same as a ‘safe Republican state’ when Trump isn’t the candidate on the ballot.

The movie here is the same one, only with more precincts.

The DCCC have placed 58 seats that are held by members of the GOP on the board. They have included Lauren Boebert. They have included Richard Hudson—the person who runs the National Republican Congressional Committee. They have included Alaska’s at-large seat. The NRCC is aiming for 36.

One side is placing more squares into play, while the other is writing larger cheques to avoid losing the squares it already has.

Redistricting was meant to be the solution. Republicans redrew maps in several states, hoping to secure a double-digit seat gain. But the environment swallowed most of that gain. In Texas, Florida, Ohio, and North Carolina, districts drawn to be safe are still competitive. You can’t create a coalition again by gerrymandering.

The 2024 majority is not a bloc.

Money cannot remedy this. That is the point.

A living majority coalition enters the midterm elections and grows; a dead majority coalition does so and has to rely on a slush fund.

From the American Research Group, between September 16th and 20th: approval of Trump stood at 29 percent and disapproval at 68 percent. On the economy, approval was 26 percent and disapproval 71 percent. That is not a favorable outcome mid-term. That is a ditch.

NBC reports that 43 percent of voters say their vote in Congress signals opposition to Trump, while 27 percent say it signals support; among independents, almost half say it signals opposition and a small number say it signals support.

The Economist’s account of the 2024 electorate contains the most straightforward sentence: 76 percent of those who support Harris intend to vote for the Democratic Party in November. 65 percent of Trump’s supporters plan to vote Republican. One in five of his own 2024 voters now disapprove. Most of the others are not changing their minds; instead, they are shrugging, staying home, or writing ‘undecided’.

The New York Times and Siena, among likely voters, showed Hispanics supporting the Democratic Party 68 to 25, those under 30 64 to 29, and independents 57 to 34; these groups were the ones they presented as the new durable majority. Twenty months later, they don’t behave as if they form such a majority.

The coalition in 2024 wasn’t a conversion; it was a rental, and the receipt was due.

Money can secure the airwaves, but it can’t bring back people who have already left.

What the headline is for

The purpose of the story about the “GOP money advantage” is to keep you from looking at the map. That is the misdirection.

When you examine the committees, the Republicans are richer. When you look at MAGA Inc., the Republicans are richer. When you consider the twenty largest donors, most have opened the vault for the GOP even though the party is performing poorly in public polls. That isn’t a working-class revolt; it’s a bidding war.

If you examine the candidates’ accounts, Democrats are collecting more money from people who still plan to live in the area.

When you look at the target lists, Democrats are gaining seats while Republicans are holding on to seats that didn’t need a defense budget.

If you examine the coalition, the majority in 2024 is beginning to break away from the demographic groups that they had boasted about.

So when a consultant gets on television and says “but the money,” translate it:

They’re spending more because the coalition has already collapsed; it is not because they are about to run the table.

They’re paying ransoms based on the previous map.

That is not strength. That is a dead majority with a credit card, not a living coalition.

Follow the map. Not the florist budget for a funeral. That is the only way to read this race.

-Maine
🫎🇺🇸🦞
X
Instagram
Facebook
TikTok
BlueSky
Site: wonkstack.com

Two new polls were released today. Neither shows a 14-point Republican advantage.

ReconMR/Siena has James Talarico up six, 49–43. Emerson/Nexstar/The Hill has him up one, 47–46, with a positive favorability score against Ken Paxton’s underwater numbers. Mason-Dixon earlier this week: 46–43. 270toWin’s five-poll average as of this morning: Talarico 47.4, Paxton 44.4.¹

This is not a close race that favors Republicans, given Texas’s history. A Democrat is consistently leading in a state Donald Trump won by 13.7 points just 22 months ago. As the Texas Tribune’s tracker noted this morning, Talarico is showing larger leads than Beto O’Rourke ever achieved against Ted Cruz in 2018. This is the clearest sign yet that the race is moving away from the 2024 baseline.

The cause is clear: the 2024 coalition is fracturing along the same lines that once unified it, and the polling below shows how that fracture is now showing up in the race.

Composition of the 2024 coalition

Certified 2024 result: Trump 6,393,597 (56.14 percent), Harris 4,835,250 (42.46 percent). Margin: 13.68 points and 1.56 million votes.³ In 2020 it was Trump 52.06, Biden 46.48—a 5.58-point race.⁴ The Republican presidential margin in Texas more than doubled in four years.

Two groups were responsible.

Hispanic Texans. Edison’s Texas exit poll put Trump at 55 percent of Latino voters, up from 41 percent in 2020 and 34 percent in 2016.⁵ Democratic ID among Texas Latinos had already fallen from 65 percent in 2016 to 45 percent by December 2024.⁶ The Valley flipped. Starr went Republican for the first time since the 1890s. Hidalgo went Trump. That was the earthquake.

College-educated Texans. Nationally, Pew had Harris winning degree-holders by 16 points.⁷ In Texas, Edison had college graduates tied 49–49. Non-college went 61–38 for Trump. White college graduates still leaned Republican, 57–41.⁸ Republicans did not win the professional class. They stopped the bleeding. That is the difference between a six-point state and a 14-point state.

To summarize: a spike among Hispanic voters combined with a stalemate among college-educated voters created the majority. That same coalition is now breaking apart, and recent polling indicates those gains are not transferring to Ken Paxton.

The Hispanic majority has shifted.

Emerson’s director, Spencer Kimball, said the quiet part this morning: “Texas Hispanic voters, who broke for Trump in 2024, now break for Talarico by 15 points, 54 percent to 39 percent. These voters disapprove of the president 53 percent to 41 percent.”⁹

This result aligns with other findings that reinforce the same shift:

Cook and Univision surveyed Rio Grande Valley districts that showed the strongest support for Trump in 2024. Statewide, Hispanic likely voters favor Talarico by a two-to-one margin. Trump’s job approval among Texas Hispanics is 33 percent. Eighteen percent of Hispanic Trump voters from 2024 now disapprove of his performance. Nearly two-thirds are less likely to support a candidate who backs his deportation policy. The economy remains the top issue, surpassing immigration by a two-to-one margin.¹⁰

James Henson at the Texas Politics Project cautioned last winter that using 2024 as the Latino baseline “is going to overestimate Republican turnout. Period.”¹¹ His assessment was accurate. The 55 percent Latino support for Trump was specific to that candidate and moment, and that is why it does not extend to Paxton.

The stalemate among college-educated voters has also ended, as the next surveys show.

August Texas Politics Project crosstabs: four-year degree Talarico 47–37, post-grad 50–38, independents 37–11, moderates 47–16, women 46–33.¹²

Mason-Dixon this week: women 52–36; Talarico, ages 18–34, 51–37; independents 51–30.¹³

According to ReconMR, independents now favor Talarico 64–24, up from 53–30 in July. In Dallas–Fort Worth, Talarico leads with 58 percent, while Paxton leads in rural Texas with 60 percent. These results show the same pattern: the coalition that lifted Trump is no longer holding together in the same way.

Paxton continues to lead among white voters, men, East Texas residents, and the non-college rural base. While these groups matter, they are no longer enough to win. White eligible voters are no longer the majority in Texas; Hispanic eligible voters now comprise one-third and are increasing. If Hispanic voters participate as they did in 2024, Paxton will win easily. If they vote as they have in September 2026, he will not.

The September board

Emerson provides the most recent data this week, shifting from Paxton leading by one point in August to Talarico leading by one after the GOP midterm convention in Dallas. In that survey, Talarico’s favorability is 48–42, while Paxton’s is 40–48.⁹ A candidate trailing in favorability in their own party’s stronghold cannot consider the seat secure.

Cook reclassified the race as a Toss-up in August. As Jessica Taylor noted, this contest would not be competitive if John Cornyn were the nominee.¹⁵ Paxton defeated Cornyn 63.8–36.2 in May, but one-third of Republican primary voters opposed him. ReconMR’s strong independent support for Talarico and the consistent 5–12 percent Republican shift toward Talarico reflect that reluctance more clearly.

Fundraising data demonstrates mounting concern. Talarico raised approximately $72 million compared to Paxton’s $16.8 million through mid-year, and held $21.5 million in cash to Paxton’s $1.8 million.¹⁶ Major Republican PACs began significant spending in September only after polling indicated a competitive race. Large financial investments are made when a race is at risk, not when it is secure.

The path to an upset is clear, and it follows from the shifts above and the thresholds below.

Texas remains structurally Republican. A Talarico victory would not represent a fundamental shift, but rather a targeted campaign against a weakened nominee during a midterm election. Democrats at 90-plus. He is there because the race is about exploiting that opening, not remaking the state.

1. Win independents by 20 to 40. He is there, and ReconMR says the gap is widening.

2. Take Hispanics from 45–55 Trump back to 55–40 Talarico. Every serious survey this month has him at or past that.

3. Win college-educated voters instead of tying them. The crosstabs already show it.

4. Peel enough Cornyn Republicans to cover the rural blowout.

If these five objectives are met, the 2024 electoral map will not secure a win for Paxton. Midterm elections typically disadvantage the president’s party. Trump’s Texas majority rested on economic and border issues, but the same coalition is now fraying. While economic concerns remain, the border message is now tied to an impeached attorney general with a negative favorability rating, while Talarico is focusing on child care and cost-of-living issues.

For three decades, Democrats have anticipated gains in Texas. In most cycles, they selected candidates who were not well-matched against their Republican opponents. This cycle, they have nominated a candidate focused on cost of living, running against the one Republican who could turn a 14-point Trump victory into a competitive race that requires major GOP investment.

The 2024 landslide, the shift among Hispanic voters, and the college-educated vote patterns were all significant. But those trends have now reversed, as the polling above and the board below show. So assuming this seat is secure based solely on Trump’s previous victory is no longer accurate.

Talarico is not relying on a potential coalition shift; the polling above shows it has already occurred. That is why he leads now.

-Maine
🫎🇺🇸🦞
X
Instagram
Facebook
TikTok
BlueSky
Site: wonkstack.com

Notes

1. ReconMR poll of 614 likely voters, Sept. 8–11, 2026, Fort Worth Star-Telegram, Sept. 17, 2026; Emerson College Polling / Nexstar / The Hill, 1,000 likely voters, Sept. 12–14, 2026; Mason-Dixon / Telemundo, Sept. 8–10, 2026; 270toWin Texas Senate average, Sept. 17, 2026.

2. Texas Tribune poll tracker, updated Sept. 17, 2026.

3. Texas Secretary of State, Official Canvass Report, November 5, 2024 General Election; Associated Press / Texas Tribune certified totals.

4. The American Presidency Project, 2020 Texas presidential results.

5. Edison Research / National Election Pool Texas exit poll, November 2024, as reported by the Texas Politics Project and Axios San Antonio.

6. Jim Henson and Joshua Blank, “Trends in Latino attitudes in Texas foreshadowed Trump’s gains in 2024,” Texas Politics Project, January 13, 2025.

7. Pew Research Center, “How voting patterns changed in the 2024 election,” June 26, 2025.

8. Edison Research / National Election Pool Texas exit poll, education breaks (Washington Post Texas exit-poll interactive).

9. Emerson College Polling / Nexstar, Sept. 12–14, 2026; Spencer Kimball comments reported by USA TODAY Network, Sept. 17, 2026.

10. Cook Political Report and TelevisaUnivision poll of 1,000 Texas Hispanic likely voters, OnMessage Inc. and Hart Research, August 26–September 3, 2026.

11. James Henson, quoted in CNN Politics, December 7, 2025.

12. University of Texas / Texas Politics Project statewide poll and crosstabs, August 5–13, 2026.

13. Mason-Dixon Polling & Research / Telemundo, Sept. 8–10, 2026, USA Today and El Paso Times.

14. ReconMR / Siena, Sept. 8–11, 2026, Fort Worth Star-Telegram.

15. Jessica Taylor, Cook Political Report, May 26 and August 20, 2026.

16. Texas Tribune, campaign-finance reporting, July 20, 2026; FEC filings through June 30, 2026.

The MAGA base is still standing.
The majority that won 2024 is not.
Young men, Hispanics, Black switchers, and independents rented Trump on a price promise.
Twenty months later the receipt came due — and they’re gone.

Donald Trump’s 2024 victory was not due to a nationwide shift toward MAGA ideology. Instead, he secured support from voters outside his core base, including Hispanic men, young men, some Black voters, and working-class households who had traditionally voted Democrat. These voters were not joining a movement; they were seeking tangible outcomes such as lower grocery prices, improved border security, and reduced political turmoil.

Twenty months later, that expanded support has disappeared. Recent national surveys do not indicate a gradual decline or realignment; instead, they show the additional 2024 voters returning to previous patterns.

The MAGA core remains intact, but the broader 2024 coalition does not. These are distinct groups. “Broken beyond repair” does not mean Republicans have disappeared or that the 2028 election is decided. It means the majority coalition from 2024 has already fractured, with insufficient time this cycle to rebuild it.

This analysis draws on multiple sources, including New York Times/Siena, Reuters/Ipsos, CBS/YouGov, Quinnipiac, Economist/YouGov, Pew, and AP VoteCast for 2024. Approval averages range from the high 30s to about 40 percent, with some likely-voter polls reporting higher figures. Subgroup data, particularly for Hispanic, Black, and 18–29-year-olds, have smaller sample sizes and greater margins of error. The argument relies on consistent patterns across all available crosstabs, not on any single data point.

The coalition was never a conversion

First, consider the actual composition of the 2024 electorate.

Trump’s voters were still mostly white and still mostly without a four-year degree. Pew Research Center put white adults without a four-year degree at 51 percent of his coalition — the smallest share of that group in any of his three campaigns.₂ That is the tell. He did not win because the old base got bigger. He won because the old base held and he added just enough of everyone else.

The additional voters determined the election outcome:

· Pew’s validated-voter analysis found Hispanic voters went from a 25-point Biden blowout in 2020 (61–36) to a near split in 2024: Harris 51, Trump 48.₃ That was the largest partisan move of the cycle.

· Black voters doubled their Republican share, from 8 percent in 2020 to about 15 percent in 2024, per Pew and AP VoteCast.₄ The Council on Foreign Relations later called it the largest Republican share of the Black vote since Nixon.₅

· Young men flipped the youth math. AP VoteCast found Trump took nearly half of voters under 45 and lifted his 18–29 share from 36 percent in 2020 to 46 percent.₆ Young women did not come with them.

· Working-class voters across race became the Republican Party’s center of gravity. VoteCast showed white working-class men remaining the floor — Trump at 68 percent, a 38-point margin — while Hispanic and Black wThis group defines the 2024 coalition—not just the core supporters, but those who enabled the popular-vote victory. This distinction is central to the analysis.n matters, because the argument turns on it.

Their support was transactional, driven by concerns about inflation, border security, and a perception that current leadership viewed them as the problem. Trump promised quick solutions, and they responded accordingly.

When promised results do not materialize, transactional support diminishes. This is the primary reason the coalition has started to erode.

The case for “the coalition is fine”

It is important to consider the opposing perspective, as it is also supported by data.

Rasmussen’s daily tracker on Sept. 14 had 45 percent of likely voters approving and 54 percent disapproving. Harvard-Harris in late August was at 44 percent among registered voters. RMG Research was also in the mid-40s. Those houses are why a Republican operative can say, without lying, that “the averages are cooked, and he’s still in the game.”₈ Republicans still approve in the low-to-mid 80s. Times/Siena has white non-college likely voters at 53 percent approval. The people who were MAGA in 2016 are still MAGA.₉

Border security was also his relatively bright spot for most of the second term — the issue least underwater when the economy and Iran were sinking everything else. If you believe 2024 was mostly a border election, you can argue the new voters got the thing they wanted, and the rest is media weather.₁₀

This represents the strongest argument for coalition stability. However, the data continues to indicate otherwise.

Even the most favorable national polls show him below majority support. A 45 percent approval rating does not indicate a stable coalition; it reflects the upper range of an unpopular presidency. No poll with demographic breakdowns shows the 2024 expansion groups still supporting him. While some polls previously reported higher support among Hispanics and young voters, these figures remain well below the levels needed for a majority. The issue is not a lack of voters, but the loss of the additional support that created a majority.

Border security is no longer sufficient to maintain the coalition. By August, Strength in Numbers/Verasight reported negative ratings for every tracked issue, including border security. Gallup found immigration approval at 35 percent nationally and 21 percent among Hispanics. PRRI reported 35 percent favorable and 61 percent unfavorable on immigration handling. The issue that was expected to retain new voters is no longer effective.

The expansion layer has already peeled off

Focus on the current data from September 2026, rather than the results from November 2024.

The headline range is no longer in dispute.

Reuters/Ipsos, after last week’s RepublicanThere is now broad agreement on the current approval range. from a career-low 33 percent in late August.₁₃ CBS News/YouGov, Sept. 8–11: 39–61.₁₄ Quinnipiac, Sept. 3–6: 33–59.₁₅ Economist/YouGov, updated Sept. 15: 35 approve, 60 disapprove, net minus 25.₁₆ Silver Bulletin as of Sept. 11: net about minus 19.5.₁₇ RealClearPolitics and Ballotpedia: roughly 39–40 percent approve.₈ Different houses, same floor. That is midterm-wipeout territory.

The cleanest current demographic cut is Times/Siena, 1,503 likely voters, Sept. 8–13. Treat the subgroup point estimates as directional. The pattern is the point.₁₈

Young voters have inverted.

Times/Siena, ages 18–29: 18 percent approve, 78 percent disYoung voters have reversed their previous support.t. CBS in June had under-30 approval at 30 percent—already a collapse from 2024 —and it kept falling. Economist/YouGov spent the summer with net approval among 18–29 around minus 45.₁₉ A group that gave Trump roughly half its vote is now in open revolt.

Hispanic voters are no longer part of the 2024 coalition.

Times/Siena likely voters: 26 percent approve, 69 percent disapprove. Economy: 18 percent. The registered-voter cut is about 30. CBS in June: 31 percent. A 48 percent presidential share does not live at 26–31 percent job approval. That is reversion.

The Black share of support has already returned to historical Republican levels.

Times/Siena likely voters: 11 percent approve, 86 percent disapprove. Economy: 17 percent. CBS in June: 18 percent. The protest spike is over.

Independent voters have withdrawn their support.

Times/Siena: independents 27 percent approve. PRRI in June put favorability among independents at 25 percent, down from 37 percent in September 2024.₂₀ Independents put him over. They are not staying.

Those four groups were the 2024 expansion. They are why a popular-vote win existed. They are also the groups now printing 2020 numbers, or worse.

This is what is meant by ‘broken.’ The Republican Party remains, but the additional support that created a majority has disappeared just seven weeks before the midterm. The difference between the base and the broader coalition is therefore significant.

Two Trump-voter numbers, not one

At this point, it is important to present both key data points for a comprehensive analysis.

Times/Siena’s likely-voter sample: 82 percent of 2024 Trump voters still approve of the job he is doing.₁₈

The Economist, using YouGov: one in five of his 2024 voters now disapprove. Only 65 percent of 2024 Trump voters plan to vote Republican this fall, versus 76 percent of Harris voters planning to vote Democratic. Among the defectors, YouGov finds 10 percent of Black Trump voters and 7 percent of Hispanic Trump voters already saying they will vote Democratic, plus 5 percent of white Trump voters under 30.₁

Those figures do not cancel each other. They measure different things.

Siena surveys likely voters in September regarding presidential approval, capturing those who remain engaged. YouGov surveys the broader pool of 2024 Trump voters about their intentions for November, including those who are less engaged. Approval does not equate to turnout, and likely voters do not represent the entire coalition. While core supporters remain loyal, the coalition’s periphery—comprising younger, non-white, and less MAGA-aligned voters—is increasingly disengaged, as reflected in YouGov’s findings.

If only Siena’s data were considered, the coalition would appear stable among likely voters. However, if YouGov’s assessment of midterm intentions is accurate, the 2024 additional voters do not need to switch parties to weaken the coalition; they simply need to abstain at higher rates than Harris voters. This dynamic can result in losing the House, even if the coalition appears strong on paper.

The base is not the coalition

White voters without a college degree still give Trump his best numbers. Times/Siena: white non-college 53 percent. CBS in June: 52 percent. Republicans: low-to-mid 80s. PRRI: 79 percent of Republicans still favorable. Strong MAGA identifiers in Economist/YouGov remain north of 90.₉

This group represents a movement, but it does not constitute a majority.

And even the base is softer than the speeches. Fifty-three percent approval among white non-college voters is not 65 to 68 percent of them on Election Day. By June 13 the New York Times had white working-class voters disapproving of his economic handling in Fox (33 percent approve), CBS (39), NPR/PBS/Marist (40), CNN (43), and Times/Siena (47). Cost-of-living approval among blue-collar whites in Times/Siena was 36 percent.₂₁ The floor is cracked. It has not collapsed. The expansion layer has.

A coalition is built in layers. When the upper layers are lost, what remains is only the core, not the full coalition.

The coalition fractured because the promised outcomes were not delivered.

The data suggests the primary factor was not a reversal of cultural alignment, but rather unmet expectations regarding cost of living.

Economist/YouGov: 72 percent of Americans say tariffs have raised prices; 64 percent disapprove of how he is handling the issue. Among young and minority voters: 74 and 71.₁ Times/Siena cost-of-living approval is in the mid-20s nationally. Economy approval among likely voters is 39 percent overall — 19 percent among 18–29, 18 percent among Hispanics, 17 percent among Black voters.₁₈ That is the coalition-killer. You cannot run a working-class realignment while the working class thinks you made the cart more expensive.

Pew and the Beyond MAGA typology found that the 2024 electorate was more diverse, as many first-time voters participated as a form of protest. According to Beyond MAGA, only 38 percent of Trump voters considered being MAGA important.₂₂ While hardline supporters exist, they were never the majority. When protest voters feel their expectations are unmet, they tend to disengage rather than become activists.

Iran sits on top of that. Many in the new coalition voted for cheaper energy and less entanglement. The Economist found 57 percent of voters think the Iran decision was a mistake, and that the post-war approval drop hit Black, Hispanic, and young white voters first.₁ That is correlation, not a signed affidavit from every switcher. It is also the simplest story the crosstabs allow: prices first, war second, identity last.

Current polling for the midterm indicates a clear trend: Reuters/Ipsos reports Democrats leading Republicans 44 to 37, the largest gap since Trump returned to office.₁₃ Ballotpedia’s September 14 index shows Democrats with a six-point advantage.₈

Why the coalition cannot be rebuilt in time

Coalitions typically fracture in two ways: either the core supporters defect, or the expanded layer withdraws, leaving only a vocal minority.

This is the second one. You do not fix it with a rally in seven weeks.

It is not possible to regain support from Hispanic men with only 18 percent economic approval through rhetoric alone. Similarly, 18-to-29-year-olds with 18 percent job approval and independents with 25 percent favorability are unlikely to be persuaded by campaign promises, such as a $5,000 dividend, which Reuters/Ipsos found was rejected by 63 percent of respondents, including 40 percent of Republicans.₁₃

Restoring the coalition would require a noticeable decrease in the cost of living before the election, which is a matter of policy, not messaging. Public opinion on this issue is already clear and consistent across polls. Young voters, Hispanic voters, Black swing voters, and independents have expressed their dissatisfaction. Even some 2024 Trump voters have indicated their intent through midterm polling, despite still identifying as Trump supporters.

Those who continue to support Trump until November are his core base, not the broader 2024 coalition. They represent what remains of the original coalition.

While the political brand may endure, achieving a majority in this cycle is no longer feasible.

The 2024 Trump coalition was a temporary alliance based on economic concerns. These voters have evaluated the results and are not renewing their support before Election Day. No new group of supporters is likely to emerge in time.

This is not a temporary setback; it marks the end of the coalition that secured victory.

Subscribe now

Notes

₁ The Economist, “Donald Trump’s disintegrating coalition,” Sept. 10, 2026; YouGov polling for The Economist.

₂ Pew Research Center, “Behind Trump’s 2024 Victory: Turnout, Voting Patterns and Demographics,” June 26, 2025; Pew Research Center, “Detailed demographic profiles of Trump and Harris voters in 2024,” June 26, 2025.

₃ Pew Research Center, “How voting patterns changed in the 2024 election,” June 26, 2025.

₄ Pew Research Center, “How voting patterns changed in the 2024 election,” June 26, 2025; AP VoteCast, 2024.

₅ Council on Foreign Relations, “The Trump Coalition,” 2026.

₆ Associated Press, “AP VoteCast: How Donald Trump built a winning 2024 coalition,” Nov. 8, 2024.

₇ Wall Street Journal, “Trump and Republicans Win With Working-Class Voters Across Racial Lines,” citing AP VoteCast, 2024.

₈ Rasmussen Reports Daily Presidential Tracking Poll, Sept. 14, 2026 (45% approve / 54% disapprove among likely voters); Ballotpedia polling index, Sept. 14, 2026 (presidential approval average 40%; generic ballot Democrats +6), including Harvard-Harris (44%, Aug. 28–30) and RMG Research (44%, Aug. 31–Sept. 8).

₉ New York Times/Siena College, Sept. 8–13, 2026; CBS News, June 17–19, 2026; PRRI, June 2026; Economist/YouGov MAGA-supporter crosstabs, 2026.

₁₀ Fox News poll, Jan. 2026: 52% approved of Trump on border security (then his only net-positive issue in that survey).

₁₁ Rasmussen Reports weekly likely-voter crosstabs, June 14–18, 2026 (Hispanic 51%; ages 18–29, 37%).

₁₂ Strength in Numbers/Verasight, Aug. 14–19, 2026 (underwater on all 12 tracked issues; border security net −5); Gallup, June 2–26, 2026 (immigration approval 35% nationally, 21% among Hispanics); PRRI, 2026 (immigration handling 35% favorable / 61% unfavorable).

₁₃ Reuters/Ipsos national poll, published Sept. 14–15, 2026 (1,143 adults; MoE ±3). Prior wave: Aug. 28–31, 2026 (33% approve). Generic ballot: Democrats 44, Republicans 37. On the $5,000 dividend: 63% disapproved.

₁₄ CBS News/YouGov, Sept. 8–11, 2026 (2,460 adults; MoE ±2.3).

₁₅ Quinnipiac University, Sept. 3–6, 2026 (970 registered voters; MoE ±3.9).

₁₆ The Economist/YouGov presidential approval tracker, updated Sept. 15, 2026.

₁₇ Silver Bulletin, Trump approval average, updated Sept. 11, 2026.

₁₈ New York Times/Siena College national poll, Sept. 8–13, 2026 (1,503 likely voters; registered-voter crosstabs published Sept. 15, 2026). Likely-voter job approval: 38% approve, 59% disapprove. Subgroup n’s are smaller: 18–29, n=250; Hispanic, n=170 registered / ~113 on some issue splits; Black, n=193 registered.

₁₉ CBS News poll, June 17–19, 2026 (2,519 adults; MoE ±2.4). Economist/YouGov tracking, June–September 2026, as reported by The Economist and Newsweek.

₂₀ PRRI, “Competing Visions of America at 250,” June 2026. Trump favorability: 34% overall; 25% among independents; 79% among Republicans.

₂₁ New York Times, “Trump Is Losing Ground With White Working-Class Voters on the Economy,” June 13, 2026.

₂₂ Beyond MAGA, “A Profile of the Trump Coalition” (more than 10,000 Trump voters, concluding early 2026).

A country that will not train its own welders, network engineers, and surgical techs will import them, then call the result diversity or a skills shortage depending on the channel. Both explanations are a dodge. The fix is the same bill: My American Workers First Act.

Bo French viewed the University of Texas student section and perceived an invasion.

John Cornyn saw in Bo French a sign of internal conflict within the Republican Party.

Donald Trump saw both situations as an opportunity to gain political leverage without legislating, since doing so would require choosing between Elon Musk’s labor model and American workers. That tension frames the week’s political fight.

This week in Texas centers on a culture war over university identity, a labor market with longstanding issues, and an administration promoting ‘AMERICAN WORKERS FIRST’ while suggesting Americans lack talent. That mix sets up the debate that follows. The throughline is simple: identity politics, labor scarcity, and policy inconsistency.

I published The American Workers First Act in November 2025. While the data has changed, the underlying issues remain. This update keeps the policy focus on American workers and highlights ongoing obstacles. I also discuss pathway to getting it passed.

For the statute, proceed to Part III. For context on the broader debate, begin here. These two paths separate the policy text from the larger dispute. Use both to follow the argument.

Part I — French’s argument is flawed, but the underlying concern is valid. This section starts with the football controversy and then widens to labor. The point is not his method; it is the issue beneath it.

Start with the football controversy, while keeping the broader labor system in mind. The point is to move from the campus dispute to the larger machine. That broader system is the real subject.

On September 13, Bo French, GOP nominee for Texas Railroad Commissioner and former Tarrant County GOP chair, posted an image from the Longhorns student section after Texas defeated Ohio State. He claimed that UT graduation ‘looked like this,’ expressed disbelief, and stated, ‘the problem is now obviously far worse than anyone imagined.’ He further asserted that social media reveals ‘how many Americans have been displaced by foreigners.’

John Cornyn, recently defeated in a primary and focused on his reputation, publicly condemned French’s comments as intolerant and racist, questioning whether other Republicans would ignore such rhetoric. Greg Abbott had already criticized French’s suitability for office. Democrats also criticized French. Responses to Cornyn’s statement included comments such as ‘America is for Americans’ and more extreme views.

It is important to acknowledge that while French’s approach is flawed, the underlying grievance is legitimate.

UT Austin, fall 2025, from the university — not from a Telegram channel:

• 55,000 students

• 80.5% Texas residents

• 10.2% from other U.S. states

• 9.3% international (5,140 on the official books)

• Freshman class of 9,900, and Texas law already requires 90% of those freshmen to be Texas residents.

• International enrollment is a graduate story: 4,873 international graduate students, 1,923 undergraduates.

• Texas Global’s wider net (Option III, English Language Center): 6,877 international students. India first at 1,660. China. A concentration of South Asian students in the section is unsurprising. Collin County and Fort Bend have sent many high-achieving students to Austin through the Top 10% rule. Friends often sit together, and broadcast cameras can distort perceptions. None of this requires F-1 visas, nor does it justify assumptions based on appearance.

Cornyn is correct that one cannot assume students are foreigners based on appearance. However, despite supporting hire-American rhetoric in the Senate, he did not advance corresponding policies. Public condemnation is not a substitute for legislation. Keep the focus on the bill.

The machine is not the football team. It is this:

F-1 student visa (uncapped, university-controlled) → CPT during school → OPT (12 months) → STEM OPT (24 more months) → H-1B lottery (85,000) → employment-based green card (backlogged for Indian nationals for decades).

H-1B gets the hearings. OPT is the quiet on-ramp. OPT has no annual numerical cap. The employer pays a student-adjacent wage. The university keeps the tuition. Congress gets to say the cap is “only 85,000,” while the pipeline keeps expanding. The policy stake is how this pipeline shapes wages, training, and hiring incentives.

Texas is where you can see it without a spreadsheet. 97,043 international students in 2024/25 — third in the country. India: 43.9% of them. The volume campuses are UNT, UT Dallas, UT Arlington, Texas A&M, then Austin. That is DFW and the I-35 corridor. That is the labor market Ken Paxton wants to run for Senate on “the American dream,” without ever describing it. The policy stake is whether that market keeps relying on OPT or turns toward domestic hiring.

Nationally, Indian enrollment in the U.S. fell 6.9% from February 2025 to February 2026 — 378,787 to 352,644 — after F-1 refusals hit 41%. That did not end the fight, because the fight is about a twenty-year stock of workers in specific metros, not this fall’s incoming class. The policy stake is whether lawmakers address that stock or treat it as a temporary flow.

Part II — Update the board. Then name the people sitting on it. Now that the machine is on the table, update the record and name the players. The next step is to identify who is driving the system. Keep the board in view.

Now that the machine is on the table, update the record and name the players. Keep the board in view as the discussion turns to the people behind it. Move from structure to actors.

Two things are true. The lottery-fraud era is receding. The program is still a concentrated foreign-labor channel. USCIS FY2025: 406,348 approvals, 283,772 for India (69.9%), 49,161 for China (12.1%). Computer occupations are still the majority. Most approvals are extensions. Temporary, on paper. A parallel career system, in practice. That is the record before turning to the employers. The policy stake is whether this system sets hiring norms in place of domestic training.

When Tata and Infosys reduce new cap cases, they do not invest in local workforce development. Instead, they extend existing arrangements, offshore work, and anticipate automation to reduce entry-level positions. This is not a skills strategy, but a temporary staffing adjustment. The administration has promised a manufacturing revival but has instead delivered increased automation and reliance on foreign labor. Evaluate both the promises and the outcomes.

The hypocrites, by name

Donald Trump campaigned on this issue, and his Department of Labor promotes the slogan. However, he has told firms that Americans lack talent and that H-1B visas are essential for staffing. The real issue is the absence of a training pipeline, which companies have not developed. Prioritize American workers, not a labor model that undermines them.

Elon Musk stated in December 2024 that he would strongly advocate for H-1B visas. His position supports access to lower-cost, globally sourced talent. While he is entitled to this view, it should not be presented as a patriotic imperative or used to challenge legislative oversight.

Vivek Ramaswamy’s comments suggest a belief that native-born students are culturally disadvantaged. This perspective discourages investment in domestic training and instead favors importing talent under the guise of merit. This is not a genuine skills argument, but rather a dismissive stance.

Lori Chavez-DeRemer’s initiatives, such as Operation Firewall and raising the prevailing wage, are useful tools but do not address the need for a domestic workforce. The Department of Labor can enforce regulations inconsistently. These measures should change incentives, not be presented as comprehensive solutions. The policy stake is whether they build capacity or merely adjust the margins.

Bo French should not be considered for statewide office if his approach relies on racial assumptions. Addressing legitimate H-1B concerns requires legislative action, not public complaints. He should focus on policy or withdraw. The policy stake is whether grievance turns into reform.

John Cornyn’s condemnation was appropriate but inadequate. Despite his seniority and influence, he did not advance hire-American legislation. That inaction is part of the party’s internal challenge.

Chip Roy and Eli Crane. Roy’s American White-Collar Worker Jobs Act — wage-based selection, hire-American attestation, layoff lockout, end of OPT as a shadow work visa — is the restriction half of this paper. Crane’s pause-and-rewrite is the blunt instrument. They are the only two Republicans in this story who wrote something that can be scored. Use them to change the system, and move from critique to action. The policy stake is whether any of these proposals become enforceable constraints.

Ken Paxton. He is in a competitive Senate race against James Talarico. While he references the American dream, Texas’s technology sector relies heavily on OPT. To appeal to working-class voters, he should support hire-American policies and GI Bill 2.0 rather than partisan rhetoric. Say the quiet part out loud about H-1B composition. Saying it is not passing it. Florida does not get to lecture Texas while the same pipeline feeds both states’ hospital systems and IT benches.

Part III — The American Workers First Act. My November 2025 proposal remains central. This week’s update adds a campus-focused provision and aligns the bill with existing legislative vehicles. Stay clear on the policy objective:

Title I — Domestic capacity

Expand registered apprenticeships with a national target of one million active positions. Priority areas include software, cybersecurity, semiconductor technology, industrial maintenance, electricians, HVAC-R, welding, allied health, and transmission.

If the only path to becoming a mid-level engineer requires a four-year degree, unpaid internship, and referral, the system will inevitably import talent to fill gaps. Paid 24-month apprenticeship tracks with employer accountability address this issue.

Offer short-term, stackable vocational programs at community colleges, ranging from three to twelve months, designed for individuals with existing work or family commitments. Tie funding to successful job placement.

Trades capital: booths and instructors, not another PowerPoint from the Department of Labor.

Require clear identification of labor shortages before awarding grants. Publish shortage data and reclaim funds from unsupported programs.

Title II — GI Bill 2.0

The original approach worked because it was generous, time-limited, and tied to institutions that had to perform. Build the civilian version.

• Who: unemployed for four weeks or more due to layoff, automation, or trade. Veterans first.

• What: $30,000 training, $1,500–$2,500 stipend, health bridge, placement.

• How long: 36 months, stackable. 30-day clock. Not a two-year WIOA odyssey.

• Scale: one million beneficiaries, feeding five million placements.

Both progressives and populists can support this initiative. Retain this provision to facilitate passage of the broader bill. The policy stake is whether this common ground helps move the larger package.

Title III — Hire American, with audits

Adopt the approaches proposed by Roy and Crane rather than creating less effective alternatives.

1. Wage-weighted H-1B selection. Highest real wage first.

2. 30-day American recruitment on a public board at the offered wage. DOL random audit. Lie, and you are debarred for three years.

3. Layoff lockout. Occupational layoffs in an SOC code in the last 12 months mean no new H-1B or OPT in that code. The Disney-in-Florida file: Ensure that prevailing wage standards are enforced, eliminating the use of Level 1 wages as a loophole. Provide fee exemptions for small firms, recognizing that a 20-person business in Austin differs significantly from large multinational corporations. Infosys.

Title IV — The campus pipeline the original paper underweighted

This provision is particularly relevant to Texas and will likely face opposition from universities, underscoring its significance.

• OPT wage floor equal to the H-1B floor for the same SOC and metro.

• Public OPT employer file: headcount, SOC, wage, layoff history.

• Cap the duration of STEM OPT unless the position pays at or above the 75th percentile for that occupation in the relevant metropolitan area. Recognize extended training periods as employment.

• Any graduate CIP code where F-1 For any graduate program where F-1 students represent a significant proportion, publish data on domestic applications, offers, F-1 student share, and OPT placements. Prioritize transparency, with caps considered if issues persist. For programs that function as work-visa mills. Hit the money.

Title IV does not include measures based on visual assessments or racial profiling at events, nor does it address social dynamics unrelated to policy.

Title V — Pay for it without a tax hike

$10–15 billion a year. The annual cost is $10–15 billion, to be covered by budget offsets rather than speculative funding. WIOA, Job Corps, and YouthBuild cannot show placement. The Adult formula alone is still in the high-$800 millions. Stop paying to warehouse people.

• Direct H-1B and OPT employer fees into a Domestic Skills Investment Fund. Use any surcharges to train American workers, ensuring transparency and accountability.

• Restrict the tax deductibility of offshore replacement labor in the occupational codes targeted by the Domestic Skills Investment Fund.

• Require employers to provide a 30–50% match for apprenticeship positions.

• Pay-for-success against UI savings, with clawbacks attached.

Studies show apprenticeships yield approximately $1.47 in returns for every dollar invested, with payback periods of five to seven years. This program must remain deficit-neutral to succeed in reconciliation.

The goal is to place 5 million additional Americans in shortage occupations by 2036 and reduce structural H-1B dependence by 20–30%. The objective is not to eliminate foreign workers or open caps, but to ensure the country can meet its own workforce needs.

Part IV — How this passes, and who has to move

A mid-2026 signing is no longer feasible. Adjust the legislative strategy to the current timeline.

• Attach Title III to Chip Roy’s existing bill, which has already been introduced.

• Include Title IV as an amendment, taking advantage of recent developments that have made its passage more attainable.

• Present Titles I and II in the Education & Workforce and HELP committees as companion skills legislation. Ensure the bill is scored and included in the next reconciliation window.

• Pilot the program in 15 states during the first year, including Texas. Establish program sites in Austin, Dallas, Houston, and relevant industrial locations, focusing on implementation rather than publicity.

House Republicans can enact restrictions this winter, but maintaining their coalition requires more than reducing the lottery without investing in workforce development. Senate Democrats may continue to cite concerns about racism if substantive alternatives are not presented. Implementing GI Bill 2.0 addresses these issues directly.

Trump can sign this tomorrow and give real meaning to the administration’s stated priorities. Alternatively, he can keep prioritizing corporate interests over American workers. His actions will demonstrate his true priorities. Place to run immigration policy. It is a perfect place to start a fight that produces no welders, no network engineers, and no surgical techs.

Bo French started the controversy; John Cornyn responded with moral arguments; Donald Trump seeks to leverage it politically; Elon Musk supports policies that lower labor costs; Vivek Ramaswamy frames it as a cultural issue; and Ken Paxton calls it the American dream. Chip Roy and Eli Crane have at least contributed substantive proposals.

I wrote the other half.

The proposal aims to create 5 million jobs, implement hire-American rules with audits, introduce GI Bill 2.0, and reform OPT and H-1B programs, all without increasing income taxes.

The stands will look like America. The payroll should too. Call the fucking vote and let’s FINALLY and ACTUALLY put the American Worker first.

— Maine

You cannot lecture people back into a coalition you abandoned — and a party that talks like it has never paid a grocery bill does not get to call itself the party of work.

The New Democratic Party has positioned itself as a broad-based coalition dedicated to advancing the interests of working people across diverse backgrounds. Central to its renewed strategy is the intentional effort to rebuild alliances with key demographic groups, such as Hispanic voters who migrated toward Trump in 2024 but are now shifting back, illustrating the party’s broader commitment to regaining trust among historically foundational constituencies.

Over the past year, I have maintained that, following 2024, many Democrats shifted their focus to image at the expense of substantive policy. I have consistently advocated for redirecting attention toward class issues. As a result, the party lost its foundation—Midwestern states, unions, and blue-collar voters. To regain their trust, Democrats must demonstrate humility and honesty rather than rely on media appearances or advice from those without firsthand experience.[1] Coalition building demands genuine engagement with previously neglected groups.

In April, I referenced Bill Clinton’s 1991 Cleveland speech, highlighting the need for opportunity, responsibility, and community as a ‘New Choice’ for working families, not simply another ideological dispute. The political context in 2026 remains similar: widespread economic anxiety persists, and working-class voters are open to substantive outreach. Priorities for reconnecting with these constituencies include rewarding work, expanding apprenticeships, pursuing fair trade with labor standards, and addressing regional economic security concerns.[2]

This defines the Blue Collar Blue Tent: a coalition, not a purity test. A party that treats the cost of living as a branding issue will keep losing those most affected.

The candidates are diverse: a Presbyterian seminarian in Texas, a Methodist pastor in Kansas, a fifth-generation logger from Allagash, a firefighter in the Lehigh Valley, an ironworker near Cleveland, a smokejumper in Montana, an auto-shop owner in Washington, a Marine drill instructor from Arizona, a lieutenant governor raised on SNAP and Section 8 in Minnesota, and a labor organizer from New Jersey. They represent varied backgrounds, faiths, and perspectives, yet share a commitment to working people. If this message reiterates my earlier assertion that a working-class campaign is reclaiming the party from elites, that is deliberate.[3]

Sending lawyers to speak with plumbers and pipefitters, then expressing surprise when the message fails, is ineffective.

The working class should not be depicted solely as men in workwear. Nearly half of working-class voters are women, who are often most affected by challenges related to rent, childcare, and healthcare.

Any strategy that excludes women is not a working-class strategy; it is merely performative.

Faith, forests, factory floors, and the women already leading this movement

James Talarico in Texas is a former teacher and Presbyterian seminarian. He does not run as “the resistance.” He runs as top versus bottom. Taxes, costs, childcare, the PRO Act, a tariff rebate, make corporations and the rich pay. He has Republicans endorsing him. That is not an accident. It is a coalition play in a state Democrats were told was unwinnable.[4][5]

Adam Hamilton in Kansas is the pastor of one of the largest United Methodist churches in the country. He is running from the center on purpose: love your neighbor, both extremes miss it, cut the tariffs that hammer farm country, restore healthcare credits, ban stock trading in Congress. His ad against socialism and against Marshall’s price-and-cuts record is the radical-centrist version of the same tent.[6][7]

You can love your neighbor and still hate the grocery bill. That’s not moderation. That’s adulthood.

Troy Jackson is the Maine chapter of this argument. Fifth-generation logger. Union guy. Former Senate president. I said after the Platner collapse that the nominee was Jackson — and that it probably should have been Jackson the whole time. Crowd sizes started looking familiar. Collins asked for four televised debates; he said yes—the latest public polls I posted this week: Jackson 48, Collins 44 in YouGov. CNN/SSRS had it at 48–45.[8][9][10][11]

Maine doesn’t need another consultant who discovered “working people” on a Zoom. It needs the guy who actually cut the wood.

Then there are the women who already fit the narrative — and who the coverage still treats as a side category. Marie Gluesenkamp Perez (WA-03) co-owns an auto repair shop with her husband. Her father immigrated from Mexico. Her mother’s family is loggers and carpenters. She built a house by hand because the budget allowed it and brought a kid to work because daycare cost too much. She wins a Trump district by talking supply chains, headlights, working waterfronts, and the fact that the middle class is an endangered species. That is the Blue Dog, wrench-in-the-toolbox version of the tent. She is on the ballot again in November.[35][36]

JoAnna Mendoza is the Democratic nominee in Arizona’s 6th. Raised in a Pinal County farmworker family. Enlisted at 17. Twenty years in the Navy and Marines, including as a drill instructor. Single mom. She is running on costs, price-gouging, Social Security, and rural hospitals against Juan Ciscomani in a toss-up; Trump won by a point. Farmworker kid to drill instructor is not a consultant’s “working-class messenger.” That is the biography.[37][38]

Peggy Flanagan, the DFL Senate nominee in Minnesota, grew up with a single mother on food stamps, Medicaid, childcare assistance, and Section 8. Different-colored lunch ticket at school. White Earth Nation. She talks about paid family leave, universal school meals, and “the many versus the money.” She beat a better-funded moderate in the primary.[39][40]

Analilia Mejia won Mikie Sherrill’s New Jersey seat in a special and is running for the full term. Daughter of a Dominican machinist and a Colombian seamstress. Labor organizer. Helped win New Jersey’s $15 wage and paid sick leave. First Afro-Latina in Congress. That is the Hispanic working-class lane in one person: shop-floor parents, union fights, kitchen-table wages.[41][42]Lived it, beat it, laminated it—every time.

Add the labor incumbents who are already acting as infrastructure for this cohort. Nikki Budzinski in Illinois and Val Hoyle in Oregon — both steeped in the labor movement — served as honorary chairs of the Blue Collar Brigade. Elissa Slotkin in Michigan keeps saying the shrinking middle class is a national-security problem and that tariffs tax people who buy cars and hockey sticks. Different tone than Jackson. Same fight over who the economy is for.[12][43]

The House challenger class still includes firefighter Bob Brooks in PA-07, ironworker Brian Poindexter in OH-07, smokejumper Sam Forstag in MT-01, and Randy Villegas in CA-22 — the educator who grew up in the family auto shop and beat the DCCC’s preferred moderate. Minnesota’s Kaela Berg, the flight attendant and union leader who ran her campaign between flights, lost her primary. She still proved the type exists.

The problem is not a shortage of working-class women. The problem is treating them as extras in a story about guys with wrenches.[12][13][34]On the trail, the platform keeps repeating itself: groceries, rent, premiums, childcare, power bills. Who pays. The PRO Act. Paid leave. Ban members of Congress from trading stocks. Build housing. Restore healthcare credits. Kitchen-table populism, not a sermon.[14]

If your first sentence is a culture war and your third sentence is rent, you already lost the room.

A broad coalition necessarily entails disagreement on certain issues.

This is the part the ideologically trapped cannot stand, on either end. Some Democrats — especially in very blue districts — will lean into culture issues because that is what their constituency wants. Safe seats in New York, parts of California, deep-blue corners of the Twin Cities and Denver are not Maine, Kansas, southern Arizona, or southwest Washington. Voters there will nominate people who talk about ICE, Gaza, gender policy, and “the groups” in a register that would lose a logger town or a farmworker precinct by 20 points.

That is representation, not a betrayal. You do not run Brooklyn like you run Allagash.[44]At the same time, moderates will distance themselves.

Hamilton’s “both extremes” ad is not an accident. Gluesenkamp Perez voting with Golden to condemn socialism is not an accident. Talarico drawing Republican endorsers in Texas is not an accident. They are answering their voters, who want costs down and do not want to be drafted into someone else’s seminar. If you demand a swing-district Democrat eat the safest-seat menu, you are not building a majority. You are running a book club with a PAC. That is what a big tent is—not being ideologically trapped and not requiring a loyalty oath on every issue. Understanding we will not agree on every fight — abortion framing, immigration enforcement, Israel, what “woke” even means in a union hall — and refusing to pretend that disagreement is heresy.

Disagreement is not heresy. Poverty is not a vibe. Upward mobility is the point. The foundation still has to be one thing: putting working people’s upward mobility first. Wages that rise. Housing you can buy. Healthcare that does not bankrupt a family. Childcare that does not eject a parent from the labor force. A union you can join without getting fired. A tax code that does not treat a firefighter worse than a billionaire’s accountant. Apprenticeships and trade school treated like real paths, not consolation prizes. That is the load-bearing wall. Culture fights can sit on different floors of the same building. They cannot replace the foundation.

You can fight about the wallpaper. You cannot fight about whether the house has a floor.

I have been saying this since last fall: the opening is class messaging. Bring unions, blue-collar, and working-class people back. The people who cannot hear that — because they are too busy policing tone, pronouns, or purity — are the problem I called “too arrogant and aloof. Too much time on TV and not with voters.”[33]Cable news is not a precinct. Twitter is not a union hall. And your donor Slack is not America.

A party that can hold Mejia and Gluesenkamp Perez, Flanagan and Hamilton, Jackson and a Brooklyn socialist, without making any of them the official mascot, can govern a country this big.

The coalition extends beyond men in Carhartt bibs.

The blue tent is a broad coalition. It includes Black voters who never left. It includes union households that did. It includes working-class women who carry the grocery bill and the childcare bill. And it has to include Hispanic voters — the fastest-growing slice of the electorate, and the group that moved hardest toward Trump in 2024. Pew’s validated-voter work showed the Democratic edge among Latinos collapsing from about 25 points in 2020 to a few points in 2024. Trump’s share among Hispanics jumped into the high 40s. Latino men split or flipped. That was real. Anyone who pretended it was “just turnout” was coping.[15]

If your theory of 2024 requires half the electorate to have been imaginary, your theory is the problem.

They are marching back. Not all the way to 2018. Back enough to matter in a three-seat House map.Hart Research for TelevisaUnivision surveyed 1,500 Latino registered voters across 17 battleground House districts in August. Democrats lead the generic ballot 58–35. That is a 23-point edge — more than double Harris’s 11-point edge with the same sample’s recalled 2024 vote. In three Texas districts that were 44–44 in 2024, Latinos now break 56–36 Democratic. Florida swing districts, Latinos went from Trump +1 to Democrat +13. Spanish-dominant news consumers flipped from a one-point Trump lean to a 14-point Democratic lean.

About 11 percent of self-identified 2024 Trump Latinos now say they will vote Democratic for Congress. Trump’s job rating in that battleground sample: 36 approve, 63 disapprove. Cost of living: 65 percent disapprove. Immigration enforcement: 62 percent disapprove. And on the issue that helped Trump with Latinos two years ago — prices — they trust Democrats 62–36.[16][17]

A Cook/Univision in Texas is even louder at the candidate level. Among Hispanic likely voters, Talarico leads Paxton 63–30. Hinojosa leads Abbott 60–33—Trump approval among Texas Hispanics: 33 percent. Economy, not immigration, is the top issue by two-to-one. Nearly two-thirds say they are less likely to back a candidate who supports the deportation policy as it’s being run. One in three Latino Trump voters in a separate Univision/YouGov Texas poll now express doubt or regret.[18][19]

Equis has been tracking the same curve all year: Trump’s ratings on the economy and cost of living among Latinos collapsed (cost of living net -52 in their summer memo). Potential defection among Latino Trump voters rose. Democrats hold a double-digit generic lead nationally, among Latinos, and in competitive districts. Young Latino men — the 2024 swing prize — are deeply underwater on prices. UnidosUS found a quarter of Latino Trump voters saying they would not vote for him again. Equis’s read after 2025 off-years: the comparison point for 2026 may be 2022, not 2024. Better for Democrats than the GOP peak. Not yet the old supermajority.[20][21][22]They rented Trump on prices. The bill came due. Leases expire.

That is the tent. A logger in Aroostook, a shop owner in Skamania County, a drill instructor from a farmworker family, a seamstress’s daughter in North Jersey, and a teacher-seminarian in Austin are not running five different parties. They are running at the same voters: people who work, people whose rent and groceries moved, people who tried Trump on prices and ICE and did not get the deal they were sold. Hispanic working-class voters are not a side quest. They are the working class. Act like it.

The polls say the issues are Democratic. The brand is still the fight.

National issue polls match the tent’s script.
Reuters/Ipsos: Democrats preferred on cost of living 36–28, the widest gap since they started asking last fall. Same series: Democrats preferred on the economy for the first time in nearly a decade.

Economist/YouGov: Democrats +7 on handling the economy. Focaldata: inflation and economic concern worth about 2.3 points toward Democrats since 2024.

BGSU likely voters: Democrats +22 on affordable healthcare, +18 on Social Security, +15 on housing, +12 on cost of living.

KFF: Democrats +13 on healthcare costs, bigger on Medicaid, ACA, Medicare. Generic ballot aggregators through this week: Democrats generally +4 to +7.

I posted the Senate board Thursday: Talarico 48–44, Jackson 48–44, Brown up in Ohio. That is what a cost-of-living midterm looks like when the White House owns the prices.[23][24][25][26][27][9]In August I wrote that people were finally seeing what I said at the start of the term: Trump would cook himself and the GOP; Democrats needed to force them to govern their mob and then watch; midterm summer would arrive with Trump in the 30s, independents bolting, Democratic voters motivated, Republicans demoralized. That is the tape.[28]

Don’t snatch a midterm from the jaws of a cost-of-living collapse because you wanted to win an argument on cable.

Two honest caveats.

One: issue advantage is not class realignment. CBS still finds both parties underwater on whether their policies help with costs. Democrats just look less like the ones lighting the bill on fire. White non-college voters have moved off Trump’s economy and have not all moved onto a Democratic ballot. Pew finds 60 percent of Americans now say “working class” describes them — and Republicans still claim the label more than Democrats. Immigration and crime remain GOP strengths.[29][30][31]Being less hated on groceries is not the same thing as being trusted with people’s lives. Don’t confuse a bounce with a conversion.

Two: if the foundation is upward mobility, the culture split becomes manageable. If the foundation becomes a national culture war with a Democratic logo on it, Hispanic men, union households, and working-class women will hear the war and not the wage. The midterm is sitting on the kitchen table. Do not knock it off to prove you’re the smartest person in the group chat.

Objectives for this coalition

I want Democrats who can stand in a union hall in Lewiston, a parish hall in Wichita, a shop in Bakersfield, a volunteer firehouse in Bethlehem, a VA clinic in Tucson, and a school cafeteria in St. Louis Park and say the same true thing: you work, the bill went up, the people at the top wrote the rules, we will change the rules. They do not have to say it the same way. They do not have to vote the same way on every amendment. Brooklyn can be Brooklyn. Rural Washington can be rural Washington. The test is whether working people’s shot at moving up is the first sentence, not the footnote.

First sentence: wages, rent, care, work. Everything else is chapter two. That is a bigger tent than “the groups.” It is also a more American one.

Black voters, Hispanic voters moving off Trump, union households, rural Mainers, working-class women, evangelical-adjacent Methodists who still believe Micah 6:8 is a policy document — that is a majority coalition if you do not blow it with arrogance or with a demand that every district recite the same script. Troy Jackson still has to beat Susan Collins. Talarico still has to beat Ken Paxton. Hamilton still has to clear a century of Kansas math. Mendoza still has to flip AZ-06. Flanagan still has to close in Minnesota. Brooks, Poindexter, Forstag, Villegas, Gluesenkamp Perez, and Mejia still have to finish the job. The polling on the issues says voters already prefer the Democratic answer on costs, healthcare, housing, and retirement. The polling on Latinos says the 2024 realignment was a lease, not a deed.

The job between now and November is simple enough that I have been repeating it since last fall: organize working-class Americans. Show up. Talk class. Don’t take the whale money and then pretend you work for the crew. Don’t get ideologically trapped when the opening is sitting on the kitchen table. Don’t build a “blue-collar” brand that forgets half the people who clock in are women. And don’t confuse a big tent with a single hymnbook.[32][33][34]

Show up where the shift change happens. Not where the fundraiser happens.

The New Democratic Party, if it exists, will not be one face. It will be a blue tent full of people who clock in — and who are allowed to disagree about everything except whether those people get to get ahead.

— Maine

@TheMaineWonk

Footnotes

1. @TheMaineWonk

2. “This type of delusion is why Democrats lose… We’ve lost the Midwest, unions, blue-collar and working people,” X, October 28, 2025, /TheMaineWonk/status/1983148598607962430.

3. @TheMaineWonk

4. “1991: Clinton’s Cleveland speech demanded a Democratic ‘New Choice’… 2026: Trump holds the White House and affordability anxiety dominates,” X, April 10, 2026, /TheMaineWonk/status/2042733579264798948.

5. @TheMaineWonk

6. “This is the New Democratic Party. Going back to its roots. Real, blue-collar, working-class people…,” X, August 23, 2025, /TheMaineWonk/status/1959290598361268280.

7. Jordan King, “James Talarico Reveals List of 20+ ‘Republican Leaders’ Endorsing Him,” Newsweek, September 10, 2026, https://www.newsweek.com/james-talarico-reveals-list-of-20-republican-leaders-endorsing-him-12425657.

8. “James Talarico,” Ballotpedia, campaign themes and 2026 Senate race, https://ballotpedia.org/James_Talarico.

9. Michael Gryboski, “Adam Hamilton releases ad opposing socialism and open borders,” The Christian Post, September 10, 2026, https://www.christianpost.com/news/adam-hamilton-releases-ad-opposing-socialism-and-open-borders.html.

10. “Adam Hamilton (Kansas),” Ballotpedia, affordability agenda and campaign platform, https://ballotpedia.org/Adam_Hamilton_(Kansas).

11. @TheMaineWonk

12. , “Shorter Maine Senate: The Dem nominee is

13. @TroyJackson207

14. . And it probably should have always been Troy Jackson,” X, July 16, 2026, /TheMaineWonk/status/2077904645544948133.

15. @TheMaineWonk

16. “Fresh Senate Polls Today: TEXAS Talarico 48 / Paxton 44… MAINE Jackson 48 / Collins 44,” X, September 10, 2026, /TheMaineWonk/status/2098155469772825016.

17. “Maine and Michigan Senate races are neck and neck, CNN polls find,” CNN, September 9, 2026, https://www.cnn.com/2026/09/09/politics/cnn-polls-maine-senate-michigan-senate-abdul-el-sayed-susan-collins.

18. “2026 Maine Senate Polls,” Pollsmax average and YouGov, September 2026, https://www.pollsmax.com/senate/maine/.

19. Lisa Kashinsky, “Dem candidates form ‘Blue Collar Brigade’ to boost fundraising,” Politico, June 30, 2026, https://www.politico.com/news/2026/06/30/union-democrats-house-fundraising-00980980.

20. Blake Jones, “The insurgent Democrat taking on David Valadao,” Politico, September 8, 2026, https://www.politico.com/news/2026/09/08/david-valadao-challenger-randy-villegas-01066275.

21. Ned Rust, “Swing Voters Want Kitchen-Table Populism, Not a Sermon,” Jacobin, September 11, 2026, https://jacobin.com/2026/09/survey-voters-populism-affordability-messaging.

22. “Latino Voters Are 2026’s Biggest Swing Bloc — And Neither Party Has Won Them Over,” Latin Times, August 14, 2026 (citing Pew validated-voter analysis of 2020–2024 Latino shift), https://www.latintimes.com/latino-voters-are-2026s-biggest-swing-bloc-neither-party-has-won-them-over-598565.

23. “Exclusive poll: GOP’s gains among Latinos fade in House battlegrounds,” Axios, September 8, 2026, https://www.axios.com/2026/09/08/poll-latino-voters-trump-gains-fade.

24. “Latino Voters in 17 Swing Districts Are Swinging Hard Away From Trump — POLL,” Latin Times, September 10, 2026, https://www.latintimes.com/latino-voters-17-swing-districts-are-swinging-hard-away-trump-poll-599130.

25. “Cook Political Report and TelevisaUnivision Texas poll of Hispanic Voters,” Cook Political Report, September 9, 2026, https://www.cookpolitical.com/survey-research/2026-texas-hispanic-voters-univision-poll.

26. “This Is What Texas Voters Think Looking Ahead to 2026,” N+ Univision / YouGov, September 9, 2026, https://www.univision.com/noticias/politica/poll-univision-texas-2026.

27. Equis Research, “Memo: Summer 2026 Economy Poll,” July 6, 2026, https://www.weareequis.us/research/memo-summer-2026-economy-poll.

28. Equis Research, “Latino State of Play: 2025 Elections and New Equis Polling,” https://www.weareequis.us/research/latino-state-of-play-2025-elections.

29. “Many Latino voters have turned away from Trump, but Democrats aren’t necessarily winning them over, new poll finds,” CBS News, May 27, 2026 (UnidosUS / BSP / Shaw survey), https://www.cbsnews.com/news/latino-voters-poll-trump-democrats-midterms-immigration/.

30. “Polling shows Democrats winning on cost of living ahead of midterms,” The Hill, August 26, 2026 (Reuters/Ipsos), https://thehill.com/homenews/campaign/6051227-democrats-favored-cost-of-living-affordability-concerns-midterms-survey/.

31. Jason Lange, “Voters Prefer Democrats Over Republicans on Economy for First Time in a Decade,” Reuters via U.S. News, August 3, 2026, https://www.usnews.com/news/world/articles/2026-08-03/democrats-lead-republicans-on-economy-as-trump-approval-falls-reuters-ipsos-poll-finds.

32. “Cost of living fuelling Democrat strength in new midterms polling,” Focaldata, September 11, 2026, https://www.focaldata.com/blog/cost-of-living-fuelling-democrat-strength-in-new-midterms-polling.

33. BGSU Poll, Democracy and Public Policy Research Network, September 2026, https://www.bgsu.edu/arts-and-sciences/democracy-and-public-policy-research-network/bgsu-poll.html.

34. “KFF Health Tracking Poll: Health Care Costs, Expiring ACA Tax Credits, and the 2026 Midterms,” KFF, January 29, 2026, https://www.kff.org/public-opinion/kff-health-tracking-poll-health-care-costs-expiring-aca-tax-credits-and-the-2026-midterms/.

35. @TheMaineWonk

36. “They are realizing what I warned you about at the start of admin: Trump is inevitably going to cook himself & GOP… Indies bolting & Dem voters motivated,” X, August 5, 2026, /TheMaineWonk/status/2085007649540559204.

37. “CBS News poll analysis on cost of living and whether Americans see answers from political parties,” CBS News, July 30, 2026, https://www.cbsnews.com/news/cbs-poll-analysis-cost-of-living-democrats-republicans/.

38. “Which Americans see themselves as working class?” Pew Research Center, August 31, 2026, https://www.pewresearch.org/politics/2026/08/31/who-sees-themselves-as-working-class/.

39. “Trump Is Losing Ground With White Working-Class Voters on the Economy,” The New York Times, June 13, 2026, https://www.nytimes.com/2026/06/13/us/politics/trump-white-working-class-voters-economy.html.

40. @TheMaineWonk

41. , “The time to organize working-class Americans and blue-collar people is now,” X, October 13, 2025, /TheMaineWonk/status/1977843612529328239.

42. @TheMaineWonk

43. “This is the problem with the ideologically trapped. Can’t even realize it’s the best opportunity to focus on class messaging… Too arrogant & aloof. Too much time on TV & not with voters,” X, May 5, 2026, /TheMaineWonk/status/2051458248323084586.

44. “Everyone wants to win working-class voters. But have they realized many are women?” Salon / The 19th, September 9, 2026, https://www.salon.com/2026/09/09/everyone-wants-to-win-working-class-voters-but-have-they-realized-many-are-women-partner/; see also “2026 could be breakout year for blue-collar Democratic women,” Roll Call, October 8, 2025, https://rollcall.com/2025/10/08/working-class-women-democrats-midterm-elections/.

45. “Marie Gluesenkamp Perez,” Ballotpedia, https://ballotpedia.org/Marie_Gluesenkamp_Perez; “Marie Gluesenkamp Perez,” Wikipedia, https://en.wikipedia.org/wiki/Marie_Gluesenkamp_Perez.

46. “This Rural Congresswoman Thinks Democrats Have Lost Their Minds. She Has a Point.” The New York Times, January 12, 2026, https://www.nytimes.com/2026/01/12/opinion/marie-gluesenkamp-perez.html; “2 House Democrats vote to salvage GOP agenda,” Politico, September 1, 2026, https://www.politico.com/news/2026/09/01/two-house-democrats-vote-to-salvage-gop-agenda-01060168.

47. “NEW POLITICS ENDORSED CANDIDATE JOANNA MENDOZA WINS DEMOCRATIC NOMINATION IN ARIZONA’S 6TH,” New Politics, July 21, 2026, https://www.newpolitics.org/news/new-politics-endorsed-candidate-joanna-mendoza-wins-democratic-nomination-in-arizonas-6th-congressional-district.

48. David Iversen, “Juan Ciscomani and JoAnna Mendoza bring dueling pitches to Arizona 6th District debate,” KTAR, September 10, 2026, https://ktar.com/arizona-politics-news/election/juan-ciscomani-joanna-mendoza-cd6-debate.

49. “What to know about Peggy Flanagan, Minnesota’s DFL nominee for U.S. Senate,” Star Tribune, August 12, 2026, https://www.startribune.com/what-to-know-about-peggy-flanagan-minnesotas-dfl-nominee-for-us-senate/601877977.

50. “Peggy Flanagan Wins DFL Primary for U.S. Senate,” Minnesota DFL, August 11, 2026, https://dfl.org/peggy-flanagan-wins-dfl-primary-for-u-s-senate/; “2026 United States Senate election in Minnesota,” Wikipedia, https://en.wikipedia.org/wiki/2026_United_States_Senate_election_in_Minnesota.

51. “Analilia Mejia,” Wikipedia, https://en.wikipedia.org/wiki/Analilia_Mejia.

52. “Congresswoman Analilia Mejia, the first Afro-Latina to Serve in Congress, Sworn In to Congressional Black Caucus,” Insider NJ, September 1, 2026, https://www.insidernj.com/press-release/congresswoman-analilia-mejia-the-first-afro-latina-to-serve-in-congress-sworn-in-to-congressional-black-caucus/; Alex Nussbaum and William Westhoven, “Who is Analilia Mejia?” NorthJersey.com, February 6, 2026, https://www.northjersey.com/story/news/politics/elections/2026/02/06/who-is-analilia-mejia-nj-primary-for-district-11-congress/88541237007/.

53. Adam Wren, “This Senator Has a Very Midwestern Plan to Save the Democrats,” Politico Magazine, August 2, 2026, https://www.politico.com/news/magazine/2026/08/02/elissa-slotkin-michigan-primary-democrats-midwest-01020467; Sen. Elissa Slotkin, “Slotkin Supports Bill to End Trump’s Tariffs…,” September 3, 2026, https://www.slotkin.senate.gov/2026/09/03/slotkin-supports-bill-to-end-trumps-tariffs-that-cost-michiganders-more-than-any-other-state/.

54. Ruy Teixeira, “Instead of a working-class reckoning, Democrats got the DSA,” USA Today, September 12, 2026, https://www.usatoday.com/story/opinion/2026/09/12/democrats-dsa-midterm-elections-socialism-capitalism/91676903007/; “The DSA Class of 2026: 11 Democrats Running as Socialists,” IVN, June 29, 2026, https://ivn.us/the-dsa-class-of-2026-11-democrats-running-as-socialists/.

For three decades, Susan Collins has demonstrated Maine’s tendency to split tickets. She won by 37 points in 2014, 23 in 2008, 17 in 2002, and still by 8.6 in 2020 against Sara Gideon, in the state’s most expensive Senate race, while Maine voted for Joe Biden by 9.¹ ²

However, two recent national polls indicate that this pattern is changing.

CNN/SSRS, August 31–September 6: Jackson 48, Collins 45 among 880 likely voters (±3.7).³

YouGov, September 2–8: Jackson 48, Collins 44 among 1,335 likely voters; Jackson 46, Collins 44 among 1,378 registered voters (±4.9).⁴

Neither poll shows a decisive lead, as both results fall within the margin of error. Taken together, however, they highlight a significant shift. Despite a contentious Democratic nomination process and a late replacement candidate, Troy Jackson remains ahead, and the coalition that previously supported Collins no longer does.

Maine is not a politically balanced, 50–50 state.

Registration still leans Democratic: about 34% Democrat, 30% Republican, 32% unenrolled.⁵ Harris carried Maine 52.4–45.5 in 2024.⁶ Collins is the only Republican senator from a state Donald Trump has never won.⁷

The midterm electorate reflects this political landscape. CNN reports Hannah Pingree leading Bobby Charles 52–39. YouGov shows likely voters favoring Pingree 47, Charles 36, and independent Rick Bennett 2.⁸ These results do not suggest that only Collins is performing well while the rest of the ticket struggles.

Trump in Maine:

• CNN/SSRS: 31% approve, 67% disapprove

• YouGov: 39% approve, 60% disapprove⁹

Despite different polling methods, the challenge remains the same. A Republican incumbent cannot rely on a president with 31% or 39% approval in a state that recently supported Harris by seven points.

Swing voters have shifted.

Maine elections are decided by unenrolled voters and by women. In 2020, exit polls showed Collins winning women 49–46 and crushing Gideon among men 55–36. Women were 59% of that electorate. Collins survived because the state’s largest group still gave her a plurality.¹⁰

CNN/SSRS now:

• Women: Jackson 54, Collins 39

• Independents: Jackson 46, Collins 42

• Men: Collins 53, Jackson 41¹¹

Jackson leads women by 15 points. Independents, who comprise about one-third of voters and often determine outcomes, also support him.

CNN’s Trump-message question is the tell:

• All likely voters: oppose Trump 46%, support Trump 17%

• Women: oppose 53%, support 15%

• Independents: oppose 44%, support 11%

• Jackson voters: oppose 85%¹²

Collins previously positioned herself as a Republican distinct from Trump. Currently, 37 percent of her voters view their Senate vote as a message supporting Trump, while 8 percent see it as a message against him.¹³ She is winning among Trump supporters in Maine, but that group is not large enough to secure victory.

Maine's key demographic groups are not closely divided.

CNN:

• Under 45: Jackson 58, Collins 33

• College graduates: Jackson 61, Collins 35

• White college graduates: Jackson 61, Collins 36

• Ages 35–49: Jackson 52, Collins 39¹⁴

YouGov shows similar age trends among likely voters: Jackson leads 51–32 among those aged 18–34 and 49–41 among those 35–49. Collins leads 50–45 among voters aged 50–64 and 46–42 among those 65 and older.¹⁵ Her support is concentrated among older voters, reflecting the 2014 map. However, this is insufficient if younger, college-educated voters participate in large numbers.

The quiet CNN number is seniors: 65-and-over is Jackson 48, Collins 47.¹⁶ When the incumbent is tied with seniors in Maine, the firewall is cracked.

This election is a referendum on both Collins and Trump.

CNN Jackson voters:

• 71% voting against Collins

• 29% voting for Jackson

• Women backing Jackson: 76% against Collins¹⁷

Collins voters are the mirror: 74% for her, 25% against Jackson.¹⁸

YouGov further indicates that 99 percent of Jackson voters have made up their minds, with only 1 percent open to switching to Collins. Among Collins voters, 95 percent are committed, while 5 percent may consider Jackson.¹⁹ This time, late-deciding voters do not automatically benefit the incumbent.

According to CNN, Jackson remains relatively undefined, with 36 percent favorable, 42 percent unfavorable, and 22 percent undecided. Collins, by contrast, is well known but faces a 41 percent favorable and 51 percent unfavorable rating.²⁰ Gideon, despite a $70 million campaign, was a familiar candidate who still lost among women voters. Jackson does not need overwhelming popularity if the election centers on opposition to Collins and Trump.

Why this is not 2020

Gideon led in most 2020 polls but ultimately lost by 8.6 points. A lead of three or four points does not guarantee victory.²¹ Two key factors have changed since then.

First, Trump is currently president. According to CNN, 46 percent of voters say their Senate vote is a message against him. Collins voters approve of Trump by 66–30, while Jackson voters disapprove by 99–1.²² YouGov reports Trump’s approval at 39–60. Previously, Collins benefited from Mainers who disliked Trump but supported her; that group has diminished.

Second, the Dobbs decision occurred after 2020. CNN finds that nearly two-thirds of Maine voters view the 2022 ruling negatively, with about half reporting anger over the decision. Most of these voters support Jackson.²³

Third, this trend isn't limited to a single poll. Since Jackson became the nominee, multiple polls show him leading: UNH 49–46, Fox 48–46, Hart 49–45, CNN 48–45, and YouGov 48–44 among likely voters. Aggregators as of September 10 show Jackson ahead by two to three points.²⁴ Despite Platner’s withdrawal, the race has not shifted back to Collins.

The honest limit

YouGov’s four-point lead for Jackson is within the 4.9-point margin of error. According to CNN, Collins still leads among men, non-college voters, and Republicans. Jackson’s favorability remains negative. A late campaign shift or lower turnout among college-educated women could still benefit Collins, as she has overcome similar challenges previously.

However, two polls now show consistent trends. Maine’s largest registered group is Democratic. In CNN’s poll, unenrolled voters and women favor Jackson by significant margins, including a 15-point lead among women. Younger voters also strongly support Jackson in both surveys. The governor’s race is similarly favorable for Democrats. Trump’s approval is at 31 or 39 percent. CNN finds the Senate vote is a 46–17 anti-Trump message, and YouGov reports Jackson’s voters are firmly committed.

No previous Collins challenger has benefited from this combination of factors. Gideon had financial resources and favorable polls but did not face Trump as president with low approval, the Dobbs decision, a 15-point lead among women, support from independents, a tie among seniors, and confirmation from multiple national polls.

Jackson now holds these advantages, placing him in a stronger position. While this does not guarantee victory, the race dynamics have fundamentally changed.

Notes

1. Dave Leip’s Atlas, 2020 Maine U.S. Senate: Collins 50.98%, Gideon 42.39%.

2. Washington Post, 2020 Maine: Biden 53.1%, Trump 44.0%. Politico seat history: 2002 R+17, 2008 R+23, 2014 R+37.

3. CNN/SSRS, Aug. 31–Sept. 6, 2026; CNN Politics, Sept. 9, 2026; DocumentCloud crosstabs.

4. YouGov, Sept. 2–8, 2026, 1,335 LV / 1,378 RV, ±4.9. Political Wire; Wikipedia polling table; YouGov PDF ttw_me_20260902.pdf.

5. Independent Voter Project, 2026 Maine registration: D 33.7%, R 29.7%, unenrolled 32.3%.

6. Official 2024 Maine presidential: Harris 52.4%, Trump 45.5%.

7. Collins is the only Republican senator from a state Trump has not won in three presidential races.

8. CNN/SSRS governor: Pingree 52%, Charles 39%. YouGov LV: Pingree 47%, Charles 36%, Bennett 2%.

9. CNN/SSRS Table A1; YouGov Q4 (20% strongly + 19% somewhat approve; 57% strongly + 3% somewhat disapprove).

10. NBC News 2020 Maine Senate exit polls.

11. CNN/SSRS Table SENVOTEME.

12. CNN/SSRS Table TRUMPVOTE.

13. Ibid., Collins-voter column.

14. CNN/SSRS Table SENVOTEME.

15. YouGov age stubs, likely voters.

16. CNN/SSRS, 65+: Jackson 48%, Collins 47%.

17. CNN/SSRS Table FORCANDMED.

18. CNN/SSRS Table FORCANDMER.

19. YouGov firmness questions, Sept. 10, 2026 reporting.

20. CNN/SSRS favorability tables.

21. RealClearPolitics 2020 Maine Senate polling vs. final.

22. CNN/SSRS Tables A1 and TRUMPVOTE.

23. CNN Politics write-up of the SSRS poll, Sept. 9, 2026.

24. UNH July; Fox August; Hart August; CNN/SSRS September; YouGov September. 270toWin / RCP / FiftyPlusOne averages as of Sept. 10, 2026.

Welcome to WONKSTACK.

The main political voices don’t reflect the wider nation; the country aligns more with the values of Teddy Roosevelt than either party admits.

I’m launching WONKSTACK because, in 2026, it is no longer possible to summarize American politics accurately with a simple phrase.

The political spectrum is still seen as a straight line: the far left, the center, and the far right. This view encourages division rather than understanding.

That map is wrong.

Today’s leading political parties are not the traditional left or right. Rather, they consist of two groups, each backed by funding and media algorithms, which have pulled the extremes even further apart and ignored the middle ground. One concentrates on identity politics, while the other concentrates on personality and donor interests. Both parties make use of the working majority for their own purposes.

The people call those in the genuine American middle class radicals.

I call this about-faceentrism.

Radical centrism isn’t about compromise for its own sake or avoiding offense; in my definition, it means the following:

The prevailing opinion, one that focuses on working people and the middle class, opposes unfair systems, and rejects divisive strategies—is now seen as extreme by both major political groups.

If it seems familiar, it should. Theodore Roosevelt used that tactic more than a hundred years ago and called it a Square Deal.

Defining “radical” in this context

The term “radical” in this context doesn’t imply extremism or inflexible standards; rather, it means tackling problems at their source.

The main problem is not a hidden desire for extreme ideologies. Rather, many voters have noticed that both parties put special interests before their communities and future, and use culture wars to divert attention.

As I have stated previously on X, and will reiterate here:

• Extremism is to be found in both parties. Say that out loud.

• Oligarchs make donations to and influence policy-making in both cases. Say that too.

• We will not agree on all issues, nor will we disagree on all of them. The existing political system prevents this kind of nuance from being acknowledged.

Political experts frequently regard support for policies such as Medicare for All, higher taxes on the wealthy, and ending foreign wars as merely exceptional. In fact, this shows that the groups referred to as the ‘base’ or the ‘left’ usually share similar material interests. The divisions that are promoted are generally distractions meant to aid fundraising.

That is why I call it radical centrism rather than moderation; nowadays, moderation usually means preserving the status quo, and Roosevelt made it clear that the Square Deal was different.

By saying that I am in favor of the Square Deal, I do not merely mean that I support fair play within the present rules of the game, but that I am in favor of altering those rules to secure greater equality of opportunity and greater equality of reward for equally good service. Fair play is ineffective when the rules are unfair, and altering the rules so that work is rewarded aligns with a fundamental American tradition that is frequently ignored. Roosevelt’s three C’s were not on the left or the right; they stood for the public interest.

Historians compress the Square Deal into three C’s:

1. Control of corporations

2. Consumer protection

3. Conservation

Take these principles out of the Progressive Era; they remain the same.

Regarding the control of corporations, Roosevelt did not oppose business as such, but he did oppose the idea of powerful interests determining laws for the public. Although he believed economic combinations were inevitable, he thought the public should watch them rather than be controlled by them. The aim of breaking up trusts was to ensure the government remained accountable to citizens, not corporations.

In today’s context, private equity is harming communities, pharmaceutical companies are setting and colluding to keep prices high, technology firms are controlling public discourse and working with the federal government to censor and monitor speech, and, of course, undisclosed money affects both sides. To address these problems, we don’t need to fix the economy; we need a government that can curb excessive influence and establish those protections.

Consumer protection—the problems pointed out in The Jungle were genuine and serious, and as a result, the Meat Inspection Act and the Pure Food and Drug Act were passed. These laws prevented sellers from putting consumers in danger and then treating that as a form of freedom.

In today’s terms, this means treating housing as an investment rather than as a place to live, charging excessive fees, engaging in predatory medical billing, causing volatile food and energy prices through speculation, and taking advantage of personal data; the Square Deal maintains that consumers have a right to fair treatment.

Conservation—Roosevelt set aside 230 million acres of forest land, refuges, and parks because a country should not give up its natural resources for temporary benefit.

In practice, this means protecting fisheries, working waterfronts, public lands, water resources, and infrastructure. Stewardship should be regarded as a way of life, not merely as an abstract idea. A nation that cannot preserve both its industries and its environment has no credibility.

What matters is that, to support affordable healthcare, housing, or environmental protection, you don’t need to accept all of a particular group’s beliefs.

The fact that there is no such thing is deliberate, which is why both main groups oppose this method. One backs economic reform only if it comes with its cultural program, while the other changes the subject when economic issues arise and threaten its interests.

Roosevelt used the power of the nation to advance the public interest, without assuming labor was always right or capital always wrong. He got involved in the 1902 coal strike and opposed monopolies to ensure the country met the needs of its citizens.

It is progressive in the conventional sense because the government acts as an impartial referee for the majority rather than as an advocate for any one group.

Why the middle had to become “radical”

Centrism has acquired a bad image because, for many years, it has generally served donors’ interests, accepted unbalanced trade agreements, caused constant conflicts, allowed finance to go unchecked, and been involved in divisive culture wars.

That method is centralism, not the same.

Everyday issues still form the heart of American life: affordable housing, access to healthcare, education without excessive debt, stable jobs, secure infrastructure, responsible resource use, and the fight against corruption.

Most voters back these policies unless they are tied to any party; they seem “radical” only when political extremists treat loyalty as normal.

If the political spectrum is to be depicted, it now looks like a vise: on one side, justice is talked about while failing institutions are protected; on the other, the working class is said to be supported, but private interests are advanced. People looking for fairness end up in the middle.

Both sides frequently criticize the middle group, yet challenging a defective system is just what responsible citizens should do.

What WONKSTACK is for

It is not a newsletter that presents both sides, and it is not a political party organ.

WONKSTACK exists to do the unfashionable work:

• Identify any undue influence—no matter what the source may be: money, the media, consultants, or platforms- regardless of one’s political affiliation.

• We should examine a policy on its core rather than merely how it is presented. If a policy benefits only a small number of people even though it claims to be populist, we will point it out. Similarly, if a program is described as radical simply because it opposes monopolies, we will address that issue too.

• We should consider the middle-class view; blue-collar workers are not political symbols. They are ordinary people dealing with real problems.

• It is important to keep a point of view rooted in Maine; unlike those involved in politics, a local community’s outlook is worth having.

Economic populism is now in the middle of politics. For a long time, the working class has known that both parties are to blame for unfair policies, and they want to be listened to, understood, and given evidence, not drawn into partisan disputes.

WONKSTACK provides the detailed discussion that short online posts cannot.

The Square Deal

The Square Deal offered neither an easy life nor the idea that only people in power should set the rules.

A century later, the trusts have new names, and the muckraking arrives as a screenshot, but the assignment has not changed:

Take control of the companies that have grown beyond the citizen.

Keep consumers from being exploited.

Once something has been lost, we MUST preserve it.

One does not need to belong to any faction to serve working people; what is needed instead is the courage to act with integrity.

Although both extremes might still regard this method as radical, the middle ground has always been where most Americans actually lie.

What is new is that holding this view now requires more courage than simply choosing a side.

Here is the mission. That’s WONKSTACK.

— Maine

Radical Centrist. Policy Wonk. Blue Collar Advocate.
🫎🇺🇸🦞

Find Me on X: TheMaineWonk