Boss Trump: American Caudillo

In November 2024, Starr County, Texas — 97% Hispanic, 83% non-English-speaking at home, a place where Democrats won by 73 points as recently as 2012 — voted for Donald Trump by 16 points. It was not alone. Across the Rio Grande Valley, across Miami-Dade, across the bodegas of the Bronx and the barrios of Queens, Trump achieved something no Republican had managed in a century: he made massive inroads with the communities least integrated into English-speaking American civic life.

The question that has consumed political analysts ever since is: Why?

The standard explanations — inflation, masculine affect, TikTok — are not wrong. But they miss the deeper pattern. What Trump did in 2024 was not new. It was very, very old. He ran a classical clientelist campaign: showing up in person, offering tangible favors, performing loyalty to specific communities, and asking for loyalty in return. He was not the first American politician to do this. He was just the first one in a long time to do it at national scale.

He was, in a word, a boss. An American caudillo.

The Data: Language Predicts Everything

To understand what happened in 2024, forget race. Forget "Hispanic voters" as a monolith. The single most powerful predictor of a county's swing toward Trump between 2020 and 2024 was not its share of Latino residents, or its poverty rate, or its foreign-born population. It was the share of residents who speak a language other than English at home.

Using Akashic Edge's database of 165 years of American election results joined to Census ACS language data across 3,102 U.S. counties, the population-weighted correlation between a county's share of non-English speakers and its 2020-to-2024 Republican swing is r = 0.76 — an extraordinarily strong relationship for a single demographic variable predicting a nationwide partisan shift. Population weighting matters because the effect is concentrated in large, diverse counties: the Bronx, Queens, Miami-Dade, the Texas border.

For comparison, the population-weighted correlation with foreign-born share is r = 0.71, with Hispanic share r = 0.68, and with college education essentially zero (r = -0.04). Language — or more precisely, distance from the English-speaking mainstream — is the variable that unifies the South Texas border, the Bronx, Miami-Dade, Queens, and the Central Valley of California.

The Gradient

Broken into deciles by share of non-English speakers, the pattern is monotonic and dramatic:

Across the first nine deciles — counties where most people speak English — the swing toward Trump was a modest and fairly uniform R+2.5 to R+4. But in the top decile, where more than a third of residents speak a language other than English at home, the swing more than doubled to R+7.6.

This isn't just "Hispanic counties." Apache County, Arizona — 48% non-English-speaking, but mostly Navajo, not Hispanic — swung R+14.8. Queens County, New York — 52% non-English-speaking, heavily Chinese, South Asian, and Caribbean — swung R+21. The thread that connects Starr County to the Bronx to Osceola to Miami-Dade isn't ethnicity. It's linguistic distance from the American mainstream.

The Parade of Bosses

To understand why this pattern emerged, you have to understand what Trump actually did in these communities. Not what he said on cable news. What he did on the ground.

The Bodega Visit. In April 2024, fresh from a Manhattan courtroom, Trump detoured to a bodega in Hamilton Heights — a heavily Hispanic section of Harlem — where clerk Jose Alba had been charged with murder after defending himself from an attacker. Trump met with bodega workers, listened to their complaints about crime, and promised to help. The Bodega Association of America, representing thousands of immigrant-owned corner stores, welcomed him. It was pure ward-boss politics: identify a community grievance, show up in person, promise to fix it. Tammany 101.

The Church Circuit. Trump's campaign deployed Hispanic pastors across South Texas, South Florida, and the Central Valley of California to vouch for him in churches. He appeared personally at 180 Church in Michigan and at events with Black evangelical leaders in the South Bronx. Among Hispanic evangelicals — 15% of all Latinos and growing fast — nearly half already leaned Republican. Trump didn't need to convert them. He just needed to show up. He showed up.

The Gold Sneakers. At a sneaker convention in Philadelphia in February 2024, Trump unveiled $399 gold high-tops embossed with his name. The symbolism was not subtle: conspicuous consumption as aspiration, the kind of brash wealth display that resonates in communities where material success is the most legible form of status. The sneakers sold out. Later, he released $499 Bitcoin-themed sneakers that sold out within three hours.

The Crypto Patronage. At the Bitcoin 2024 conference in Nashville, Trump promised to fire the SEC chair, create a "strategic bitcoin reserve," and make America "the crypto capital of the world." Once elected, his family launched World Liberty Financial and the $TRUMP memecoin, earning hundreds of millions in personal wealth. Crypto firms that donated to Trump saw federal investigations halted. Executives who backed his ventures received pardons. This is patronage — the exchange of tangible favors for political support — operating at industrial scale.

None of this is how a normal presidential campaign works. It is, however, exactly how Tammany Hall worked.

Tammany's Ghost

In the 1850s, over 130,000 Irish immigrants arrived in New York in the wake of the Great Famine. They were poor, Gaelic-speaking, Catholic, and utterly excluded from Anglo-Protestant civic life. Tammany Hall — initially hostile to immigrants — realized it needed their votes. The machine's bosses, especially William "Boss" Tweed, offered immigrants something that respectable reformers never would: material help, immediately, with no moral strings attached.

Tweed made sure the immigrants had jobs, found a place to live, had enough food, received medical care, and even had enough coal money to warm their apartments during the cold of winter.

Tammany didn't ask the Irish to assimilate first. It didn't ask them to learn English, to adopt Protestant values, to demonstrate civic virtue. It said: You need something? I can get it for you. Now vote for my guy.

The parallels to Trump's approach are precise:

The key insight — which historian Terry Golway makes in Machine Made — is that Tammany's corruption and its social services were inseparable. The machine could help people because it had patronage money. It could mobilize voters because it controlled favors. The positive and negative aspects were two sides of the same coin.

Trump's model works the same way. The crypto patronage, the transactional pardons, the personal brand merchandising — these are not bugs in the system. They are the system.

The Caudillo Template

The comparison extends beyond American machine politics. In Latin America, the tradition of the caudillo — the charismatic strongman who rules through personal authority, patronage, and a direct bond with "his people" — has a long and well-studied history.

As early as 2016, Foreign Affairs observed that Trump's "braggadocio, demagoguery, and disdain for the rule of law" placed him squarely in the caudillo tradition — more Peron than Mussolini, more Chavez than Franco. The caudillo doesn't operate through ideology. He operates through personal relationships. He bypasses Congress, the media, and his own party to forge a direct connection with his base. His power is charismatic and non-transferable.

This is why the "Trump = fascist" frame has never quite worked. Fascism is institutional — it builds a party, a bureaucracy, a youth movement. Caudillismo is personal. It dies with the caudillo. And this distinction matters enormously for understanding what the 2024 results mean — and what they don't.

For immigrant communities that feel poorly served by American institutions — the healthcare system, the education system, the legal system — the appeal of a patron who gets things done is not irrational. It's the most rational political calculation available when you don't trust institutions to work for you.

The Communities That Moved

Let the data speak. Here are the 15 most non-English-speaking counties in America (pop > 5,000) and their presidential margins across four cycles:

Look at the trajectory. These aren't sudden converts. The erosion began in 2016, accelerated massively in 2020, and completed in 2024. Starr County's 89-point swing from 2012 to 2024 is the largest single-county partisan shift in modern American history.

But these aren't all the same kind of community. The Bronx (53% non-English, 55% Hispanic, 35% foreign-born) swung R+22.5. Queens (52% non-English, 28% Hispanic, 48% foreign-born — heavily Asian) swung R+21. Passaic, New Jersey (48% non-English, 44% Hispanic, 33% foreign-born) swung R+19. Miami-Dade (71% non-English, 69% Hispanic, 55% foreign-born) — a county of 2.7 million people — flipped outright from D+7 to R+11.

The 1920s Mirror

The data above tells a story that happened once before — in reverse.

In the 1920s, the Republican Party dominated American politics with a coalition eerily similar to the Democrats' recent one: they were the party of cities, of industry, of the educated professional class, and — critically — of immigrants. Harding won 61% of the popular vote in 1920, carried every non-Southern state, and drew massive support from Irish, German, and Italian immigrants who had turned against Woodrow Wilson over the League of Nations and wartime Prohibition.

This wasn't a soft lean. Look at Akashic Edge's county-level results from the 1920 presidential election in what are today America's most linguistically diverse counties:

In 1920, the Bronx voted Republican by 32 points. Queens by 43. Cook County — Chicago, the great immigrant city — by 49. These were the bedrock of the GOP coalition. Immigrant communities, especially Irish and Italian Catholics, voted Republican because Lincoln freed the slaves, because the party stood for industry and opportunity, and because Tammany's Democratic machine — however helpful — was associated with corruption and parochialism.

Then Republicans made a fatal error. The Immigration Act of 1924, signed by Calvin Coolidge, imposed the most sweeping restriction of immigration in American history — national-origin quotas designed to freeze the ethnic composition of the country. The message to the Italians, Poles, Jews, and Greeks who had been drifting toward the GOP was unmistakable: We'll take your votes, but we don't want more of you.

The breaking point came in 1928. Al Smith — Catholic, wet, a child of the Lower East Side — ran for president and the immigrant urban vote flipped overnight. The Bronx swung 71 points. Manhattan 55. Brooklyn 61. Boston 56. Even Cook County, Republican by 49 points in 1920, shrank to R+6. Smith lost the election, but he permanently realigned the immigrant cities. Four years later, FDR would cement the shift into a coalition that lasted six decades.

The Off-Year Paradox: It Happened Before

Here is where the 2025 backlash data gets interesting — because the exact same thing happened in the 1920s.

Analysts point to Democrats' D+15 overperformance in 2025 special elections as evidence that Trump's gains are illusory. But the 1920s GOP faced identical off-year backlash while immigrant communities were actively realigning toward their party — and the realignment stuck anyway. The backlash and the realignment were separate phenomena, operating simultaneously in different parts of the electorate.

Akashic Edge's database of 1920s governor and Senate races tells the story:

Read that carefully. In 1922 — just two years after Harding's historic landslide — Democrats won the New York governor's race by 15 points. They swept New Jersey. They flipped a Michigan Senate seat in a state Harding had carried by 51 points. This is almost exactly the pattern of 2025, where Democrats are winning special elections in districts Trump carried by 20+ points.

And yet: the GOP won the next two presidential elections anyway. Coolidge won by 25 in 1924. Hoover won by 17 in 1928. The off-year backlash did not prevent continued Republican presidential dominance. And — crucially — the immigrant-city realignment continued during this period regardless of the off-year results. The Bronx drifted from R+32 (1920) to R+3 (1924) to D+39 (1928) on a steady trajectory, completely independent of what was happening in New Jersey governor's races.

The lesson: off-year backlash and long-term realignment can coexist. They operate in different electorates. The suburban professionals who swing against the governing party in special elections and the non-English-speaking communities drifting toward a new partisan home are different people responding to different forces. One does not disprove the other.

The Case That Trump Is Hoover

With that context, here is the case that Trump's gains will collapse — not because of the backlash, but because of the betrayal.

The immigration betrayal. The Republicans of the 1920s courted immigrant votes while legislating against immigrant communities. Trump is doing exactly this. He won Starr County, then launched the largest mass deportation program since the Eisenhower era. He flipped Miami-Dade, then revoked Temporary Protected Status for hundreds of thousands of Venezuelan and Haitian immigrants — many of them Miami-Dade residents. He gained 10 points among Hispanics nationally, then empowered ICE to conduct raids in schools, hospitals, and courthouses in Latino neighborhoods.

The 1924 Immigration Act didn't produce an immediate backlash in immigrant communities. The Bronx still moved toward the GOP between 1916 (D+8) and 1920 (R+32). The betrayal built slowly. But when Al Smith offered an alternative in 1928 — someone who looked like them, who spoke like them, who came from their world — the reservoir of resentment burst all at once. The Bronx swung 42 points in a single cycle.

Trump's immigration policies are building that same reservoir. The question is when — not whether — it breaks.

The non-transferable coalition. The 2025 special elections show a genuine weakness in the Republican position. Democrats have overperformed 2024 presidential margins by 13-16 points across 46 contested races. They have flipped 25 GOP state legislative seats. Republicans have flipped zero. Iowa SD-35 — Trump +21 in 2024 — elected a Democrat by 10.5. Florida CD-1 — Trump +38 — saw a 23-point Democratic overperformance. Spanberger won Virginia by 11. Sherrill won New Jersey by 14.

Yes, the 1920s GOP also faced off-year backlash and still won presidential elections. But here's the difference: Coolidge and Hoover won those elections with their own appeal, not Harding's. Trump's coalition appears to be non-transferable in a way the 1920s GOP's was not. Tammany survived Boss Tweed because it had a deep bench of ward bosses. Trump has family members, social media, and a handful of loyalists. No Republican running in 2025 can replicate what Trump did in 2024 — because the bond is personal, not institutional.

No institutional deliverables. FDR didn't just win immigrants with charisma. He built things — Social Security, the Wagner Act, the CCC, the FDIC. These were tangible benefits that bound immigrant communities to the Democrats for generations. What has Trump built for the non-English-speaking communities that swung to him? Crypto deregulation benefits tech bros, not bodega owners. Tariffs raise prices at Walmart. The border wall threatens their relatives. DOGE is cutting the social services they rely on. The caudillo showed up, made promises, and delivered — what, exactly?

The Case That Trump Is FDR

And yet. The case for a durable realignment is stronger than Democrats want to admit — and the 1920s off-year data actually helps the case.

The shift is gradual and persistent — exactly like the 1920s. The New Deal coalition didn't appear overnight. The Bronx drifted from R+32 (1920) to R+3 (1924) to D+39 (1928) to D+51 (1932) — a four-cycle arc that was completely independent of what was happening in off-year elections. The Texas border is following the same trajectory. Starr County: D+73 (2012), D+60 (2016), D+5 (2020), R+16 (2024). Four consecutive elections, each more Republican than the last, each by a larger increment.

The 1922 midterm backlash — D+15 in the New York governor's race — did not slow down the immigrant-city realignment at all. The Bronx kept drifting Republican right through 1924. If the 2025 special election backlash follows the same pattern, it tells us nothing about whether Trump's gains in non-English-speaking communities will persist.

The underlying forces are structural, not personal. The r = 0.76 population-weighted correlation between non-English speakers and Republican swing is not about Trump's charisma. It's about something deeper: the failure of Democratic institutions to serve linguistically isolated communities. These are communities where the healthcare system doesn't speak your language, where the school bureaucracy is impenetrable, where the legal system is terrifying, and where the Democratic Party's offer is — what? More bureaucracy? More means-tested programs with 40-page applications in English?

The appeal of a patron who cuts through the bullshit is not irrational when the institutions have never worked for you. And that appeal doesn't disappear when the patron does — it just waits for the next patron.

Hispanic evangelicalism is durable institutional infrastructure. Roughly 15% of all Latinos identify as evangelical or born-again, and among them, Republican identification was already approaching 50% before Trump. The evangelical church network is exactly the kind of institutional base that the caudillo supposedly lacks — pastors who know their congregations personally, who vouch for candidates from the pulpit, who organize voter drives after Sunday service. This is not Trump's personal machine. It's a pre-existing social network that he activated but didn't create — and it will persist after he's gone.

Class is replacing race as the primary axis of American politics. The 2024 election wasn't really about Hispanic voters moving right. It was about working-class voters of all races moving right. Non-college white voters have been trending Republican since the 1960s. Non-college Black voters shifted toward Trump by roughly 8 points in 2024. And non-college Hispanic voters — the core of the non-English-speaking communities in this analysis — shifted by 20+. If the realignment is class-based rather than personality-based, it will survive Trump's departure.

Democrats have no Al Smith. The 1920s realignment required two things: a Republican betrayal (the 1924 Immigration Act) and a Democratic champion who spoke directly to immigrant communities (Al Smith in 1928, FDR in 1932). Today's Democrats have the betrayal — mass deportation, TPS revocations, ICE raids — but they have no Al Smith. The party's leadership is aging, its bench is thin, and its cultural affect — educated, professional, institutionalist — is precisely the thing that alienates linguistically isolated communities. Who will walk into a bodega in the Bronx and speak to the owner in terms he understands? Kamala Harris lost these communities by historic margins. There is no obvious successor who would do better.

Where This Is Going

The 1920s parallel provides a framework precise enough to make a prediction.

In the 1920s, two forces operated simultaneously: off-year backlash against the governing party (1922, 1926, 1930) and long-term realignment of immigrant communities (1920 → 1932). The backlash happened in the suburbs and among swing voters. The realignment happened in the immigrant cities. They were different electorates moving for different reasons. Both were real. Neither cancelled the other out.

The same thing is happening now. The 2025 special election backlash is real — D+15 is enormous, and it will probably produce a significant Democratic wave in 2026. But the realignment of non-English-speaking communities is also real — r = 0.76, three consecutive presidential cycles of movement, and structural forces that predate and will outlast Trump.

The critical lesson from the 1920s is that realignments are not decided by the party gaining the new voters. They are decided by what happens next. The GOP gained the immigrant cities in 1920. It lost them by 1932 — not because the Democrats were brilliant, but because the Republicans betrayed their new coalition with the Immigration Act and then failed to deliver during the Depression. FDR won the immigrants because Coolidge and Hoover lost them first.

We are at the equivalent of 1924 in this cycle. Trump has gained the non-English-speaking communities. The off-year backlash is fierce. But the immigrant-community realignment is still moving in the Republican direction, just as it was still moving Democratic in the 1920s despite GOP off-year wins.

What decides the outcome is what happens in the next four to eight years — specifically:

Our assessment: the most likely outcome is a partial realignment that stabilizes below Trump's high-water mark. Hispanic evangelicals and Cuban-Americans will likely remain Republican — the shift there is ideological and institutional, not personal. But the dramatic border-county swings — Starr's 89 points, Maverick's 86, Webb's 77 — are inflated by the caudillo effect and will partially revert. The non-English-speaking working class will settle somewhere in the middle: no longer reliably Democratic, but not permanently Republican either. A contested battleground, like the Midwest after the New Deal coalition fractured in the 1960s.

The 1920s teach us that realignments are not won with charisma. They are won with institutions — and lost with betrayal. The GOP has the charisma. It does not yet have the institutions. And the betrayal is already underway.