The Outsider Perspective Issue 500 | The Beltway Outsiders

If you’d like to read this issue on my website, click here! If you’d like to sign-up and receive this in your inbox each week, click here! Read past issues here.

Good Friday Morning! Voting in the first Fat Marmot Week opens August 24. The Rocky Mountain Biological Laboratory sits in Gothic, Colorado, and it has been studying yellow-bellied marmots since 1962, which makes it the second-longest-running study of its kind. The marmots spend every summer eating until some of them weigh three times what they did in spring. Then they sleep for five months. The public gets a bracket and votes on who did it best.

The lab needs money. Federal research funding tightened, so the scientists opened an OnlyFans account called OnlyMarms and posted wholesome marmot videos on it. Emily Renkey, a UCLA scientist on the project, explained the origin: their advisor was watching a show about OnlyFans and thought “marmots also have money troubles, what if we started an OnlyFans?” By early August the account had more than 2,100 followers and had raised about $6,000, after OnlyFans took its 20 percent.

A sixty-four-year-old alpine ecology program is now partly funded by people subscribing to see a fat rodent. It is also not a joke to the marmots. Dan Blumstein, the UCLA professor who runs the Marmot Project, says the animals “give us a window on the health of high-alpine ecosystems,” and about a hundred of them died in one recent bad year.

It’d make more sense to dump more money on to the Marmot OnlyFans account than it would to pay pollsters in the Midwest right now. We’re witnessing some extreme polling misses in Democratic Party primaries. I’ll get into that and more below – links to follow.

Quick Hits:

El-Sayed said he intends to live by Sharia until the day he dies. Andrew Kerr and Jessica Schwalb of the Washington Free Beacon went back to a 2009 New York Times story on Sharia-compliant mortgages and found Abdul El-Sayed in it. Michigan’s Democratic Senate nominee was a 24-year-old medical student then, buying a $123,000 condominium in Ann Arbor, and he explained why he financed it the way he did. “Ultimately, the question is, when I die and I stand before God and go through everything I did in my life, I don’t want to say I did it the easy way instead of the Shariah-compliant way. Not because of fear but because of obligation.” His campaign told the Beacon he “believes in the separation of church and state” and “opposes any effort by any religious community to write their faith laws into the laws of our country.” It’s quite the turn from Democrats. In 2017 Dianne Feinstein told Amy Coney Barrett that “the dogma lives loudly within you, and that’s a concern,” and Dick Durbin asked her outright whether she considered herself an orthodox Catholic. Now Democrats are pushing a Sharia-compliant man for Senate.

Where you can find me this week

Please subscribe, rate, and review The Horse Race on YouTube — the reviews help listeners, and readers like you find me. Make sure to sign up for the Conservative Institute’s daily newsletter and The American Almanac.

AOC Is Freezing Her Eggs, And The Left Isn’t Telling Women The Odds – Conservative Institute

Democrats Should Ask Cubans Before Calling This A Red Scare – Conservative Institute

Democrats Lost Billions And Punished Everyone Who Noticed – Now They Want Your Vote – Conservative Institute

The Polls Are Overselling The Socialists

On July 29, the Marquette Law School Poll released its survey of the Democratic primary for governor of Wisconsin. Francesca Hong, a state representative from Madison and a self-described democratic socialist, led with 38 percent. Former Lieutenant Governor Mandela Barnes sat second at 16. Milwaukee County Executive David Crowley came in third with 7 percent. A third of the electorate, 34 percent, had not decided.

Crowley won on August 11 with 39.8 percent.

Charles Franklin runs that poll. He is the most respected survey researcher in Wisconsin, and two weeks before the vote his poll had the winner in single digits. On August 12 he published an autopsy of his own work. His finding: “the vast majority of late-deciding voters went to Crowley and very few went to Hong.”

The undecideds did not split down the middle. They moved together, they moved late, and they moved to the man polling at seven.

The field changed underneath the survey. Marquette went into the field July 22 and came out July 27. Barnes dropped out afterward and took his 16 percent with him. Crowley re-entered the race. Governor Tony Evers endorsed him, and 36 percent of Democratic primary voters told Franklin the endorsement made them more likely to vote that way. Everything that decided the race happened in the window the poll could not see.

Four more elections in eight days broke the same way.

The socialists really are taking over the party. That is not the question.

The fight inside the Democratic Party is real, and I have spent months documenting it. The Democratic Socialists of America hold real seats, run real primaries, and beat real incumbents. Zohran Mamdani runs New York. Abdul El-Sayed is the Democratic Senate nominee in Michigan. The Democratic National Committee is rewriting its presidential calendar to stop the left, which is not something a party does to a faction it has already beaten.

How many actual voters stand behind that faction is a separate question, and the instruments that answer it failed four times in eight days.

None of that makes November a Republican year. My read stands. Democrats are favored to take the House. The Senate leans Republican and probably stays Republican. Sean Trende, the senior elections analyst at RealClearPolitics, said in June that “it’s just really hard to imagine a situation where the Republicans don’t end up netting losses in the House, flipping the chamber.” He predicted a 51-49 Republican Senate. Jessica Taylor of the Cook Political Report told NPR in April that Democrats gain one to three seats, short of the four they need.

Polling error is not a partisan gift. Silver Bulletin, Nate Silver’s outfit and the successor to FiveThirtyEight’s ratings work, measured 2025 and found the polls carried “an average 6.1-point bias toward Republicans.” Trende’s warning covers the rest: “It’s ultimately a loser’s bet, trying to guess exactly what the partisan composition of the polls should be and adjust it accordingly.”

If November lands inside the polling averages the way 2018 and 2022 did, I was wrong and August was a primary-season quirk. I will say so in this space.

Four races in eight days, and the polls missed every one

Michigan, August 4. The RealClearPolitics average had Abdul El-Sayed, the Bernie Sanders-endorsed former Detroit health director, ahead of Congresswoman Haley Stevens by 10.2 points. The last three surveys had him up 14, 15 and 10. He won by one.

Wisconsin, August 11. State Navigate, a nonprofit polling operation, published a final survey giving Hong a 22-point lead. She lost by half a point on unofficial returns.

Minnesota, August 11. Minnesota Democrats run under the party’s own name, the Democratic-Farmer-Labor Party. The last independent public poll of its Senate primary, SurveyUSA for KSTP, had Congresswoman Angie Craig ahead of Lieutenant Governor Peggy Flanagan by two. Flanagan won by nineteen. The Democratic firm GQR had Flanagan up 13 and came close, so Minnesota is one bad poll rather than a whole field. It still ran twenty-one points the other way from Wisconsin on the same night.

Hawaii, August 8. State Senator Jarrett Keohokalole’s own campaign polling had him ahead of Congressman Ed Case by seven to nine points in late July. Case won by twenty.

Two overstated the left. One understated a progressive. One understated a moderate.

Nate Silver put the historical baseline at nine points of error for presidential primaries since 2000. That number does not cover these races and should not be used to wave them off. Presidential primaries run twenty states at once on a calendar that moves every week, with candidates entering and quitting on national television. A Senate primary in one state is a smaller, slower, more stable thing. The Wisconsin firm keeps its own report card, and its average miss runs 8.2 points in primaries against 2.7 in general elections.

Against that standard, a 22-point miss in a governor’s race is not the normal cost of doing business. Neither is 21 in a Senate primary, or 27 in a House race. Three of those inside eight days points at the samples, not at bad luck.

In 2016, Wisconsin’s last-week deciders broke 59-30

Undecided voters do not stay undecided. They decide late, they decide close to the vote, and they decide together.

The American Association for Public Opinion Research is the trade group American pollsters belong to. It audits the industry after a bad cycle. Its 2016 postmortem put the numbers in a table. In Wisconsin, voters who decided in the final week made up 14 percent of the electorate and broke for Donald Trump 59 to 30. Pennsylvania broke 54-37. Florida, 55-38. Michigan, 50-39. Courtney Kennedy, who led that review, named the consequence: “polls in September, October were conducted too early to detect this.”

Four states, one direction, the same week.

The pattern held. In 2020, voters who decided in the last week went for Trump 64-31. In 2024, NPR reported that last-few-days deciders broke for him by six and last-week deciders by twelve.

Primaries move harder, because party loyalty is not there to hold anyone in place. On Super Tuesday in 2020, NBC News found that nearly three in ten Democratic primary voters decided in the final days. Among those late deciders, Joe Biden beat Bernie Sanders 47 to 18. Twenty-nine points, inside a week, in a race the whole country had watched for a year.

David Gopoian and Sissie Hadler studied late deciders across three decades of national election surveys. Their choices, the two found, are “nearly random in character.” Christopher Wlezien of the University of Texas wrote the standard work on how campaigns move opinion. The late shift is real rather than noise, and it lasts through election day.

Nobody can tell you which way the last week breaks. Everybody can tell you it breaks one way, in most places, at once.

A night that loses Texas does not win Michigan

On August 13, Fox News released battleground surveys of Ohio and Michigan, fielded August 6 through 10. Its Texas survey came out July 30, from fieldwork done July 23 through 27.

Fox has Mike Rogers, the Republican former congressman who ran the House Intelligence Committee, beating El-Sayed in Michigan 51 to 47. Fox also has Democrat James Talarico beating Attorney General Ken Paxton in Texas 51 to 48.

Both of those cannot be true on the same night.

Texas is a Republican state. Michigan is a swing state that Democrats usually carry. An election that hands Democrats a Texas Senate seat is a national Democratic wave, and a national Democratic wave does not stop at the Michigan state line and hand Republicans a Senate seat there. The reverse fails the same way. A night good enough for Republicans to take Michigan is a night Texas never comes into play.

Late deciders break one direction, in most states, in the same week. No version of November exists where the last week breaks for Democrats in Texas and for Republicans in Michigan.

Fox has former Senator Sherrod Brown beating appointed Senator Jon Husted 53 to 45 while Vivek Ramaswamy, the biotech investor who ran for president in 2024, beats Democrat Amy Acton for governor 50 to 48. Fox measured the mechanism and printed it: Brown draws 11 percent of Ohio Republicans, 25 percent of non-MAGA Republicans, and 14 percent of Brown’s own voters cross over to back Ramaswamy. In Texas, a quarter of non-MAGA Republicans say they will vote for Talarico. Republican voters have not defected at those rates in three cycles, and they will not start in a midterm against a Democratic Party running El-Sayed and Mamdani as its new face.

These numbers are junk. Not slanted and not spun, just wrong in a way that makes every individual race in the set untrustworthy.

The sample is the likeliest culprit, and August is the likeliest reason. All three Fox surveys polled registered voters, not likely voters. A registered-voter sample in August carries a large bloc of people who have not started paying attention and will not vote. Wisconsin and Michigan just proved how badly that goes wrong. Two pollsters built samples that were too young and too enthusiastic, and both missed by double digits. Fox is fielding into the same electorate with a looser screen.

El-Sayed is losing a third of his own party, and Fox found it

Underneath the Michigan topline, El-Sayed takes 98 percent of Democratic Socialists of America supporters and 91 percent of Democrats, and 69 percent of non-DSA Democrats and independents. Fox’s own writeup states that he “is underperforming with Black voters, women, and suburban voters compared with former Vice President Kamala Harris in the 2024 presidential election.” Rogers voters are nearly three times as likely to cross over for Democrat Jocelyn Benson in the governor’s race as El-Sayed voters are to cross for Republican John James, 14 percent against 5.

A Democratic Senate nominee losing a third of the Democrats who are not socialists is the same weakness the primary polls never found.

Mary Radcliffe, an analyst at State Navigate, published this five days before Wisconsin voted, under the heading “The highest propensity voters are the least likely to support Hong.” Among people who voted in all four of the last four primaries, Hong led by 8. Among people who had skipped all four, she led by 50. Radcliffe called it “perhaps another reason to believe her lead here is inflated” because “Hong’s strongest support is among those with the least practice at the voting booth.”

The crosstabs held the answer. The headline number buried it.

Chaz Nuttycombe runs the firm. He had already tightened his weights after Michigan, cutting the youngest voters and adding ten points of seniors. He missed by 22 anyway, then graded his own final Wisconsin survey an F in public. Adam Carlson, who polled Michigan for a group supporting El-Sayed, reached the same conclusion: “more enthusiasm does not necessarily equal more votes.”

Silver named the voters the instruments keep missing: rural whites and working-class Black voters, “two groups that are notoriously hard to reach in surveys.” Patrick Ruffini of Echelon Insights, one of the sharpest polling minds on the right, named the mechanical failure: “polls are still a crude measure for consolidation, which only happens at the very end.”

The socialist coalition answers surveys and does not turn out. Every one of these misses runs through that gap.

They call it a workers’ party. Four percent of them are Black.

The Democratic Socialists of America surveyed its own membership in 2021 and published the results. About 13,000 members answered. Four percent of them are Black. Seventy-seven percent are white alone. More than 80 percent of respondents over 25 hold a bachelor’s degree. Among members over 24 who have not retired, 21 percent belong to a union, down from 31 percent in 2013.

That is a workers’ party where the college graduates outnumber the union members four to one, in a party where Black voters decide the nomination. The survey is self-selected and five years old. It is also the only count the group has ever published of itself.

The socialists did not build a workers’ movement. They built a faculty lounge with a mailing list, and for a decade they have insisted that the people who keep voting against them will come around once the message improves.

Sanders tested that theory first, and the record is not close. In the 2020 South Carolina primary, Biden took 61 percent of Black voters, who made up 56 percent of the electorate, against 16 percent for Sanders. Among Black voters over 65 it was Biden 78, Sanders 9. Four years earlier in Michigan, Hillary Clinton won about two-thirds of Black voters in a state Sanders carried. He won the state and lost the group. Nobody gets to blame the machine for a constituency he lost in a state he won.

Zohran Mamdani, now the mayor of New York City, ran the same experiment and got the same answer. In the June 2025 Democratic primary he lost majority-Black precincts by sixteen points to former Governor Andrew Cuomo. The Intercept, which is not a conservative outlet, reported it: “Cuomo won more than half of the votes in majority-Black precincts, while Mamdani got about 34 percent.”

He won the general election in November, and the class numbers are worse. E.J. Dionne, the liberal Brookings Institution columnist, read the exit polls: Mamdani beat Cuomo among college graduates 57 to 38, and took 40 percent of voters without college degrees against Cuomo’s 48. Among voters making under $30,000 a year, Cuomo won 48 to 42.

The democratic socialist lost the poorest voters in New York City to a Cuomo.

The city’s own Campaign Finance Board counted citywide turnout at 41.6 percent, the highest in a New York City election in more than 50 years. Manhattan hit 48.5 percent. The Bronx, the poorest and most heavily Black and Hispanic borough in the city, hit 30.3.

The biggest turnout surge in half a century skipped the poorest borough by eighteen points against Manhattan. Brownstones voted. Housing projects did not.

Juan Gonzalez and Matthew Wilson of the Great Cities Institute at the University of Illinois Chicago went through all 1,291 precincts in the 2023 runoff that elected progressive Brandon Johnson. Among registered voters: white turnout 61.1 percent, Black turnout 29 percent, Latino turnout 20.5. Johnson won 88 percent of the Black voters who cast a ballot. The coalition that put him in office voted at double the rate of the people he governs.

Missouri Democrats made the same choice twelve days ago. Congressman Wesley Bell beat former Congresswoman Cori Bush 59 to 37 in Missouri’s first district, which runs through St. Louis. Axios reported the alignment: Bush ran with Sanders, the Democratic Socialists of America and Justice Democrats behind her. Bell ran with House Democratic leadership and the Congressional Black Caucus. Twenty-two points.

Isaac Pablo Duarte joined DSA in 2023 and is one of the few Black members of its Bronx and Uptown Manhattan branch. He told Grassroots Thinking: “DSA doesn’t do well at all at engaging with Black people and Latinos.”

They know. They have known for years. They decided the voters are the problem.

Jacobin says the calendar exists to stop them

The socialists are going to lose 2028 on a schedule somebody else set.

I wrote two weeks ago about the Democratic National Committee’s Rules and Bylaws Committee putting South Carolina first on the 2028 presidential calendar and cutting Iowa out entirely. Forty-nine members voted on which state leads. South Carolina took 26, Nevada 19, New Hampshire 4. Nevada goes second, then New Hampshire, New Mexico, Michigan and Virginia. The full DNC meets in Austin this weekend to ratify it.

It will ratify. Joe Biden moved South Carolina to the front in 2024 over the same objections from the same states. New Hampshire held a rogue primary anyway. The party stripped its delegates and then quietly seated them. The DNC does what its sitting leadership wants on calendars, and the calendar is the one thing it still controls.

The left knows exactly what the calendar does. Branko Marcetic writes for Jacobin, the flagship magazine of the American socialist movement, and he told Axios that with the new order “the whole rationale is stopping a left-wing insurgent.” A member of the DNC’s own rules panel told the same reporter the calendar could “shut down” a progressive candidate. Christale Spain, who chairs the South Carolina Democratic Party, says Alexandria Ocasio-Cortez could absolutely win her state’s primary.

Biden won South Carolina 48.6 to 19.8 in 2020. Tom Steyer quit that night. Pete Buttigieg and Amy Klobuchar quit within three days and endorsed Biden before Super Tuesday. Black voters cast a majority of the ballots that ended the race.

Put a candidate who runs 34 percent in majority-Black precincts into that state first, and the movement dies before New Hampshire votes.

Democrats are fighting a real war with fake maps

On Thursday the DNC’s resolutions committee voted 25 to 12 to endorse abolishing Immigration and Customs Enforcement. That vote came at the same Austin meeting where the party was choosing which state gets to end the socialist candidacy first.

Those two votes are the same fight. The socialists have the energy and the primary wins. The establishment has the rulebook and is using it.

Neither side has an accurate count of the voters. Their instruments missed by nine, twenty-one, twenty-two and twenty-seven points inside a single week, and missed hardest on the exact question both sides are fighting about.

The socialists are winning the argument inside their own party. The polls keep telling them they are winning the country too. Wisconsin, Michigan and Missouri were elections rather than surveys, and all three said otherwise.