Like the parrot, this newsletter is not dead — it’s just been resting.

And while it has been resting, I’ve been busy. In addition to doing my job, I’ve been working on a book proposal (more about that, I hope, in a future post) and writing for various publications, mainly Business Insider. Here are the pieces I’ve published since my last update to this newsletter:

• A piece on the spread of homeowners associations, and why it should worry us.

• A piece on how restrictive zoning is pushing more people into the path of climate disasters.

• A call for America to embrace “weird” housing.

And my personal favorite: An essay for The Nation about how the 2024 election demonstrates the extent to which the rest of the United States has come to resemble California. As I was writing this one, it occurred to me that there’s probably an entire version of When the Clock Broke to be written just about California since the 1960s or 70s. You can think of my own essay as the rough outline for such a narrative.

That’s not at all. I also have a piece coming out for Dissent’s winter issue that I’m excited about. And even more exciting: sometime within the next 1-2 weeks, I’m going to become a father.

Needless to say, the latter news has left me both elated and terrified. Mostly, I am overwhelmed. It took my wife and I a while to get to this point: we struggled with infertility for about 18 months before finding a doctor who could actually help us. (I’ve started and then deleted about three different newsletter drafts revisiting Vance’s unsavory fixation on “childless cat ladies” in light of our experience.) Adding to my sense of being in over my head, our daughter is going to arrive within a week of an existential Election Day.

That Election Day is the subject of the below post. But before I get to it, one more quick note: I do hope to get on a more regular cadence with this newsletter in the coming months. My hope is that the discipline of writing these posts on some sort of schedule will prevent my brain from totally atrophying during parental leave. Maybe that’s wildly unrealistic. Maybe your personal experience as a parent demonstrates that it is wildly unrealistic! If so, I invite you to keep your observations to yourself.

And with that, let’s get to the subject at hand.

And Now For the Hopium

With just a few days to go until Election Day, the polls are basically where they’ve been ever since Joe Biden stepped down as the Democratic nominee and anointed Kamala Harris as his successor: the 538 national poll average still shows Harris just slightly ahead of Trump, and the state poll averages show the candidates within a percentage point of one another up and down the Blue Wall. The 538 model shows a virtual coin toss election, though it currently gives Trump the slight edge. The New York Times poll tracker shows similar results.

For obvious reasons, this has caused a lot of anxiety among liberals and, in some cases, panic. I’m not immune to this panic; if anything, the fact that I’m about to have a daughter has amplified it. But when I take a step back and look at things dispassionately, I have to admit they look pretty good for Harris. Yes, the polls are tight, but I’ve become persuaded that they are systematically underestimating Democratic strength.

One reason for believing this: the strong likelihood that pollsters are herding. Josh Marshall has a good summary of what that means here:

It’s difficult to prove this is what’s happening, but the uniformity of poll results is pretty suggestive. If Trump and Harris really have equal electoral strength and there is no herding going on, you would expect a normal distribution of poll results converging around a 50/50 split. So maybe one poll shows Trump +3 and another shows Harris +4, but when you average all of them together you get a near tie.

But that’s not what we’re seeing. The polls that give +3 or +4 to one of the candidates in this cycle are either extreme outliers or junk polls from Trumpy firms; the majority of the polls are mysteriously near the exact average. Instead of being bell curve-shaped like you might expect, the distribution of poll results looks like a single neatly stacked column. That’s an implausible result, to put it mildly — unless pollsters are herding.

The Changing Electorate

That raises the question: why are pollsters herding? I suspect they’re being cautious to avoid getting burned. The reputation of the polling industry is at a low ebb after the misfires of 2016 and subsequent cycles, so no one wants to stick their neck out and publish a counterintuitive finding. Relatedly—and I think this is the real nub of the issue—the standard assumptions that pollsters typically bake into their polls no longer hold true, and no plausible alternatives have revealed themselves.

Pollsters are grappling with the same problem that has, since 2016, confronted journalists and Americanists (political scientists who focus their attention on the American political system). All three groups consist largely of people who came of age and entered their respective professions somewhere between the beginning of the Nixon administration and the end of the Obama administration. Their mental model of American politics tends to be rooted in the prevailing assumptions and behaviors of this era.

That mental model no longer works; American political economy has changed so dramatically in just a few years that conventions and truisms from less than a decade ago are now antiquated. The political order that most of us grew up in—call it the neoliberal order—is over.

The more thoughtful and sensitive journalists and academics tend to recognize this and grapple with how they should change their approach. Some reach further back into American history for analogues; others look to other countries that have gone through similar upheavals. Most of the worst Trump-era reporting and commentary has come from people who are unable or unwilling to revise their basic assumptions in the face of a radical new reality.

Pollsters face an even thornier dilemma than reporters and political scientists. They need to develop a plausible composite model of the electorate, including some educated guesses about who counts as a likely voter and about the demographic composition of America’s two major political coalitions. And in doing this, they have very little to go on other than their own intuitions and the past few decades or so of electoral history.

But the end of a political order disrupts political coalitions and voter behavior in unpredictable ways. My theory is that the collapse of the neoliberal political order has damaged pollsters’ ability to identify the shape of America’s political coalitions, and perhaps fundamentally broken their ability to determine who is or is not going to show up on election day.

Consider the following. Trump’s campaign has said it is specifically targeting low-propensity voters in its turnout efforts; in the first post-Roe presidential election, turnout among women could look very different, and a lot more partisan, than in previous years; more broadly, a race between a Black/South Asian woman on one side and a putschist who is running on a platform of mass ethnic cleansing is, needless to say, unprecedented. Meanwhile, as of this week, more than 600,000 early votes in Georgia alone came from people who didn’t vote in 2020, including both new Georgia residents and first-time voters. How many of them were “likely voters” according to pollsters’ most educated guesses?

It’s impossible to say for sure how any of this will shake out. I’m not so certain that even modeling the electorate based off of 2016 and 2020 results will yield accurate polling results. Too much is still in flux; while one political order is dead, I don’t think we’ve arrived at the next one yet. We’re still in an interregnum. My hunch is that the results of the 2024 election will tell us a lot about what America’s post-neoliberal order will look like for the next generation or two.

Good Vibrations

This is not to say that polls are totally useless. But with so much uncertainty in the air, I think we need to supplement any quantitative model of the election with non-quantitative indicators in order to get a more accurate picture. And I think the non-quantitative indicators—the vibes, if you will—suggest that Harris is stronger than the polls indicate.

Consider the reports that the Trump camp’s Elon Musk-led GOTV campaign is in shambles. Or compare the relative crowd sizes for Harris and Trump, and look at their respective endorsement lists: Harris is assembling a popular front that extends from Angela Davis on the left to Dick Cheney on the right, while Trump has now featured Hulk Hogan at both the Republican Convention and his Nuremberg Rally reenactment in Madison Square Garden. Most tellingly, look at the tells from each campaign. Team Harris is acting like it has the initiative, while the Trump campaign’s recent moves betray a stunning lack of confidence: preemptive accusations of voter fraud, frantic efforts to shore up their Republican base, bizarre candidate schedules that seem designed to keep the principal far away from undecided swing state voters.

None of this means that Harris has this locked down; the people around Trump evidently didn’t think he was going to win in 2016 either, and look how that turned out. It’s entirely possible that instead of underestimating the American electorate’s pro-democracy and pro-Roe sentiment, pollsters are missing a hidden groundswell of MAGA fervor. But my gut tells me this is less likely.

My gut is naturally less precise than the polls. But the precision of the polls is false precision. It’s based on assumptions about the electorate that, at the most basic level, are similarly based in pollsters’ gut intuition. That doesn’t mean we should disregard polls entirely, but it is a mistake to treat them like pure mathematics. They’re not astrology, but they’re not physics either.

If either candidate wins by a significantly wider margin than the poll averages indicate, I hope this will prompt some journalists to rethink how they cover elections. Much of the reporting on the 2024 race has been either monomaniacally focused on poll results or guided by whatever reporters think they can glean from crosstabs. I suspect this is because many journalists, especially those who lack any sort of fluency in quantitative methods, think of polling as an “objective” read on the state of the race. If they outsource their judgment to whatever the numbers tell them, no one can accuse them of making editorial decisions based on their own unconscious bias.

But this is itself, of course, a kind of bias. It also misleads readers about the limits of most polling and obscures the sort of background work that goes into producing topline numbers. Reporters and critical readers alike should avail themselves of the whole range of analytical tools—quantitative and qualitative alike—while acknowledging the limitations of them all.