Hopium comes at a high price

Democrats ignored the polls at their peril.

I finally got a post-election mental reset as my partner and I took a long weekend in the Catskills. It served as a reminder of a few things: 1) that Upstate New York is one of the most beautiful parts of the country, 2) that you ought to pack plenty of water when you go on a 6-mile hike even when it’s cold out … and 3) that it’s time to transition out of “election aftermath mode” here at Silver Bulletin into our medium-term plan, which involves covering a wider variety of topics as we hammer out our long-term plan.

The next two posts planned after this one have nothing to do with the election or even politics at all: one on the NBA and one about Substack. In the next week or so, we’re also going to run an election aftermath SBSQ to tie up loose ends: there’s still time to submit questions. But then we’ll aim to run another SBSQ in early December in which election-related questions are prohibited.¹

But this newsletter? Yep, it’s one more election aftermath post.² In fact, it’s a little bit … actually a lot … of a hot take. 🔥🔥🔥

The polls were … fine

How did the polls do this year? Our official take on that will have to wait until election results are certified and we update our pollster ratings. But I’ve got a piece up at The New York Times with a preliminary answer. And the answer is basically … fine. Not great, but fine:

It may even feel as though we’re Ping-Ponging between radically different futures, never quite certain what lies around the bend. Yet on the whole in 2024, polling did not experience much of a miss and had a reasonable year. Ms. Harris led by only one point in my final national polling average. And Donald Trump led in five of seven key states, albeit incredibly narrowly. The final polling averages were correct in 48 of 50 states.

The final Times/Siena national poll (including third-party candidates) had Mr. Trump one point ahead. There was plenty of data to support a Trump win.

The story explores why people are perpetually so dissatisfied with polls even when they have a really good year like in 2022 — or a fine year like this one. And there are a few obvious answers to that.

One is that they’ve missed in the same direction (underestimating Trump) in three presidential elections in a row. I don’t know how many times this has to happen before we go from saying “well, it’s pretty normal for a coin flip to come up the same way X number of times consecutively” (in three coin tosses, you’ll get a string of either three heads or three tails 25 percent of the time) to “man, we really have a problem here.”

I don’t mean that as a rhetorical question. In 2028, there will be more pressure than ever on pollsters not to publish results that lowball Republicans and not to publish Selzer-esque outliers. But to the extent there was something unique about Trump himself, he won’t be on the ballot — Republicans have shown no tendency to outperform their polls in Trump-era elections when Trump wasn’t running. So if I had guess today, I wouldn’t expect to make any major changes to the model in 2028 after it got the story right in important ways. But we’ll look at things around the margin, like whether national polls deserve more weight relative to state polls or whether the “fundamentals” (which were more favorable to Trump than the polling averages) should be weighted higher. Fortunately, I have 3+ years before making any decisions about this stuff.

Another reason for polling skepticism is simply that we’re in an era of very close elections. Nobody cares much when a candidate who was ahead by 11 points wins by only 7 — or romps to victory by 15. But if Kamala Harris was ahead in Wisconsin by 1 point and loses by 1 instead, that will seem like a big deal — even though the polling miss (2 points) was only half as wide as the 4-point error in the hypothetical election I described.

Maybe the media hasn’t quite internalized the lesson that a 1- or 2- or 3-point polling lead very much isn’t safe — within the normal range of polling error. Though I think the coverage has gotten better. The typical polling story these days is something like this: a headline saying “Harris leads in Michigan”, but then body copy pointing out that it’s only a 2-point lead, well within the margin of error, plus there are other uncertainties in polling. Maybe it’s time to foreground the caveats and soften the headlines, even though “race polling within margin of error” won’t click very well. But overall, I think the mainstream media did a good job portraying this as a close and uncertain race.

Poll denialism is now afflicting Democrats, too

As for the partisan media, though … that’s much less true. Precisely because they were so close, the polls in this election served as a Rorschach test. If you came away with the impression that either Harris or Trump was a heavy favorite, you probably need to re-examine your media diet — because you were being spun.

In 2022, the polling never justified the “red wave” narrative — though, to be fair, the mainstream media participated in the narrative, too, not just conservative outlets. So, if you were reading the news, you’d come away with the impression that Republicans were further ahead than they really were, and therefore that there was a bigger polling miss than there really was.

This year, it was mostly Democratic-friendly outlets that were painting a misleading outlook. Check out some of the tweets from Simon Rosenberg of Hopium Chronicles on the eve of the election, for instance:

Rosenberg isn’t giving you probabilities, but these are strongly-worded tweets that would leave you with the impression that Harris was a heavy favorite. His claim that “there [was] no data right now suggesting he’s winning” was absolute bullshit when highly-rated pollsters like AtlasIntel basically nailed the outcome dead to rights, showing Trump narrowly ahead in all seven key swing states. Meanwhile, Harris’s lead in national polls had all but disappeared, and Trump led in the final state polling averages in enough states to win the Electoral College. (Granted, by just 0.2 points in Pennsylvania in our numbers, for instance.) I don’t know whether prediction markets count as “data,” but they certainly aren’t “voodoo” — they’re more reliable than the other indicators Rosenberg cherry-picked, like Harris having bigger crowds.

Rosenberg and Tom Bonier and other hopium merchants have influence. And people in the progressive media pond who consume the hopium through osmosis have a lot of influence. Take Heather Cox Richardson, the Boston College history professor and the Substack GOAT as in Greatest Of All Time: she has an incredibly popular newsletter that I subscribe to, “Letters From An American”.

Richardson doesn’t purport to be a polling expert or a data maven. But I read her because she often has smart, slightly offbeat things to say — and because she’s an articulate and reliable benchmark for a certain worldview that’s popular in progressive spaces. Not terminally online like younger progressives, but wiser, academic, knowing. Here’s what she wrote on Nov. 6 once the results were in:

These results were a surprise to everyone. Trump is a 78-year-old convicted felon who has been found liable for sexual assault and is currently under indictment in a number of jurisdictions. He refused to leave office peacefully when voters elected President Joe Biden in 2020, instead launching an unprecedented attack on the U.S. Capitol to stop the counting of electoral votes, and said during his campaign that he would be a “dictator” on his first day in office.  

Pollsters thought the race would be very close but showed increasing momentum for Harris, and Harris’s team expressed confidence during the day. By posting on social media—with no evidence—that the voting in Pennsylvania was rigged, Trump himself suggested he expected he would lose the popular vote, at least, as he did in 2016 and 2020. 

But in 2024, it appears a majority of American voters chose to put Trump back into office. 

The results were a “surprise to everyone”? They shouldn’t have been a surprise to anyone, actually, provided they were reading Silver Bulletin or other nonpartisan polling analysis, or mainstream outlets like the Times, and certainly not if they were consuming any right-wing media. And it wasn’t just that the polls had been close: Trump had already been elected president before!

To be fair to Richardson, maybe by “surprise,” she’s taking a broader historical sweep, referring to the results relative to what you might have expected in the halcyon days of mid-November 2020 through early January 2021, when the election was called for Joe Biden, then Pfizer released its favorable COVID vaccine data, then Democrats won two Senate runoffs in Georgia. It seemed for a fleeting moment like there was a “return to normalcy.” But the good times didn’t last long. Not even 24 hours. The very day after the Georgia races on January 5 was … January 6. Then, only a handful of Republicans voted to impeach Trump. Then there was the Delta variant and a new wave of COVID anxiety, a period of extremely high inflation — and Biden’s tanking popularity.

2022 created false hope for Democrats

But then came the 2022 midterms, which were quite good for Democrats relative to the opposition party's typical performance. They also weren’t great. Democrats lost the popular vote for the House by about 3 points and control of the lower chamber. They did come up clutch in most of the major Senate and gubernatorial races, though not all.

People like Rosenberg treated this solid election for Democrats as though it was Waterloo, a death knell for both Republicans and polling aggregators. In a TV appearance just before the election, Rosenberg called FiveThirtyEight’s averages “essentially Republican propaganda.”

Look, I think it’s good to recognize that polling errors can run in either direction. But to a certain class of Democratic commentators, 2022 became proof that polls were now biased against Democrats, even though the polls had been historically accurate in 2022 — a fact Rosenberg literally refused to accept in a contentious New Yorker interview — and had been significantly biased against Trump in 2016 and 2020.

The Rosenbergs of the world also had influence in the White House. Prior to the midterms, Biden’s decision to run for re-election was treated as an open question — to the point where Gavin Newsom was basically running 2024 campaign ads for a presumed general election matchup against Ron DeSantis.³ After the midterms, the Biden camp concluded that the polls couldn’t be trusted and talk of a retirement or a primary challenge died down.

Commentators who shrugged off voter concerns about Biden’s age — and didn’t repent — can’t be trusted

As late as April 2024, Rosenberg asserted that Biden’s age — an incredibly persistent problem in the polls and an incredibly obvious one to anyone who watched his public appearances or observed his lack thereof — was actually “also an asset for him.” The #Resistance historian set — Richardson and Timothy Snyder and, yes, Allan Lichtman — contributed to the chorus, while liberal media critics compared the media’s focus on Biden’s age to that on Hillary Clinton’s emails in 2016.

The Biden-Trump debate in June ought to have been a reality check: about as decisive proof as you can possibly get of having been wrong, outside of an election itself. Some commentators like Matt Yglesias quickly acknowledged their error. But among the #Resistance crew, there was no self-reflection: no recognition that if they’d been so wrong about Biden, their epistemics might be poor and they probably needed to reconsider what else they might be wrong about. Instead, almost to a person — once more rational voices in the Democratic Party prevailed and Biden was shoved aside — this group seamlessly and unapologetically transitioned from insisting it was a huge mistake to oust Biden to being equally certain that Harris would win despite her mediocre electoral track record. 

As I wrote in the Times story, this wasn’t the only way in which Democrats ignored public opinion data:

And on an array of issues, Democrats downplayed inconvenient polling truths — about the public’s dislike for inflation and lax immigration policies, about Mr. Biden’s age, about Ms. Harris’s middling popularity and, yes, about the cultural phenomenon that we call wokeness.

Yes, I’m lumping somewhat unlike things together. The Federal Reserve and Biden, through massive stimulus spending made possible by those runoff wins in Georgia, ran the economy “hot,” which produced a rapid recovery in the labor market but at the cost of high inflation. That might have been justified on a utilitarian basis — more jobs are more good! — but it proved to be electoral poison, as the White House was warned about at the time by Joe Manchin and Larry Summers and the sorts of centrists that #Resistance progressives love to hate.

Democrats’ challenges on immigration were also baked in early, with Harris and other candidates racing to their left in the 2019/2020 primary and then in the first days of the new administration. I don’t want to dwell on it, and I make a specific point in this newsletter of not talking about trans issues any more than I absolutely have to. In general, I think they’re less of an electoral liability for Democrats than people often assume. But there’s a reason that you couldn’t escape the “Kamala’s for they/them” ads. Harris promising taxpayer-funded gender transition surgery even for “those in prison and immigration detention” played into basically the Holy Trinity of voter anxieties: immigration and government overspending⁴ and too much wokeness.

These were foreseeable errors, in other words. And the common theme was Democratic influencers ignoring public opinion data, which for all its faults, did provide plenty of hints about what was coming:

Polls still provide important hints, leads and hypotheses. They were basically right that Democrats’ dominance among minority groups was waning, overestimating the swing among Black voters but understating it among Hispanic voters. Asian Americans, Native Americans and the oft forgotten group of voters who identify their race as “other” also shifted toward Mr. Trump.

They were right that Democrats would experience a significant erosion among younger voters. Ms. Harris won voters ages 18 to 29 by just four points, according to the A.P. VoteCast survey — down from Joe Biden’s 25 points in 2020. But the shifts were much bigger among young voters who didn’t attend college, especially men.

Some of the biggest shifts of all came in large cities like New York. Polls showed clues about this, too. A Times/Siena poll late in the campaign had Ms. Harris winning New York City by 39 points, down from Mr. Biden’s 53 points in 2020. Her actual margin in the five boroughs? Just 38 points, although some votes remain to be counted.

And if polls did have a systemic bias against Trump, that was all the more reason to proceed carefully and dispel with the fiction that you were the new FDR.

If in October 2028, JD Vance is down by 3 points to Gretchen Whitmer in the swing states and Republicans unskew the polls, I can’t say I’d really blame them — even though Trump’s poll-outperforming magic might not transfer to another candidate, and pollsters will likely be cautious about showing meaningful Democratic leads. But it’s time for Democrats to throw the rest of their hopium pills in the toilet.

1. It’s probably best to hold your non-election questions for now so they don’t get buried.

2. I don’t mean to imply it’s the very last one. I want to start digging into some data at the county level, for instance.

3. Another assumption that didn’t age well.

4. Even though the overall expenditure would be trivial in the context of the federal budget.