SBSQ #15: Democrats have a "fool me twice" problem
Plus, what might change in the model for 2028. And prediction markets: lucky or good?
Welcome to a special edition of Silver Bulletin Subscriber Questions. We’re using it to tie up loose ends on various reader questions about the election and the model. We’ll run another SBSQ (#16) that avoids election-related questions in a couple of weeks; you can leave questions for that edition in the comments below. Ideally, then, we’ll get back to the regular schedule in which SBSQs run toward the end of the month. There’s some chance we’ll skip the late December/early January SBSQ as I’ll be traveling in Japan and Korea then — although typically I’m bad about taking vacation and actually do a fair amount of writing when I’m on the road.
Another announcement: we’ve set up a customer service address at silver.bulletin.customer.service@gmail.com. This is sort of aspirational in that with a team of just Eli and me, we’re undoubtedly going to miss some questions in busy periods. But right now, these questions have no place to go at all. We’d ask that you please try to stick to questions related to problems with your account, billing, and so forth. Inquires about, for example, business partnerships should be directed to NATE DOT SILVER DOT MEDIA AT GMAIL DOT COM instead.
One more bit of site news: we realize that our election night coverage on Nov. 5 didn’t meet everybody’s expectations, including our own. We tried doing two things, neither of which exactly worked as we were hoping. First, we had an election night model that we turned off after a few hours. We warned in advance that there was a clear possibility of this … but still not ideal.
I do want you to know that we left everything out on the field, including publishing five long stories in the period between Nov. 1 and Nov. 5. We probably needed to budget another full day or two of testing for the election night model, both to look for bugs and to see if it was producing results that passed the smell test given various real-world scenarios. (Election nights are tricky because if something breaks, there’s not really enough time to diagnose and fix it adequately, especially when you’re stressed/panicked.) But that would have meant passing on, say, the Selzer piece or the pollster herding story, which we thought were valuable.
We also had a Subscriber Chat where the signal-to-noise ratio wasn’t really right because there were just three of us (Eli, me, and special guest Matt Glassman) but thousands of reader comments, and I was distracted by other things (including trying to debug the model) for parts of the evening. We were trying to do one too many things.
Anyway, we’re in this for the long run and not for short term profits. So if you signed up specifically for the election night coverage and feel like we let you down, write the customer service account, and we’ll give you a 3-month comp. This is not a blanket amnesty provision, so we will check the dates when you signed up.1
Onto this edition’s questions. I think it’s a strong bunch. The first response is long (and free), so I’m going to put a subscribe button here in case it’s the sort of thing you appreciate.
Are voters irrational to punish parties when the other choice is even worse?
Did prediction markets and other indicators that called the race correctly just get lucky?
How has the model changed over the years? And what changes are in store for 2028?
Are Donald Trump, Joe Biden and Hillary Clinton replacement-level candidates like Kamala Harris?
Was there hidden Harris strength in the battleground states? (No.)
Are voters irrational to punish parties when the other choice is even worse?
In Dearborn, Michigan, the city with the largest share of Arab Americans in the country, Kamala Harris received just 36 percent of the vote, compared to 42 percent for Donald Trump and more than 20 percent combined for Jill Stein, write-ins and other third-party candidates. Vote shares varied heavily by precinct with Harris getting as little as 13 percent in some places. There had been nothing like this in 2020, with Joe Biden winning almost 70 percent of the vote in Dearborn.
Undoubtedly, this was a protest vote against what some Arab Americans saw as Biden and Harris being too favorably disposed toward Israel. A lot of critics have pointed out that Trump is likely to be even more hawkish on Israel, something his cabinet appointments have reinforced. So, weren’t these voters being irrational?
I’m not so sure, actually. Scott Alexander has a long analysis of their predicament at his excellent Astral Star Codex. In principle, voting is a repeated game. Voters concerned about US policy toward Palestine were trying to balance multiple objectives; on the one hand, they might recognize that Trump is worse than Harris on their issue; on the other hand, they wanted to lever Democrats into being more pro-Palestine in both the current and future elections. When you have multiple strategic objectives, the equilibrium in game theory often involves a mixed strategy where you actually randomize your choices:2
How do you preserve ability to bargain with Kamala while continuing to treat her better? You could try to add a stochastic element to your vote. Maybe you roll a dice and vote Kamala for 1-4, and Trump for 5-6. Then you tell Kamala that you’ll vote for her with certainty (rather than only 66% chance) if she does what you want in Palestine. Maybe you could generalize this: figure out what percent of your ideal policy platform Trump gives you, what percent Kamala gives you, and vote with a frequency equal to the ratio of those percentages (eg if Kamala gives you twice as much as Trump, then you vote for Kamala with 66% probability). Then you adjust your frequency up or down as candidates grant your demands.
Alexander discusses various complications with this equilibrium; I won’t cover those because I’m less interested in the abstract questions than the heuristics that voters use in practice.
And in practice, voters tend to care a lot about a sense of feeling betrayed: that a party didn’t live up to its promises (e.g. in the case of Democrats, standing up for oppressed groups) or wasn’t looking out for their interests. In recent years, the Democratic Party’s message has often been simply this: vote for us, or the country gets it, because Trump is even worse. (Actually, political parties have probably always used some form of this message: every election is deemed to be the most important election of one’s lifetime and declared to have existential stakes.)
The tactic doesn’t meet the dictionary definition of “blackmail,” but you can describe it as coercive. And in the face of coercion, people often react by saying “fuck you, I’m not playing along.” One can debate whether this is rational in the long run. But it might be — as I discuss in On the Edge, a lot of nuclear deterrence theory rests on the premise that aggrieved parties will seek revenge: they’ll press the button and launch back at Moscow even after all hope is lost.
Overall, Harris and other Democrats faced a “fool me once, shame on you, fool me twice, shame on me” problem. Voters had seen what Biden and Harris had to offer. They were promised a quick return to normalcy — that was a more tangible promise than Democrats typically offer and one reason that Biden won when Harris and Hillary Clinton didn’t. And they didn’t get it when the summer of 2021 became all about the Delta variant, inflation, and the supply chain crisis: Biden’s approval ratings turned south at that point and never recovered. Then Biden decided to govern like FDR and run again rather than be a caretaker president or a “bridge to the future.”
A related question is whether voters are rational in punishing parties for what they see as a poor performance even if they still prefer that party’s policies going forward. The Republican Lee Zeldin came within 6 percentage points of defeating Kathy Hochul in the 2022 New York gubernatorial race even though New York is a Democratic state (albeit decreasingly so). But the Democratic Party has a long history of electing corrupt and otherwise “problematic” candidates in New York. The decline of what had once been a moderate flank of Republicans in New York — look at what happened to Rudy Giuliani — gives these candidates free rein because of the lack of interparty competition. How can you deter a party from constantly nominating corrupt or craven candidates if you’re never willing to vote for the other side?3 I was one New York voter who thought Zeldin's vote against certifying the 2020 election results was a bridge too far, so I gritted my teeth and voted for Hochul. But I’m not sure if I’m happy about that vote.
Turning back to national politics, there are two times when I’ve felt betrayed by the Indigo Blob, my term for the unofficial alliance between the Democratic Party and the progressive expert class. If you’ve been reading me for a while, you can probably identify them because they’re the two huge fights I’ve had with the left in the past several years. One was with COVID stuff. When the pandemic began, I was one of those people who was like “Welp, we ought to just trust the experts here!”. Many of those experts did a great job under impossible circumstances. But I felt betrayed by a minority who were clearly using the pandemic to advance their political agendas: their utter hypocrisy in endorsing the George Floyd protests after having spent weeks telling everyone to stay home, for instance.4 And then they did profound harm with prolonged school closures.
Then there was Biden’s decision to run again. I thought this actually did present an existential risk because of an 86-year-old president’s questionable decision-making abilities in a crisis. At some point, it even became farcical, with Biden referring to Ukrainian President Volodymyr Zelenskyy as “President Putin” in a press conference meant to reassure the nation about his cognitive fitness. You’d get yelled at by a certain type of Democrat if you didn’t play along and pretend that this was normal.
After the June debate, I was willing to put my foot down; I would have voted third-party5 if Biden had remained on the ballot. Maybe this was intended to “teach the Democratic Party a lesson.” Or maybe it was an emotional reaction more analogous to revenge. But either way, being coerced into voting for a man who clearly wasn’t fit for another four years: sorry, that’s where I was going to tap out.
Other voters may feel betrayed by the Democratic Party for other reasons, particularly too much wokeness, too much immigration, and too much spending. In 2019, Harris ran far to her left. Although many pundits claim that Harris pivoted to the center this year, she was making mostly empty gestures. She backed down from many of her 2019 positions without providing any rationale or proposing much in the way of substantive policies to replace them, or doing anything to offend the various “groups” and nonprofits that dominate the Democratic Party’s policy-making infrastructure. And faced with a decision that did have real consequences — her choice of a running mate — Harris went with what the progressive wing wanted instead of the moderates.
Isn’t Trump also hypocritical and prone toward promising lots of things he can’t deliver? Sure, but his promises are also more tangible: less immigration, less crime, less inflation and most importantly, owning the libs and thumbing his finger in the nose of the establishment through the mere fact of his election.
Meanwhile, Democratic promises are hopelessly abstract. Essentially, the party brand is justice delivered by the enlightened rule of the expert class. The party platform was obsessed with the term “justice,” employing it dozens of times. But what did it look like in practice? Here’s one example:
ENVIRONMENTAL JUSTICE
For too long in this country, too many communities have been denied the basic security and dignity that comes from having clean air, clean water, and functioning infrastructure. Too many parents still lay awake at night worrying what the rain or soil or water or air is doing to their kids. That kind of inequity goes against everything that we stand for as a nation. To finally change that, President Biden’s climate agenda includes the most ambitious environmental justice actions in history, making sure that the people who’ve suffered most from legacy pollution can now share in the benefits.
During his first week in office, the President established the Justice40 initiative, with the goal of directing 40 percent of the benefits of our historic climate and clean energy investments to fenceline communities hit hardest by past pollution. Last year, he gave the program new teeth, with an executive order requiring every federal agency to designate its own Environmental Justice Officer to consider and improve the health impacts of its work on all communities.
Who is asking for this? What is an “Environmental Justice Officer?” Am I supposed to be excited that Biden created a new layer of federal bureaucracy? Does this have anything to do at all with actually reducing pollution? Or justice in any legible sense?
The party has moved away from tangible, material goals. But the abstract goals are hard to fulfill, especially concepts like justice that are seen as a continuous struggle. And then when there’s an emperor-has-no-clothes moment, like with Biden’s obvious decline and half the party refusing to acknowledge it, it calls into question the party’s moral authority as the expert class and reality-based community.
For me, “Trump’s even worse!” worked one last time and I voted for Harris — largely because of January 6 and because Trump, like Biden, is too old. But maybe some of my gut feeling that Trump would win was because I sympathized with voters’ instincts to punish the Democratic Party more than I did in 2016 and 2020. Being willing to take a short-term hit to discourage coercion or punish broken promises is probably a pretty good default, an attitude that’s close enough to rational more often than not.