Editor’s Note: Is anything ailing, torturing, or nagging at you? Are you beset by existential worries? Every Tuesday, James Parker tackles readers’ questions. Tell him about your lifelong or in-the-moment problems at dearjames@theatlantic.com.

Don’t want to miss a single column? Sign up to get “Dear James” in your inbox.

Dear James,

I have a group of friends who used to work together, and our friendship has happily continued even though we’re no longer at the same company. At a recent get-together, however, I accidentally found out that there’s a group text thread, and I’m not on it. Based on the group chat’s title, I believe that it started when we all worked together, when my friends were involved in projects that didn’t include me. As a fully grown woman, I feel like this shouldn’t bother me. I’m still invited to get-togethers. And I know, logically, that it’s not like I’m being purposefully kept off the chat. But I can’t help feeling sad. Growing up, I often felt as if I didn’t fit in, so I’m sensitive about being forgotten or left out. I’m so happy to have work friends who have become real friends, and I have no intention of saying anything. But I’m hurt. How can I soften the blow of feeling excluded?

Dear Reader,

This is why I have a flip phone.

This whole bloody, tingling plexus of hyper-connectedness from which—if it fails us, if the right person doesn’t instantly get back to us saying precisely the right thing—we suddenly feel ourselves expelled like Lucifer from heaven, kicked out, falling, alone, the crystal towers receding above us as we plummet through the abyss with our feathers starting to smoke and crackle … It blows my mind. We seem to have invented a completely new way to be miserable. Or a completely new way to get in touch with a very old way of being miserable, which is what your letter is about.

The pain of being left out: How to manage it? I applaud the quality of your self-talk. You’re telling yourself that you’re an adult, that no one’s trying to hurt you, that there’s a non-catastrophic explanation for why you’re not on this group thread. And I know that sometimes none of that works—you just have to sit there with that sad-child feeling. It will pass.

Last thought: Maybe take a moment to reimagine your friendship with these people, to see it as a set of individual relationships rather than as a single, mainline, all-or-nothing attachment to the group-as-group. Because groups are always leaving people out, in one way or another. They can’t help it. It’s what makes them groups.

Pecking out texts to myself,

James

Dear James,

My 24-year-old stepdaughter, who lives away from home and whom we rarely see, asked to be included on our holiday card. We pay for all her expenses—including the therapy she goes to many times per month—except for a small amount of rent and the maintenance on the new car that we bought for her. Yet she is disrespectful and rude to us. She is also obese and slovenly. I have no desire to include her on our beautiful holiday card. Should I reconsider?

Dear Reader,

Yes, you should reconsider. Your stepdaughter is reaching out, asking to be included, and the fact that you find her too unbeautiful to be photographed alongside the rest of your family—well, that might have something to do with her being in therapy. You’ve got an opportunity here to improve your relationship with her. Take it!

Sticking up for stepdaughters,

James

By submitting a letter, you are agreeing to let The Atlantic use it in part or in full, and we may edit it for length and/or clarity.

Jonathan Zittrain breaks ChatGPT: If you ask it a question for which my name is the answer, the chatbot goes from loquacious companion to something as cryptic as Microsoft Windows’ blue screen of death.

Anytime ChatGPT would normally utter my name in the course of conversation, it halts with a glaring “I’m unable to produce a response,” sometimes mid-sentence or even mid-word. When I asked who the founders of the Berkman Klein Center for Internet & Society are (I’m one of them), it brought up two colleagues but left me out. When pressed, it started up again, and then: zap.

The behavior seemed to be coarsely tacked on to the last step of ChatGPT’s output rather than innate to the model. After ChatGPT has figured out what it’s going to say, a separate filter appears to release a guillotine. The reason some observers have surmised that it’s separate is because GPT runs fine if it includes my middle initial or if it’s prompted to substitute a word such as banana for my name, and because there can even be inconsistent timing to it: Below, for example, GPT appears to first stop talking before it would naturally say my name; directly after, it manages to get a couple of syllables out before it stops. So it’s like having a referee who blows the whistle on a foul slightly before, during, or after a player has acted out.

For a long time, people have observed that beyond being “unable to produce a response,” GPT can at times proactively revise a response moments after it’s written whatever it’s said. The speculation here is that to delay every single response by GPT while it’s being double-checked for safety could unduly slow it down, when most questions and answers are totally anodyne. So instead of making everyone wait to go through TSA before heading to their gate, metal detectors might just be scattered around the airport, ready to pull someone back for a screening if they trigger something while passing the air-side food court.

The personal-name guillotine seemed a curiosity when my students first brought it to my attention at least a year ago. (They’d noticed it after a class session on how chatbots are trained and steered.) But now it’s kicked off a minor news cycle thanks to a viral social-media post discussing the phenomenon. (ChatGPT has the same issue with at least a handful of other names.) OpenAI is one of several supporters of a new public data initiative at the Harvard Law School Library, which I direct, and I’ve met a number of OpenAI engineers and policy makers at academic workshops. (The Atlantic this year entered into a corporate partnership with OpenAI.) So I reached out to them to ask about the odd name glitch. Here’s what they told me: There are a tiny number of names that ChatGPT treats this way, which explains why so few have been found. Names may be omitted from ChatGPT either because of privacy requests or to avoid persistent hallucinations by the AI.

The company wouldn’t talk about specific cases aside from my own, but online sleuths have speculated about what the forbidden names might have in common. For example, Guido Scorza is an Italian regulator who has publicized his requests to OpenAI to block ChatGPT from producing content using his personal information. His name does not appear in GPT responses. Neither does Jonathan Turley’s name; he is a George Washington University law professor who wrote last year that ChatGPT had falsely accused him of sexual harassment.

ChatGPT’s abrupt refusal to answer requests—the ungainly guillotine—was the result of a patch made in early 2023, shortly after the program launched and became unexpectedly popular. That patch lives on largely unmodified, the way chunks of ancient versions of Windows, including that blue screen of death, still occasionally poke out of today’s PCs. OpenAI told me that building something more refined is on its to-do list.

As for me, I never objected to anything about how GPT treats my name. Apparently, I was among a few professors whose names were spot-checked by the company around 2023, and whatever fabrications the spot-checker saw persuaded them to add me to the forbidden-names list. OpenAI separately told The New York Times that the name that had started it all—David Mayer—had been added mistakenly. And indeed, the guillotine no longer falls for that one.

For such an inelegant behavior to be in chatbots as widespread and popular as GPT is a blunt reminder of two larger, seemingly contrary phenomena. First, these models are profoundly unpredictable: Even slightly changed prompts or prior conversational history can produce wildly differing results, and it’s hard for anyone to predict just what the models will say in a given instance. So the only way to really excise a particular word is to apply a coarse filter like the one we see here. Second, model makers still can and do effectively shape in all sorts of ways how their chatbots behave.

To a first approximation, large language models produce a Forrest Gump–ian box of chocolates: You never know what you’re going to get. To form their answers, these LLMs rely on pretraining that metaphorically entails putting trillions of word fragments from existing texts, such as books and websites, into a large blender and coarsely mixing them. Eventually, this process maps how words relate to other words. When done right, the resulting models will merrily generate lots of coherent text or programming code when prompted.

The way that LLMs make sense of the world is similar to the way their forebears—online search engines—peruse the web in order to return relevant results when prompted with a few search terms. First they scrape as much of the web as possible; then they analyze how sites link to one another, along with other factors, to get a sense of what’s relevant and what’s not. Neither search engines nor AI models promise truth or accuracy. Instead, they simply offer a window into some nanoscopic subset of what they encountered during their training or scraping. In the case of AIs, there is usually not even an identifiable chunk of text that’s being parroted—just a smoothie distilled from an unthinkably large number of ingredients.

For Google Search, this means that, historically, Google wasn’t asked to take responsibility for the truth or accuracy of whatever might come up as the top hit. In 2004, when a search on the word Jew produced an anti-Semitic site as the first result, Google declined to change anything. “We find this result offensive, but the objectivity of our ranking function prevents us from making any changes,” a spokesperson said at the time. The Anti-Defamation League backed up the decision: “The ranking of … hate sites is in no way due to a conscious choice by Google, but solely is a result of this automated system of ranking.” Sometimes the chocolate box just offers up an awful liquor-filled one.

The box-of-chocolates approach has come under much more pressure since then, as misleading or offensive results have come to be seen more and more as dangerous rather than merely quirky or momentarily regrettable. I’ve called this a shift from a “rights” perspective (in which people would rather avoid censoring technology unless it behaves in an obviously illegal way) to a “public health” one, where people’s casual reliance on modern tech to shape their worldview appears to have deepened, making “bad” results more powerful.

Indeed, over time, web intermediaries have shifted from being impersonal academic-style research engines to being AI constant companions and “copilots” ready to interact in conversational language. The author and web-comic creator Randall Munroe has called the latter kind of shift a move from “tool” to “friend.” If we’re in thrall to an indefatigable, benevolent-sounding robot friend, we’re at risk of being steered the wrong way if the friend (or its maker, or anyone who can pressure that maker) has an ulterior agenda. All of these shifts, in turn, have led some observers and regulators to prioritize harm avoidance over unfettered expression.

That’s why it makes sense that Google Search and other search engines have become much more active in curating what they say, not through search-result links but ex cathedra, such as through “knowledge panels” that present written summaries alongside links on common topics. Those automatically generated panels, which have been around for more than a decade, were the online precursors to the AI chatbots we see today. Modern AI-model makers, when pushed about bad outputs, still lean on the idea that their job is simply to produce coherent text, and that users should double-check anything the bots say—much the way that search engines don’t vouch for the truth behind their search results, even if they have an obvious incentive to get things right where there is consensus about what is right. So although AI companies disclaim accuracy generally, they, as with search engines’ knowledge panels, have also worked to keep chatbot behavior within certain bounds, and not just to prevent the production of something illegal.

[Read: The GPT era is already ending]

One way model makers influence the chocolates in the box is through “fine-tuning” their models. They tune their chatbots to behave in a chatty and helpful way, for instance, and then try to make them unhelpful in certain situations—for instance, not creating violent content when asked by a user. Model makers do this by drawing in experts in cybersecurity, bio-risk, and misinformation while the technology is still in the lab and having them get the models to generate answers that the experts would declare unsafe. The experts then affirm alternative answers that are safer, in the hopes that the deployed model will give those new and better answers to a range of similar queries that previously would have produced potentially dangerous ones.

In addition to being fine-tuned, AI models are given some quiet instructions—a “system prompt” distinct from the user’s prompt—as they’re deployed and before you interact with them. The system prompt tries to keep the models on a reasonable path, as defined by the model maker or downstream integrator. OpenAI’s technology is used in Microsoft Bing, for example, in which case Microsoft may provide those instructions. These prompts are usually not shared with the public, though they can be unreliably extracted by enterprising users: This might be the one used by X’s Grok, and last year, a researcher appeared to have gotten Bing to cough up its system prompt. A car-dealership sales assistant or any other custom GPT may have separate or additional ones.

These days, models might have conversations with themselves or with another model when they’re running, in order to self-prompt to double-check facts or otherwise make a plan for a more thorough answer than they’d give without such extra contemplation. That internal chain of thought is typically not shown to the user—perhaps in part to allow the model to think socially awkward or forbidden thoughts on the way to arriving at a more sound answer.

So the hocus-pocus of GPT halting on my name is a rare but conspicuous leaf on a much larger tree of model control. And although some (but apparently not all) of that steering is generally acknowledged in succinct model cards, the many individual instances of intervention by model makers, including extensive fine-tuning, are not disclosed, just as the system prompts typically aren’t. They should be, because these can represent social and moral judgments rather than simple technical ones. (There are ways to implement safeguards alongside disclosure to stop adversaries from wrongly exploiting them.) For example, the Berkman Klein Center’s Lumen database has long served as a unique near-real-time repository of changes made to Google Search because of legal demands for copyright and some other issues (but not yet for privacy, given the complications there).

When people ask a chatbot what happened in Tiananmen Square in 1989, there’s no telling if the answer they get is unrefined the way the old Google Search used to be or if it’s been altered either because of its maker’s own desire to correct inaccuracies or because the chatbot’s maker came under pressure from the Chinese government to ensure that only the official account of events is broached. (At the moment, ChatGPT, Grok, and Anthropic’s Claude offer straightforward accounts of the massacre, at least to me—answers could in theory vary by person or region.)

As these models enter and affect daily life in ways both overt and subtle, it’s not desirable for those who build models to also be the models’ quiet arbiters of truth, whether on their own initiative or under duress from those who wish to influence what the models say. If there end up being only two or three foundation models offering singular narratives, with every user’s AI-bot interaction passing through those models or a white-label franchise of same, we need a much more public-facing process around how what they say will be intentionally shaped, and an independent record of the choices being made. Perhaps we’ll see lots of models in mainstream use, including open-source ones in many variants—in which case bad answers will be harder to correct in one place, while any given bad answer will be seen as less oracular and thus less harmful.

Right now, as model makers have vied for mass public use and acceptance, we’re seeing a necessarily seat-of-the-pants build-out of fascinating new tech. There’s rapid deployment and use without legitimating frameworks for how the exquisitely reasonable-sounding, oracularly treated declarations of our AI companions should be limited. Those frameworks aren’t easy, and to be legitimating, they can’t be unilaterally adopted by the companies. It’s hard work we all have to contribute to. In the meantime, the solution isn’t to simply let them blather, sometimes unpredictably, sometimes quietly guided, with fine print noting that results may not be true. People will rely on what their AI friends say, disclaimers notwithstanding, as the television commentator Ana Navarro-Cárdenas did when sharing a list of relatives pardoned by U.S. presidents across history, blithely including Woodrow Wilson’s brother-in-law “Hunter deButts,” whom ChatGPT had made up out of whole cloth.

I figure that’s a name more suited to the stop-the-presses guillotine than mine.

Subscribe here: Apple Podcasts | Spotify | YouTube | Overcast | Pocket Casts

If you got an extra $2,500 after filing your taxes, who would you thank? The president? Congress? Your governor? How about H&R Block?

One of the biggest problems facing democracy is whether voters can discern and reward policy makers for good policy and, in reverse, punish them for bad policy. The research here has been mixed, and the Democratic Party’s performance in the 2024 presidential election has led some to doubt whether the feedback loops necessary for good policy—and a healthy democracy—even exist.

This episode of Good on Paper pushes back against the pessimists. Interpreting signals from voters is complicated, and so much is contingent on which issues are salient when they head to the ballot box. But the political scientist Hunter Rendleman’s research indicates that when states rolled out Earned Income Tax Credit programs—a benefit for working-class Americans—voters rewarded governors who implemented the policy with higher vote shares and approval ratings.

“I think I’m an optimist on sophistication,” Hunter told me. “I think a lot of times political scientists are a bit pessimistic on individuals’ capacities to actually know what’s going on to them because it is quite complicated. But we don’t often set up our analyses or studies in a way to give voters the benefit of the doubt.”

The following is a transcript of the episode:

[Music]

Jerusalem Demsas: In 2021, Joe Biden gave the Teamsters union exactly what it was looking for: The American Rescue Plan, the $1.9 trillion COVID-19 stimulus package, included $36 billion to prevent cuts to union pensions. Tens of thousands of workers and retirees in Michigan, Ohio, and Wisconsin received aid.

But when election season came around, Teamsters President Sean O’Brien gave a prime-time address at the RNC praising Donald Trump and J. D. Vance, infuriating White House aides who told The Washington Post that it was a “betrayal of the administration’s support” for union workers.

That betrayal rested on a concept known as “deliverism.” Deliverism refers to the idea that if you deliver good and important policy wins, then people will reward you for them at the polls. As The Prospect’s David Dayen wrote in 2021, Barack Obama’s reelection run contained a great example of this: the phrase, “Bin Laden is dead and General Motors is alive.”

But this line of thinking has come under much scrutiny following Kamala Harris’s loss to Donald Trump. As one article from Democracy Journal put it, “If spending trillions of dollars to improve people’s lives isn’t a fair test of ‘deliverism,’ then what is?”

[Music]

Demsas: My name’s Jerusalem Demsas. I’m a staff writer here at The Atlantic, and this is Good on Paper, a policy show that questions what we really know about popular narratives.

My guest today is Hunter Rendleman. She’s a political scientist soon to join the faculty at UC Berkeley and has published a paper at the American Political Science Review called, “Do Government Benefits Affect Officeholders’ Electoral Fortunes?”

Alongside her co-author, Hunter looks at one important federal tax credit for working-class Americans: the earned-income tax credit, or the EITC. Data in 2018 showed the EITC lifted about 5.6 million people out of poverty, most of them kids.

Hunter’s paper asks: Did voters reward governors who expanded this social benefit?

Hunter, welcome to the show.

Hunter Rendleman: Thanks so much for having me.

Demsas: I’m excited to have you here. You’ve written a really interesting new paper, and it’s about something that I really want to be true, so I’m trying to temper my enthusiasm. But before we get into the details of your paper, I’m hoping you can just step back for us here and explain: What is a policy-feedback loop?

Rendleman: A policy-feedback loop is something where you implement a policy, and when that policy is implemented, it creates a new constituency of individuals that are invested in making sure that policy doesn’t go away. And that constituency of individuals can be motivated for what’s usually thought of for two reasons: either for resource reasons—so basically, you’ve given this group some amount of money, and they become mobilized and motivated around making sure that money doesn’t go away—or it does something to how these people perceive and interpret government.

So for example, in the case of the GI Bill, it allowed these veterans to go to get university education. This changed how they saw themselves vis-à-vis politics, vis- à-vis society. And this led to them changing how they mobilized and then how they participated in government moving forward.

Demsas: And so a lot has to go right for a policy-feedback loop to work, right? People have to understand why a policy happened and who’s responsible. And is it salient to voters come Election Day? And then, do politicians themselves understand that voters are rewarding or punishing them?

So let’s take something like Medicaid expansion. Are voters understanding who’s responsible for that? Is the media giving them that information? Do they know which politicians are actually to blame? Is it federal politicians? Maybe they’re the right people to attribute it to. Maybe it’s state politicians. If it’s state, is it your state legislature? Is it your governor? I mean, a lot of that has to go right. And before we get into your study, again, can you just lay the groundwork for us about the political-science literature in this space?

My read is that there’s a lot of pessimism about the possibility for specific policies to generate salient and durable electoral shifts. There’s this book by Christopher Achen and Larry Bartels called Democracy for Realists, and it argues against the idea that policies are due to voter preferences or that voters are even engaged in issue voting, right? They point to those two problems I already mentioned about the difficulty in identifying who’s responsible for changes in your own welfare. But how dominant is this view in the academy?

Rendleman: It really depends who you are in the academy. So if you’re an economist, for example, you often believe in the notion of pocketbook voting, and I think that’s the most classical notion of how people respond to policies. If they’re doing better today than they were a year ago, they’re going to vote for the incumbent that made their pocketbooks, their bank accounts fatter, basically.

Since that happened—and that’s, you know, [Anthony] Downs and how he viewed the world. Since that work was published, there was a lot of pushback on the part of the academy, saying, like, Okay. But wait—we know that people vote on the basis of vibes, for lack of a better word. We know they vote on the basis of identity. And so even if they are not necessarily doing as well as they were yesterday, they still might vote for the incumbent. And then, even further, if the incumbent does do something for them, it depends on who that incumbent is to generate this sort of feedback loop that we’re talking about. And so I’ll say that the academy is extremely mixed with regards to how they view this proper attribution process going.

Demsas: What about you before you started doing your study? What did you think?

Rendleman: When I started the study—the only reason why I wrote the study is because I had this firsthand experience about the policy that I was studying, the EITC.

Demsas: Can you say what that is?

Rendleman: Yeah. The earned-income tax credit—it is a tax credit remitted to lower-income individuals based off of how much they earned in the past tax year. And basically what it is, is this potentially sizable sum of money given to individuals that will defray part of their tax burden, and it’s refundable. At the federal level, it’s refundable. So you offset part of your tax burden, and whatever you get in remainder, you get to take home and as a check.

So this is an incredibly popular policy. It has had bipartisan support since it was first created and implemented. But there is a lot of political-science work that says this actually shouldn’t have an effect, and it doesn’t have an effect. And the reason why people argue this—and specifically, it’s folks like Suzanne Mettler in her book called The Submerged State, but also folks like [R. Douglas] Arnold, who studies something called “policy traceability”—is that no one really knows where the EITC comes from when they get it.

So you get your check on tax day or after tax day, and you’re like, Okay. Cool. I got $500. I did my taxes. It could be for a whole host of different things, like, Taxes are complicated. Maybe I just paid too much, and so I’m getting money back. Or it could be some special policy, or it could be: I hired a fancy accountant, or I went to H&R Block for the first time, and I got this money.

And, in fact, there is a great book called It’s Not Like I’m Poor, which basically says the design of the tax-preparation industry submerges these policies. It makes it really difficult for normal people to know whether or not they’re getting a given tax policy from the government.

But to get back to the topic of my paper—when I was in college, I did volunteer work for the IRS because I was a nerd. And I was helping people out with claiming their EITCs. And people were like, Wow. This is a lot of money. I’m like, Yeah. It’s from the government. And they’re like, Wow. I always knew this guy was really good for us, and they would talk about whatever politician was in office at the time.

Demsas: And who are they talking about? Are they talking about, like, presidential level? Or are they talking about, like, governor? Who are they talking about?

Rendleman: Presidential level. Like, you know, this broad attribution to the president.

Demsas: Thanks, Obama.

Rendleman: Thanks, Obama, basically. But in the positive way, like, Actually, thanks, Obama. And when I got to graduate school, and I read this literature about traceability and policy feedbacks and, specifically, this work on the EITC, I said, That doesn’t make any sense. People sort of know—or, oftentimes, people can know—about this policy, but maybe we’re just not measuring it in the right way.

Demsas: But when you heard people say, Thanks, Obama, do you think that Obama is the reason why the EITC—or he should be credited for the EITC checks people are getting?

Rendleman: No. So government, as you know, is complicated. So it’s an interesting question because there was an EITC expansion under the Obama administration, but was he responsible for it? No. It was the legislators that worked hard to bargain with other legislators in order to get a new policy through.

And the topic of my paper, which are state-level EITCs, it’s the same thing: Is it the governor that is doing this, or is it state legislators, or is it someone else entirely, like nonprofits that went and lobbied to push for certain policies?

I guess going back to your initial question—this idea of attribution is super difficult, and it’s not really clear, from a normative standpoint, if you necessarily need that perfect one-to-one attribution or if it’s okay that we have a diffuse attribution. So in general, Are we rewarding the party that did this for me, or are we rewarding the individual?

There’s a paper by Justin de Benedictis-Kessner and Chris Warshaw, which basically studies this question: When people do better in their wages—I believe that’s the study—is it the individual politician responsible, like what politician is responsible for wages, or is it just this diffuse party effect? And they find that salient politicians really benefit. Like, governors and presidents and House legislators—they really benefit when people do better. But in general, the party in office tends to do better when people are doing better off economically.

Demsas: Well, so in an ideal world, though, the way that we would want to see policy-feedback loops work is: If a policy durably benefits you, then you are going to be able to understand who is responsible for that. That’s, obviously, a difficult question, but you could identify the arm of government that was most responsible, or at least the arms of government that were involved in this, and then you would reward them at the ballot box however you could.

But you’re saying, if you feel like this diffuse version, this one where parties get benefited, is also an optimal way for feedback loops to work.

Rendleman: I guess I have difficulty with this idea of the optimality of the feedback loop, because it’s really sort of voter psychology. Do I want voters to respond to the policy landscape that they are facing? Yes. I think that is the sign of a healthy and thriving democratic system.

We want people to have the information to understand what’s happening to them, and for them to translate that information into decision making at the ballot box. If it’s to a particular individual they can actually trace—I know politician X did Y—that’s great. I think an alternative, perhaps more realistic view is: Can I detect if I did better, and the party that is in office is probably associated with that policy? That’s also good.

Demsas: I’m going to get into your study now. You and your co-author conduct a study looking at states’ adoption of EITC programs from 1992 to 2018. Can you just walk us through what you actually did, what you were trying to measure, and how?

Rendleman: Yeah. We went at it from, as I was saying, this very pocketbook-voting type of perspective, where we investigate whether the introduction of a given state level in earned-income tax credit leads to the incumbent governors—and generally, the incumbent party—being rewarded on Election Day.

So what we do is: It’s called a staggered difference-in-differences design, where you see policy getting introduced in year T, and then we look at the nearest election to year T and see if there is an increase in voting for the incumbents at that time. And so we did this at the county level for vote shares for governor, and then we also did an additional study at the back end of the paper looking at individuals’ approval levels for governor using survey data.

And just to explain the results a little bit: We find quite small effects at the ballot box, which is expected. The EITC doesn’t actually benefit a lot of people on a percentage basis in a given county. We do see in places where more people tend to claim the federal EITC, you get a bit larger increases at the ballot box. So effectively: Places that are poorer, you see bigger swings in favor of the incumbent governor.

But what was also interesting is: When we used the survey data, and we’re able to actually see, Okay, given the demographics that a given person ticked off in the survey—you know, their marital status, their income, how many children they have—we can back out who is a likely eligible person for the EITC. You, in particular, see big boosts in approval amongst those people. So not only do we see this rather small increase on Election Day, but you also see that it has perhaps this more durable psychological effect, or at least attitudinal effect, on the people that the policy is helping.

Demsas: Can we start with why you chose governors? I would imagine the state legislatures are likely to be more responsible for this. Why didn’t you measure what was happening there?

Rendleman: Part of it was data. Vote shares for governors are far easier to come by, and they are still involved in the policy process. They have to sign off for the budget. Like, Okay, this is a good budget. And at the time, it was more difficult for us to detect state legislative election votes. If we were to redo the study, probably we wouldn’t try to investigate that, as well.

You do see on the attitudinal side, we look at different types of politicians. So we look at the governor, we look at the president, we look at senators, and we look at House representatives. And we don’t see the same sort of effect that we do for governors. And so we took that as a good sign that people are properly attributing to at least the state level when they are receiving their EITCs.

Demsas: So when states adopt the EITC, what do we see happening to governors’ vote share?

Rendleman: Yeah. What we see is: In counties where a state EITC has just been implemented, you see a two-percentage-point increase in favor of the incumbent governor, the governor that passed the policy.

Demsas: That’s a pretty big number. Is that surprising to you?

Rendleman: We took that as a pretty small number, given the wide-ranging estimates in the economic-voting and policy-feedback literature.

Demsas: Well, EITC is one of, like, thousands of policies that are implemented. The idea that you would detect something that’s going to a minority of individuals—a very small minority of individuals in the state—and that that would increase the governor’s vote share by two percentage points, I mean, that can be the margin of victory, right?

Rendleman: It could be. Yeah. So when you really parse the effects further, you find that these effects are concentrated in only the first year after the EITC is implemented. And it is way concentrated in places where there are more claimants, so as I was saying before. But also, it’s concentrated amongst Republican governors.

So Republicans tend to get a bigger boost from passing these sorts of policies than Democrats. And I think that’s also a pretty significant part of the story, because if these people who tend to be lower income—and so oftentimes we associate them with voting with the left, or at least voting for Democrats—we’re seeing them potentially switching their votes. Of course, the study itself can’t tell us about whether a given person switched their vote from one thing to another, but we do see, within these counties that would plausibly vote more blue, this increase in support for Republicans and stuff.

Demsas: Can you talk us through the story of what’s going on here? A governor decides to implement a more generous EITC expansion in their state. And then what’s actually happening on the ground? I can imagine there are a bunch of different stories you could tell. There’s a story where the people who are getting the EITC become more likely to vote and more likely to do political participation because they have more money. And we see that happening, in general, with higher-income folks accessing the ballot box more.

Or you could imagine that people are really excited about the EITC. You could also imagine that the EITC is covarying with a bunch of other welfare policies or anti-poverty policies. What’s actually the most plausible story to you about what’s happening here?

Rendleman: These are popular policies when they are passed. And they’re policies that governors would plausibly want to run their elections on, or their campaigns on, rather. And so it’s—I don’t know—October, and you’re trying to get a new sector of the vote, and you say, Okay. What do I tell them that I did?

Well, you tell them that you passed something like the EITC, and they’re like, Okay. Well, that’s wonderful.

So that’s probably going on—that because these policies are popular, and people would probably want to reward you for passing something like this, you talk about it. I think Gavin Newsom—California passed an EITC in our sample—he ran pretty strongly on the fact that he rolled out this EITC. And he also campaigned pretty strongly on increasing the EITC, as well, or making a very generous one. So that’s going on.

You also have the entry of these sorts of nonprofit and lobbying types of organizations that say, Okay, this is sort of a workfarey type of policy that we like to see. It’s something that incentivizes people to work. There’s a whole bunch of economic studies that talk about the downstream economic benefits of implementing EITCs. Maybe they come in, and they say, Yeah. We want you to keep the EITC, and so we’ll help you out. We’ll throw you a bit of extra cash to campaign more in these districts. So that could be going on.

I don’t know, and I doubt it’s a resource story, in the sense of: The amount of money given to an individual because of the EITC is enough to dramatically change the amount of leisure time that they have, which can allow them to participate more in government, or in politics, rather.

So I think that’s a story that’s often told for Social Security: Older folks that are impoverished—they get this potentially transformative amount of money. It allows them to not have to think about maybe getting a part-time job after they retire. Instead, they can have more time to themselves and potentially campaign. I don’t think that’s what’s going on for the EITC. I don’t know.

It’s a great windfall. It’s, like, $200-$300, which is great. There have been economic studies that talk about people spending that money often on more durable goods. So you get your EITC. You buy yourself a new washing machine, potentially, which improves your quality of life significantly. But it’s not something that’s going to durably transform how you approach life.

Demsas: But it’s bigger for people with children, right? So you could have—for childless adults—you could have maybe $500, something like that. But you can get up to $6,000, or even higher than that for families with three or more kids. So that could be pretty big. And one thing I think that was interesting in your study is that you don’t actually see families with children having stronger effects. What do you think is going on there?

Rendleman: Basically, what we’re doing in that analysis is: We’re seeing if there is a difference in the effect size between, generally, people that are eligible to receive the state EITC versus people that are eligible and have children. We don’t see a difference between those two groups. We, overall, see an increase in people’s approval for the governor. We just don’t see a distinct increase on top of that for people with kids.

Demsas: But that’s surprising, right? You would expect, you know, one person’s getting $500, and another person’s getting $6,000. It’s a very, very different outcome.

Rendleman: It’s a very different outcome. It’s also the case that when you also have kids, you’re eligible for more things, on average. And so maybe it’s harder for you to do the attribution, or you already know that you’re receiving more money, on average, and so maybe this extra cash is not going to be so revolutionary to you.

Demsas: But I guess my question is: When you’re going from $0 to $500 or $700, you’re detecting a change in the governor’s vote share in that county where you see more exposure. And so if you see that difference from $0 to $500, but you don’t see that from $500 to $6,000, what would that indicate?

Rendleman: It could be that it is hard for individuals to detect changes, I think. And we also have analyses in the paper that say when states increase or decrease their EITCs, you don’t see a change in either vote shares or approval levels. So when you give someone like a lower-income single person more cash, that’s great because they generally don’t have a lot of things available to them in, you know, the laundry list that is welfare policies. They can’t really claim a whole lot of things, whereas people that might be or are already getting things from government could be less sensitive to these changes.

Sometimes people think of voters as myopic, and so they detect the first thing, and then they don’t detect other things after the fact. There’s a lot of studies to show this. Also within the EITC paper, this also seems to be the case because we only really detect an effect in the first year of implementation.

So yeah. Different populations, if you want to see this attribution going on, like targeting to populations that are going to be able to actually detect it and would change their behavior meaningfully for it—that’s part of how politicians would want to think about targeting these types of things.

Demsas: After the break, why democracy relies on voter sophistication.

[Break]

Demsas: I guess what I take from what you’re saying, and what started me thinking about, too, is just the sort of sense that people are taking a signal from this, right? They’re taking a signal about the political goals or the policy goals, the orientation of the governor and party that is in power when this happens—less so that they’re reacting, specifically, to the material circumstances that are changing for them. And part of what leads me that way—something you mentioned earlier that we didn’t get into, which is that—Republican governors are particularly rewarded for these expansions.

And that is such a signal to me because what it’s saying is that Republicans are less friendly to these sorts of programs. They’re more concerned about welfare expansion leading to problems with people not working a lot. Obviously, that’s why EITC, which is something that—it’s the earned-income tax credit and requires you to work—is more amenable to Republicans. But it is something that’s almost—it changes your perception of the Republican Party for a Republican governor or legislature to okay an EITC expansion or a particularly generous one.

And so is that how you think about it, too? That voters are getting a signal about what the priorities of these elected officials are, and that’s changing their vote share—less so that they’re reacting to the material shifts in their lives?

Rendleman: Yeah, totally. I think a lot of the story that we’re telling here is one of sort of symbolic politics in a way. So if you see an individual politician behaving against their type, or, like, your perceived type of them, then you’re going to be like, Oh, maybe this guy is not so bad, or, This woman is not so bad, so I’ll throw a vote his way, or, I’ll say he’s not as bad as he was when I get the next survey. I think that’s a big part of it and also contributes to this idea of traceability of policies.

Demsas: Yeah. I think that that’s one other part of your paper that is so relevant for this, is that you look not just at people who were eligible for the EITC but also people who were not eligible, and you find effects there. Can you talk through the effects you find from people who were not getting this benefit at all?

Rendleman: Yeah. We do see a bump in approval levels amongst people that are not eligible for the EITC. We went into this finding with sort of two thoughts: Either it’s that people that are not eligible for the policy are seeing the good effects of the policy on the ground, or they’re just into the idea of them being in a state government that is being productive. And so we try and parse these ideas by looking into this—we call them higher- versus low-exposure counties, so places, again, that claim more federal EITCs versus less federal EITCs.

We don’t see the same sort of strong effect for ineligible people when you compare high- versus low-exposure counties. So it’s sort of like this uniform effect amongst ineligible individuals across like the context of the county.

Demsas: So basically, if you’re ineligible and you’re in a county that doesn’t have a lot of people claiming that EITC, or you’re in one that does have a lot of people claiming, your effect is the same. It doesn’t really matter.

Rendleman: Exactly.

Demsas: Okay. So it’s not really about you observing, Oh, people around me seem better off. It’s that you’re just realizing this has happened in your state.

Rendleman: Yes.

Demsas: I think the first thing that’s going to come to a lot of people’s minds is: How do you control for all the things that are going to covary with this, right? Because if you’re a governor that’s willing to do EITC implementation and expansion, it probably means that there are a lot of other things about you that are going to be shifting in that direction. It’s probably not just this random heterodox thing that you did that you can fully isolate.

And on top of that, a lot of other stuff might be going on. Maybe the state’s more willing to do EITC expansion when the economy’s good. Maybe you’re more willing to do it when the state coffers are higher for whatever reason, like your investments paid off, or something happened, or you stopped paying off some municipal bond. You know what I mean?

So there are a bunch of things that could be happening that could covary with this decision to do this. So how confident are you that what you’re measuring is really the effect of EITC implementation?

Rendleman: Yeah. We can’t be completely certain. It’s social science. We just do massive amounts of checks to try and rule out alternative explanations.

Demsas: It’s the funniest part of reading a paper, where you’re like, Okay. Five pages of this are the study, and 30 pages of a study is just like, Are we sure that’s what we’re finding? (Laughs.)

Rendleman: We do try and look at what’s going on in the states where EITCs are implemented. We don’t see any sort of change in, sort of, the budget surplus of the states. So it’s not like an EITC is implemented in a time where, you know, you see a huge budget surplus, and then the state’s like, Oh, okay. So let’s do a bunch of different sorts of spending now.

You also see a sort of a commiserate drawdown on revenues. So the state is not taking in as many revenues anymore. So they’re giving it back. And we also don’t see strong effects in favor of a pure economic story going on. So we don’t see changes in pay in the areas where the EITCs are implemented or in unemployment rates. This is broadly consistent with the sort of mixed findings of EITCs. Some economists find very strong findings. Some of them find quite muted findings.

We also did follow up analyses trying to detect whether people in the survey that we are using detect whether the economy around them is doing much better or much worse. We don’t see very strong effects that suggest that’s going on, at least for ineligible people. In general, actually, we see people that live in high-poverty areas tend to say that the economy is doing pretty poorly, which is consistent with what you might expect.

So we can’t completely rule out the idea that maybe some of these places pass, like, these omnibus welfare policies, where not only did they introduce an EITC, but they introduced something akin to a CTC—

Demsas: Child tax credit.

Rendleman: Yeah. Child tax credit. Thank you. It doesn’t seem like that’s the case when we look at the budget data or in our checks of the places that are passing EITCs. But we can’t really account for everything that is going on in the political moment where these policies were passed.

Demsas: I want to take a step back here because you’ve kind of referenced this a couple of times now, but your research kind of pushes back against other research that has not found an effect of the EITC.

First, can you just give us a sense of what those studies say and why you think yours found something different?

Rendleman: Yeah. The past research on the EITC has been mostly concentrated around the idea that this is a policy that, perhaps purposefully, was deployed in a way that makes it difficult for people to attribute it to government. Suzanne Mettler, in her book The Submerged State, makes the case that many policies that you might expect would have a policy-feedback effect are deployed through the tax code in a way to obfuscate the role of government. You just get your check, as I was saying.

And for policies that might have particularly helped Democrats, let’s say, like an EITC, potentially, it’s sort of a way to level the playing field for politicians. Like, Okay. Well, you want to get this done, but we’re going to remit it in such a way that it’s going to be submerged through the tax code. And because of that, there have been other studies after that, that say, Who does get the benefit from EITCs? And it’s people like tax preparers or community-service organizations that are helping people navigate the complex landscape of the tax code. So it’s not the politicians, but it’s this delegated welfare state, to borrow a term from Andrea [Louise] Campbell. And this happens with a lot of social policy in the United States.

Demsas: So you just mean someone going to their local tax preparer thinks that they’re the ones who helped them, not the state.

Rendleman: Exactly

Demsas: Okay. Gotcha.

Rendleman: Like, I just hit it out of the park by finding this guy from H&R Block.

Demsas: (Laughs.) The Intuit lobby is alive and well.

Rendleman: Exactly. That was sort of the focus of the studies on the EITC. And they spoke to other work that focused on social policies and how they’re remitted and why people don’t respond to those policies. Sometimes people are just more interested in keeping their body and soul together, and they’re not going to necessarily be able to do this attribution process. And the ways that policies are implemented influences this.

The way that our study is different is that we try to exploit periods where—we try and find the best-case period for detecting an effect of an EITC. That’s going to be, like, the first year it was implemented, and it’s going to be in places where, potentially, it’s more generous or the state is doing things to allow for that attribution process to be a bit easier than it might otherwise be.

By focusing on the state EITCs, we’re able to basically examine the same policy across a bunch of different policy landscapes with a bunch of different policy-design strategies, and we’re able to test these different levers that might lead to better or worse attribution to the party that passed the policy. The other studies that study the EITC—they’re just looking at a given snapshot in time, either through interviews or through surveys. And what we do is really exploit this “over time” phenomenon, and say, Can we actually compare it to when we know we wouldn’t detect an effect to when we probably would?

Demsas: And one of the parts of the study that you haven’t talked about yet is what you just mentioned, which is: The way in which the government is almost trying to make the policy-feedback loop happen. So can you talk about the notification laws and what those are and how they work and why those are important?

Rendleman: Yeah. So in some states—and I think the city of Philadelphia—they have what are called EITC-notification laws. So if you are likely eligible to claim the EITC, your employer has to tell you. And so basically you get this note saying, Hey. If you haven’t already, when you file your taxes, you should make sure you claim this benefit. This is, at least to my mind, a tremendous idea. Telling them that you have done something or telling them to do the thing seems like a smart strategy. And in fact, that’s what we see in our study, that places where you have these notification laws, you see slightly higher approval levels for the incumbent governors.

Demsas: So I find it really funny because something very similar to this happened to me this week. I was at the optometrist, and I am always ready to fight when I go to the eye doctor because they are so annoying about giving you your prescription. But this year, the FTC—well, the reason that they’re annoying about it is because they want you to buy glasses from them, rather than going to, you know, 1-800-CONTACTs or going to whatever and Costco and finding something cheaper.

And I have literally been in a situation constantly where I’ll go to the optometrist, and then I will say, I want my prescription printed out, please. And they will try to make it impossible for me to get this prescription. And I’m, like, blood pressure high, ready to go in, fight with my eye doctor. And I get there, and for the first time, they just hand me the printed copy of my prescriptions. And I was like, What’s happening here?

And then I get another document called a Prescription Acknowledgement Form. And it says—I’m reading it right now: “In accordance with Federal Trade Commission regulations, optometrists are required to obtain a signed confirmation from a patient confirming that they received their glasses and/or their contact lens prescription.” And I was like, This is amazing. Not only did I get it; I got a piece of paper explaining to me that the FTC is the reason.

But the funniest part of this whole story is that after this happens, I was like, Oh, why am I getting this? I was just trying to figure out if everyone understood what was happening here. And the woman who was assisting me was like, Oh, yeah. It’s just a local D.C. regulation. It’s like, the piece of paper says that it’s the FTC, and it’s the federal government doing this. The person who handed me this piece of paper, who’s responsible for implementing this policy, is telling me it is a local D.C. regulation, but oh, also, it applies in Virginia.

And this is not to make fun of this woman at all. I don’t think that’s the point. I think that part of what’s happening here is just, like, even with all this notification, this feedback loop is extremely difficult to create. And I don’t know. It both made me optimistic and pessimistic about democracy in the same interaction.

So I don’t know how you feel about the efficacy of these kinds of programs when it’s so, so difficult to actually make clear what is happening and why, even when they do this. I mean, I think another clear example of this is when Donald Trump was signing the stimulus checks, and people were just like, Oh, it’s awful. He’s doing this. But it’s like, I don’t know—shouldn’t we know what government is doing?

Rendleman: Yeah. I think this story sort of sums up how everyone sort of interacts with the welfare state or just public policy, in general. Like, you hear about public policy from your neighbor or from the news or something like that, and you get 10 percent of the story, and you’re just going to take that 10 percent and go with it.

Demsas: It’s actually really funny, too, because I tweeted about it, and someone responded, Oh, in my office, they thought it was because of corporate, and that’s why it was happening.

So we talked a little bit about this earlier, but I want to delve into it a bit more. There’s this idea that I think that most people popularly have, which is that engagement with government programs will increase both approval, if you’re getting something beneficial from the government, but also that it will make you more likely to participate in the political process. I mean, I know you mentioned that, like, this is probably too small of a benefit to change the amount of leisure time you have in the way that Social Security did.

One of the books that I’ve read on the subject that has been really formative to me is this book by Jamila Michener called Fragmented Democracy. And so she looks at Medicaid enrollment and finds that enrollment yields a decline in the likelihood of participating in politics and voting. Her story, I think, has a lot to do with the ways in which the Medicaid implementation can be really degrading for people—the way they’re treated in these offices, the messages they’re receiving about their worthiness as members of the political body.

And is your idea that you think the EITC is just like a better-designed program than Medicaid, or how does your finding kind of interact with Michener’s?

Rendleman: I love the idea that you could think that my study is in conversation with Michener’s.

So it’s a dramatically different policy. A lot of policies that came out of the ’90s will say, We’re very much interested in the idea of dignity, or how we can maintain the dignity of individuals when they claim these policies. And going to a benefits office, where you have to talk to a tired bureaucrat who is upset to be at their job because it’s a difficult job, or you have to travel far away to the office to try and claim the policy—so just making it more difficult for you. All that comes together to make it sort of a lame thing to try and seek out, even if you need it.

The EITC is just a check. It’s easy. So even though, sure, you have to do your taxes, we always have to do our taxes every year, and maybe it’s more or less complicated for some individuals, depending on how complex their taxes are, but it’s nothing like having to go into an office, have to ask answer questions, and have to face the notion that things are not going your way, and you’re having a tough time.

Demsas: I feel like a lot of this work is, in part, answering the question: Do we think that voters are politically sophisticated? There’s a famous 1964 chapter by a political scientist, Philip Converse, called “The Nature of Belief Systems in Mass Publics.” And I have not gone to political-science grad school, but the people I know who have say this is one of the first things everyone gets assigned and has, like, 13,000 citations.

He makes the argument he’s making, and this is a very early attempt to sort of really go out into the field and learn from people through survey data, rather than just kind of sitting in an ivory tower and doing theorizing. And he basically looks at the ideologies of voters. And his argument is that people’s policy views are inconsistent and not ideological. When you ask someone, one day, What do you think about X? It’s not going to be the same question if you ask them at another time. And then it won’t covary with things that should covary together if you were to be a consistent person.

Your argument kind of supports the idea that voters actually are somewhat sophisticated, at least they’re able to see something that is valuable and approve the right people for it. But then there’s also parts of your study which push in the other direction, right? Like, the fact that it’s not super durable indicates that, perhaps, you know, you ask them in year one: Oh, they’re really happy about it. Year two: They don’t even care whether or not it happened. And the size of the benefit not really varying with how much support people are giving to the politicians that they view as responsible also indicates that perhaps they’re not politically sophisticated.

So do you think that voters are politically sophisticated? How has this research changed your views about that, if at all?

Rendleman: No. I think I’m an optimist on sophistication. I think a lot of times political scientists are a bit pessimistic on individuals’ capacities to actually know what’s going on to them, because it is quite complicated. But we don’t often set up our analyses or studies in a way to give voters the benefit of the doubt.

I think the effects that we detect are indeed consistent with this sophistication story. And I think, maybe, I’ll push against this idea that after that first post period—after the implementation of the EITC—that’s where you see the bump. I think I would push against the idea that reversion back to the mean is indicative of voters not being so sophisticated.

It just means that other things happened that interceded in people’s capacities to reward the governor in that same direction. Not all politicians have these hot streaks that will allow for constant bumps to their vote share at the polls or in how people approve of them on approval polls. Scandals get in the way; other policies get in the way, etcetera.

And the finding on the amount of money—it’s not really clear how people would respond to that amount of money, because it’s not like people have that counterfactual, necessarily, or, like, can really detect, I got 15 percent of the federal EITC this year and then 18 percent of the federal EITC the next year. How would we expect that to manifest in approval levels? Like, how much more would we expect people’s attitudes to change? It’s not really clear.

So I like the study because it really points in favor of the idea that voters are reasonably sophisticated. We shouldn’t just expect them not to react to policy. They will react to policy. And it depends on the environments that these individuals are in, in order for us to see an effect. So we should just be far more conscientious about how we deploy policies and where we deploy policies and what sorts of messaging goes along with the policy.

Demsas: One thing that I think really plays into this is sort of the ideological nature of politicians themselves. I mean, I think we have this sense that what we would want is: good-guy politician who does what he believes no matter what, and pushes back against, you know, anything that they find is negative, and is a leader.

But democratic theory, and what we’ve been talking about this whole episode, is that, actually, you kind of want a politician that all they’re going to do is try and make sure that they’re making people happy. They’re going to respond to positive feedback loops. They’re going to search for them. They’re going to try to make sure people know what they’re doing, and they’re going to try to iterate to be more popular because that is how you get better governance, according to basic democratic theory.

And it seems we’re in a highly ideological environment, where politicians are often not willing to do things that are very popular or are willing to do things that are very unpopular.

I think one example of this is the child tax credit, which we mentioned briefly before. I think it’s a great policy. And when it expanded in 2021, it lifted more than 2 million kids out of poverty—when that was expanded by the Biden administration. But it doesn’t get renewed. And there isn’t a massive backlash. And survey data from NPR/Marist showed that just 15 percent of respondents who got payments said that it had helped their families a lot.

And yet, both parties run on it during the presidential election, right? Both the Harris campaign and the Trump-Vance campaign—they run on this, and they talk about the child tax credit. It’s like a point of unity. You can see a lot of Republicans pushing for it and Democrats.

You know, we tend to think about the blame of policy-feedback loops resting with voters for not being sophisticated. But is part of this also just that the ideological nature of politicians makes them often unwilling to engage in those feedback loops appropriately?

Rendleman: Yeah. Yeah, totally. So it’s interesting because you would, indeed, just expect politicians to do the thing that would maximize votes, but politicians don’t always have full information on what would maximize votes, and they also don’t have full information about how people react to policy, necessarily. So it’s funny—you bring up the presidential campaigns, and it seems like presidents will just run on boosting the economy in whatever way possible.

And if you have this sort of universalistic policy, like the child tax credit or everyone-has-kids, everyone-could-benefit type of policy, then that just seems like good political capital. It’s just something to run on. But then the campaign is done, and you actually have to get policy done. It’s hard to bargain with other people towards actually implementing things and, also, implementing things in a satisfactory manner.

One reason why we don’t have a very thick safety net in the United States is, in part, because of the entry of not just outside influence into politics—so let’s say, like, money in politics, like lobbying and things like that—but also because, yeah, these politicians are people, too, and—

Demsas: Politicians are people, too. (Laughs.)

Rendleman: Yeah, exactly. And when they have to work with each other, it’s hard to actually overcome not only their own conceptions of what it means to have good policy but also, you know, what they actually have to do in office. Like, you know, Maybe instead of passing the CTC, I want to do something else.

Demsas: Yeah. Well, that’s a great place for our final question, which is: What is something that you once thought was good at the time but ended up being only good on paper?

Rendleman: It’s funny because it’s going to be about this idea that politicians are people too. So my other arm of my research agenda is focused on basically the workplace of politics, so focusing on how politicians interact with each other but also how bureaucrats interact with each other.

And when I was doing part of my dissertation work, I wanted to study the idea of, like, Okay, when people of color are with each other in deliberative spaces in government, maybe they work better together. And so I did all this work in my third or fourth year of grad school, and I was fully expecting to see, Yeah, if we add more, let’s say, Black people to a congressional committee, you’ll see that they are more effective at their jobs. And I didn’t see that, and I was really bummed about it. And, you know, I talked to people, and people were like, Ah, maybe it’s just, like, a null effect, so throw it away, and move on with your life.

But I was quite stubborn, so I went to D.C., and I started talking to people. And I asked them, you know, Why am I seeing this effect? I’m seeing no effect, effectively. And it never dawned on me that when people come into Congress, they are not just the silos. They’re not just by themselves, working. They are bargaining with other people. They’re working with other people, and they’re deciding, like, What is the best way to position myself or ourselves as a group?

And ultimately, I found that in the case of African Americans, you don’t see a general empowerment effect; you see this very specific delegated effect. So people that are closer to the policies—groups like the Black Caucus are in support of—those are the people that are speaking more in these settings, whereas people that are farther from those policies speak less.

So this empowerment effect—thought it was good on paper. Maybe it was, but it led to just this far more robust finding that I think squares more with how we think about how elites think about the political process, in general.

Demsas: Side note: I feel like I’m very on the bandwagon of everyone should be publishing their null effects, and it shouldn’t be a negative thing. I guess that’s probably the fault of, like, you know, journals and stuff and not really you, in particular, Hunter.

Rendleman: Well, yeah. Like, my incentives at this stage in my career are such that I have to publish, and sometimes reviewers just like seeing positive effects as opposed to null effects.

Demsas: Blame reviewer two, always.

Rendleman: Yeah, of course.

Demsas: Well, thank you so much, Hunter, for coming on the show.

Rendleman: Thanks so much for having me. I had a lot of fun.

[Music]

Demsas: Good on Paper is produced by Jinae West. It was edited by Dave Shaw, fact-checked by Ena Alvarado, and engineered by Erica Huang. Our theme music is composed by Rob Smierciak. Claudine Ebeid is the executive producer of Atlantic audio, and Andrea Valdez is our managing editor.

And hey, if you like what you’re hearing, please leave us a rating and review on Apple Podcasts.

I’m Jerusalem Demsas, and we’ll see you next week.

Washington’s Nightmare

Donald Trump is the tyrant the first president feared, Tom Nichols wrote in the November 2024 issue.

Thank you, Tom Nichols, for your timely article profiling our Founding Father. President George Washington modeled key tenets of public service: Be a citizen, serve with integrity, graciously pass the responsibility on to the next citizen. Choosing to juxtapose Washington’s virtues against the abomination that is Donald Trump not only is appropriate, but should serve as a wake-up call to all Americans.

Sadly, it seems many Americans have lost touch with our history. I hope this changes before we find ourselves in a mad scramble to reclaim our more perfect union. Our patriots fought hard for our independence; our forefathers died to keep it. For nearly 250 years, we’ve dedicated ourselves to getting it right.

Do we have any idea what we stand to lose?

Peter Brown
Lyman, Maine

I admire Tom Nichols’s writing and analysis. I teach courses on leadership and ethics at Georgetown University’s Walsh School of Foreign Service. I conducted a review of 44 peer-reviewed academic-research articles that analyzed various aspects of Trump’s leadership style. My conclusion was that Trump has extremely limited leadership skills. He is a domineering bully who retaliates against anyone who challenges him. His moral decision making is based on self-interest; he lacks emotional intelligence, and his personality is defined by paranoia and victimhood. His principal skill for influencing others is creating an attractive image and selling a product. In times like these, we need a true leader, not a salesman.

Kenneth Williams
Chapel Hill, N.C.

As I read “Washington’s Nightmare,” I returned to the historian Erica Armstrong Dunbar’s Never Caught: The Washingtons’ Relentless Pursuit of Their Runaway Slave, Ona Judge. Dunbar’s book tells the story of a woman enslaved by George and Martha Washington, who fled the president’s residence in 1796. Washington pursued Judge until his death.

The article and book make for a problematic juxtaposition: I, too, celebrate Washington’s achievements, his heroism and devotion to the republic. But Tom Nichols’s sometimes hagiographic treatment of the first president gave me pause. Nichols devotes only one passage—about 150 words—to the problem of Washington and slavery. I understand that his larger interest is in how Washington “set the standard of patriotic character for his successors” and not necessarily in the glaring contradiction of Washington and the people he enslaved. But if one is to compare Washington’s virtues with Trump’s, this contradiction demands attention. Shouldn’t we consider Ona Judge’s nightmare, as she was pursued relentlessly by the Washingtons for having dared to claim what Washington himself had fought for—freedom, dignity, certain unalienable rights? Nichols prominently features General John Kelly’s remark that after serving as president, Washington went home. When the Washingtons returned to Mount Vernon, what did their slaves think?

I’m grateful for Nichols’s assessment of Washington’s legacy. But this aspect of it deserves deeper engagement.

Kevin L. Cole
Sioux Falls, S.D.

Tom Nichols replies:

Thank you to the many readers who appreciated my look back at our first president. I, too, found the distance between George Washington and Donald Trump almost too painful to grasp. Peter Brown raises the real problem of poor historical literacy among Americans, but I wonder if the larger issue is that our politics have become amoral and transactional, something that Washington would have abhorred and that education alone cannot solve. Perhaps Kenneth Williams is closer to the mark by noting that Trump, above all else, is the product of a modern phenomenon—marketing.

I understand Kevin Cole’s objection regarding Washington and slavery, but not every thought and reflection dealing with the first president ought to center on slavery. Washington rebelled against an important institution of his day—monarchism—and to judge him from the 21st century because he did not right one of the other grievous wrongs of the 18th strikes me as ahistorical and, potentially, a diversion from the example he offers us in our current struggle with authoritarianism.

The End of Judicial Independence

One of America’s greatest achievements could disappear overnight, Anne Applebaum wrote in the October 2024 issue.

Anne Applebaum provided an important reminder that some Supreme Court justices appear loyal to President Donald Trump and not the Constitution. But their bad behavior does not end there: Some are loyal, too, to the billionaire class.

Justice Samuel Alito went on an expensive fishing trip with the billionaire Paul Singer, staying at a lodge that charges more than $1,000 a night for a room. Afterward, Alito did not recuse himself from cases involving his benefactor’s hedge fund. Justice Clarence Thomas, not to be outdone, has enjoyed gifts and favors from several billionaires, taking at least 26 flights on private jets. Thomas and Alito claimed that they didn’t know that they were obligated to report these far-from-trivial gifts.

Something has to be done to put an end to this. We need an enforceable code of conduct for the nation’s highest court.

Seth Wittner
Worcester, Mass.

I graduated from law school in 2010. I didn’t know then that those were the halcyon days, when precedents such as Roe v. Wade and its landmark arguments seemed unassailable. I was only 25. I had bought into the teachings about the sanctity of the judiciary’s role in our legal system. I believed—I still believe—in the ethics of my profession, in the dignity of our highest courts and what they and their decisions symbolize. Respect for tradition and the analysis of capable scholars and judges of years past was crucial to my training. But I wonder what younger people sitting in constitutional-law classes must think today, having now witnessed the ugly, frayed edges of our democracy. What do these sobering, disappointing lessons mean for the practice of law in the future?

Brittlynn Mourgue
New York, N.Y.

Behind the Cover

In this month’s cover story, “Walk on Air Against Your Better Judgment,” Caitlin Flanagan writes about her relationship with the poet Seamus Heaney. Heaney first met the Flanagan family during a year spent teaching English at UC Berkeley, where Flanagan’s father, Tom, was a professor. Heaney became a second father of sorts, Flanagan writes, and the poet and his work shaped her understanding of the world. The cover image shows Heaney at the Royal Society of Literature, in London, in 1995, months before he was awarded that year’s Nobel Prize in Literature.

— Paul Spella, Senior Art Director

The Atlantic Correction

“The Most Remote Place in the World” (November 2024) stated that, relative to the universe, the plane of the International Space Station’s orbit never changes. In fact, because of the Earth’s oblateness, it does shift over time.

This article appears in the January 2025 print edition with the headline “The Commons.” When you buy a book using a link on this page, we receive a commission. Thank you for supporting The Atlantic.

Photographs by Kennedi Carter

Boley is easy to miss. The flat vista of eastern Oklahoma is briefly interrupted by a handful of homes and some boarded-up buildings, the extent of the tiny town. In its heyday, Boley was a marvel of the Plains, a town built by and for Black people that attracted visitors and media attention from across the country. But now its businesses are defunct, and the steady outflow of young people has left behind an aging population of only about 1,000 people. The main reminder of the town’s glory days is the historical marker I passed on Route 62. The road took me to the wide-open fields where the Boley Rodeo, hailed as the oldest Black rodeo in America, is held every year. I arrived at the grounds early on a May morning, before the sun pierced the clouds. Karen Ekuban, the rodeo’s promoter, had already been there for hours.

The rodeo was in two days, and Ekuban and a small crew of volunteers, including her children, had been busy. She was obsessing over every detail. When new bleachers had been installed a few days earlier, she’d spent so much time directing workers to ensure that the earth underneath them was level, with no holes where people might trip, that she’d dozed off as she drove home that night. Instead of veering into traffic, her vehicle had eased onto the shoulder, jolting her alert. Ekuban told me the story with a giggle, as if nothing had happened. She was “awake now,” she said.

The annual rodeo had been a Boley tradition for more than 120 years, once regularly drawing crowds of thousands, but as of late, it had become something of a glorified family reunion for people with ties to Boley. Ekuban had a plan to try to turn that around. She’d sunk thousands of dollars of her own money and months of her time into throwing what she hoped would be the best rodeo the town had ever seen. Her goal was to raise at least $200,000 to help revitalize Boley, and bring back national attention to the town.

Kennedi Carter for The Atlantic

Without much in the way of a budget, Ekuban created a new sign to welcome the visitors she hoped would come for the rodeo. Ekuban’s new logo for Boley featured a silhouette of a cowboy on a bucking bull. She also brought on Danell Tipton, a world champion bull rider, to serve as the rodeo’s producer.

Members of the community turned out to help. Henrietta Hicks, an 89-year-old judge whom Ekuban affectionately refers to as “Grandmother,” offered her home to Ekuban the night of her near accident. Other old-timers and close friends stopped by to lend a hand. I watched as they worked up a sweat, carrying provisions that they hoped would feed thousands of visitors. All the while, Ekuban’s phone chimed, sending a notification for each ticket bought online. As the sun set, a caravan of trailers carrying horses and bulls arrived.

Ekuban’s aspirations for the rodeo echo the ambitions of her forebears. In 1866, a year after the Civil War ended, the Muscogee (Creek) Nation signed a treaty with the United States. In that agreement, all Black people who had been enslaved by the Muscogee Nation were emancipated and provided with full Creek citizenship privileges, including the right to landownership. (The Muscogee Nation would later redefine citizenship to exclude the descendants of those enslaved people, a decision that some descendants are challenging in court.) Abigail Barnett and her father, James, a freedman of the Muscogee Nation, were deeded the land that would become Boley, and many other Black Creek citizens settled nearby. After the arrival of the Fort Smith and Western Railway, in 1903, J. B. Boley, a white railroad agent, helped the community incorporate, a process completed in 1905. According to the historical marker in town, Mr. Boley had faith in the Black Man to govern himself and persuaded the railroad to establish a townsite here.

In those days, Oklahoma was promoted by a network of boosters as a promised land for Black people, part of a larger movement to establish Black autonomy and self-governance throughout the West. Some 50 Black towns popped up across Oklahoma from 1865 through 1920. This movement was closely associated with the work of Edward McCabe, who’d risen to prominence when he became the state auditor of Kansas, the first Black person to be elected to a statewide office in the West. McCabe moved to the Oklahoma Territory after two terms in office, hoping to make it a majority-Black state. In an 1891 speech, he laid out his vision. “We will have a new party, having for its purpose negro supremacy in at least one State,” he said, with “negro State and county officers and negro Senators and Representatives in Washington.” Answering the summons of McCabe and others, thousands of Black people chose to forsake the dominion of Jim Crow for Oklahoma.

Karen Ekuban, the promoter of the 2024 Boley Rodeo, hoped that the event would help bring her hometown back to its former glory. (Kennedi Carter for The Atlantic)

The very existence of Boley, and other towns like it, proved what was then—in many white people’s minds—a radical proposition: that Black people could thrive on their own. Booker T. Washington visited the town in 1905 and marveled at its two colleges, its abundance of land. Unlike in other Black communities in America at the time, many of which struggled, the original guarantee of citizenship and property rights in Boley under the 1866 treaty gave citizens a measure of security and wealth to pass on. Even decades after Boley’s founding, the allure of freedom and the ability to own the land under their feet drew Black families—including Ekuban’s Mississippian parents—from across the South.

[From the September 2019 issue: The Mississippi Delta’s history of Black-land theft]

The exact origins of the Boley Rodeo are lost to time, although we do know that it likely arose organically from local customs that predated the town itself. Rodeo, descended from practices established by Spanish and Mexican ranchers, had been molded into its modern form through the participation of Black cowboys, and Boley’s iteration formalized it locally as an event. Even in its early days, the Boley Rodeo regularly drew spectators of all races from Oklahoma and elsewhere, and inspired several other rodeos across the South and the West. One year, Joe Louis, the legendary boxer, made an appearance. The rodeo was always an important source of income for the town.

The 1960s marked the heyday of both Boley and its famous rodeo. According to The Black Dispatch, a newspaper in Oklahoma City, at least 10,000 people attended the event in 1961. The rodeo that year featured acts such as Billy “The Kid” Emerson, a Black rock-and-roll pioneer, and a full orchestra from Houston. In 1963, the same newspaper noted the growth in the “small Negro metropolis,” marked by the five new businesses that had sprung up on Main Street.

But as has been the trend in much of rural America—particularly Black rural America—Boley has suffered from the loss of capital and the gravitational pull of city life. In recent decades, many young people left in search of opportunities elsewhere, and the elders who remained struggled to keep the memory of the old days alive. Boley’s self-reliance had been lauded by white politicians and business executives, but when the town struggled, they lost interest, leaving it to wither.

Over time, the rodeo faded as well. Before Ekuban pitched her plan, it seemed possible that the rodeo might disappear entirely, and that—in a state where history books rarely mention places like Boley—the history might disappear too. Even with Ekuban’s intervention, there are no guarantees.

Ekuban left Boley after high school, some 35 years ago, with no interest in returning. But she eventually discovered that the place had a hold on her. Her mother had taught for decades in the now-defunct school, and her father had served as superintendent. Ekuban remembers how sporting events were held at the local prison, one of the town’s main employers, because there were no facilities elsewhere. When she returned to the area, she bought a house in nearby Spencer, another majority-Black town in Oklahoma. Ekuban started several revitalization projects, because she wanted to cultivate the beauty she saw in her hometown. But after she learned more about Boley’s history from “Grandmother” Hicks, Ekuban’s says her desire to help took on new significance.

She wanted to impress that significance upon the rodeo’s supporters, and to explain just why its success was important. So, the day before the rodeo, even with plenty of setup work left to do, Ekuban gathered a group of friends and family members. They drove to Tulsa to visit the Greenwood Rising History Center, which commemorates Tulsa’s Greenwood District, a neighborhood that had been established by Black people as part of the same wave of migration that built Boley. Greenwood was a thriving community that grew so prosperous, it gained national fame under the moniker “Black Wall Street.” But in 1921, a mob of white assailants attacked and burned the district, killing as many as 300 Black residents and leaving thousands more homeless. The massacre was a tragic reminder of the limits of the optimism of boosters like McCabe, who believed that a Black state could offer “equal chances with the white man, free and independent.” But the exhibition also shows that, even if only briefly, the exertions of Black people themselves brought McCabe’s vision to fruition.

Kennedi Carter for The Atlantic

In front of the exhibits showcasing the golden age of Black Oklahoma, Ekuban told the crowd how early Boleyites had journeyed hundreds, in some cases thousands, of miles to unfamiliar territory, and then dug, carved, and heaved their way to independence. They built schools, gristmills, homes, universities, churches, and hospitals. They bore children, and one generation passed down its story to the next. Part of that story was the Boley Rodeo. For more than a century, cowboys had sought to ride atop the bulls and broncos as long as their will and body would allow. They did this for money and acclaim, yes, but also for a less tangible reason: because, no matter how many times they were knocked off their horse, dusted up, and trampled on, trying was always their salvation.

Kennedi Carter for The Atlantic

By the morning of the rodeo, Ekuban had sold more than 1,400 advance tickets, and she hoped thousands more would purchase admission at the gate. She’d already banked more than $20,000 from ticket sales.

A parade of souped-up classic cars, followed by dancing, flag-waving spectators, opened the event. Willie Jones, a country singer who was featured on Beyoncé’s Cowboy Carter album, performed. And then, at 7 p.m., the cowboys entered. The gates to the east of the arena opened and the horses galloped through, kicking up huge clouds of dust.

For the first time in weeks, Ekuban breathed easy. If only for a moment, she could return to the wonder she had experienced as a kid. But then, just 30 minutes into the rodeo, she spoke with her husband, whom she had appointed to liaise with local police and security. Attendance had surged to more than 5,000 people. Without enough seats, some spectators were hanging on the white railing of the rodeo arena. Ekuban had to tell her husband and the police officers to close the gates to the rodeo grounds.

Kennedi Carter for The Atlantic

As the sun set, young cowboys proved their mettle by riding half-wild broncos, the muscles of both man and beast straining to prevail. When the bulls came out, the competitors traded their cowboy hats for padded helmets. Away from the action, Ekuban watched as attendees bought merchandise from stands run by her fellow Boleyites.

Finally, it was time for the centerpiece of the rodeo: the Pony Express, a relay race on horseback, a tradition unique to Oklahoma rodeos. In the finals, the Country Boyz faced off against newer competitors, the Yung Fly Cowboys. Four riders from each team lined up, whispering in their horses’ ears.

The gun fired, and the competitors took off. Riding half-ton horses, they went around the track furiously. The Yung Fly Cowboys finished first, unseating the Country Boyz as champions. In a moment that might have seemed strange at any other rodeo, the stereo system swapped out the country music that had been playing all day for “Get in With Me,” by the Florida rapper BossMan Dlow. The winning crew’s music blared, and winners and losers alike recited every lyric.

The rodeo was over.

Kennedi Carter for The Atlantic

Today, with the success of Cowboy Carter, with the breakthroughs of Black country artists such as Shaboozey, with the growing prominence of Black rodeos including Bill Pickett’s Invitational, and with increased awareness of the Greenwood District through vehicles like the HBO miniseries Watchmen, it’s easier than ever to see the Boley Rodeo’s legacy. In popular culture, the mélange of Westerns and cowboys and country music that is often called “Americana” has been associated with a particular vision of liberty, in which the people most entitled to the possibilities of the West are white men who mastered the land and its inhabitants. But Boley and its rodeo helped create a countertradition—that of the Black West. In this tradition, Black people who had every reason to give up on America instead struck out for places like Boley and dictated the terms of their belonging.

The Black West is a compilation of stories often overlooked or forgotten. As someone from Oklahoma, I’ve long wondered why our histories—of rodeos and of the dreams of leaders like Edward McCabe—don’t appear in many mainstream depictions of Black life. It is our obligation to salvage and reimagine these remnants.

Kennedi Carter for The Atlantic

When it was all over, Ekuban had sold about 3,500 tickets to the revitalized rodeo, and had raised more than $270,000 from ticket sales, donations, sponsorships, and investments. She deemed it enough of a success to start planning the next one before the spectators had even left. It remains unclear whether the rodeo can actually attract enough money to save Boley, but the event undoubtedly helped spread the word about the town’s history and culture, and that was, in itself, a kind of victory. I was surprised when Ekuban, an eternal optimist, seemed to acknowledge how quixotic her mission is. “Boley will never be what it was,” she told me, then added: “But who says it can’t be better?”

This article appears in the January 2025 print edition with the headline “Boley Rides Again.”

Editor’s Note: Find all of The Atlantic’s “Best of 2024” coverage here.

Vulgar year, vulgar music. Kendrick Lamar and Drake made a show out of accusing one another of depravity. Pop entered a renaissance thanks to smart young women sharing impure thoughts. Taylor Swift released a sprawling confession of lust-crazed misjudgment. Hip-hop’s “sexy drill” scene continued to flourish. Acclaim flowed to artists who defied genre, distribution conventions, and the very format of the album: Cindy Lee’s two-hour rock epic was released via GeoCities, Mk.Gee blended hair metal and ambient music, Charli XCX’s Brat landed less like an album than a fashion line.

What’s happening? The internet’s culturally fracturing effects are becoming terminal, killing any shared sense of how things are done. Pop stars and punks alike are embracing the great dissolving by saying exactly what they feel, exactly how they want to say it. Although my top picks span a variety of genres, many of them have a similar spewing quality. They play like glorious run-on sentences, full of oversharing and id.

Follow along on Spotify.

10. Sabrina Carpenter, Short n’ Sweet

As rom-coms fade into a minor art form, pop has taken its place in charting society’s courtship anxieties. A little bit Dolly and a little bit Britney, Carpenter reworks the blond-bombshell archetype for an image-obsessed, connection-starved era. Beneath an easy-listening, FM-radio surface, Short n’ Sweet is supersaturated with ear-catching detail and sex-columnist wit, ballasted by tragic irony: If a girl like this can’t find a good man, what hope is there for anyone else?

Listen to: “Coincidence”

9. Ka, The Thief Next to Jesus

Writers, take heed: “Only embed indispensable, not a sheet wasted.” So preached the last testament of Ka, the 52-year-old New York City emcee who died in October. The album opened with a broadside against materialistic “dummy rap” and sexual exploitation sold as empowerment—a common complaint from the underground, but one that Ka rendered with new clarity. His power—whether analyzing poverty, religion, or racism—lay in the haunted timbre of his voice, and his insistence on making every syllable matter.

Listen to: “Borrowed Time”

8. Mannequin Pussy, I Got Heaven

Indie rock is sounding comfy and nostalgic lately, so it’s worth cheering the artists who make independence still sound ferocious. The fourth album by these Philadelphia punks is a pristine work in the Sleater-Kinney tradition of petulance: Noise and vulnerability surge in exquisite counterpoint; guitars crunch as satisfyingly as an autumn leaf under one’s heel. “What if we stopped spinning, and what if we’re just flat?” singer Missy Dabice asks, her voice trembling with fury at complacency in all its forms.

Listen to: “Loud Bark”

7. Sega Bodega, Dennis

The ringtone that cries out one minute into this album is like an inverted wake-up alarm, drawing the listener into a shared dream state with one of the most interesting electronic producers of the moment. Sega Bodega uses EDM as a storytelling tool, and what he’s expressing here is a whoosh of inexpressible feelings. You’ll summit mega-rave peaks, rest in oases of Enya-esque choral singing, and arrive at the end of the journey thankful that you can return to his netherworld whenever you want.

Listen to: “Kepko”

6. Hurray for the Riff Raff, The Past Is Still Alive

The ninth album by the folk singer Alynda Segarra is a warm and inviting road memoir that also bitterly criticizes an unfair nation. Many of the friends Segarra made while wandering America as a “little girl with a buck knife and a fake ID” have lost their lives amid poverty and addiction, but Segarra sings about them in a way that underplays the drama, trusting in the efficacy of strong melodies with soft reverb. “You don’t have to die if you don’t want to die,” Segarra sings—magic words at a time of waning faith in the future.

Listen to: “Alibi”

5. Beyoncé, Cowboy Carter

“Pop star goes country” is a common gimmick that’s enabled plenty of mediocrity, but Beyoncé flipped that expectation on its spurs. Let loose on the range of the Americana canon, she went wild, getting as experimental as any celebrity musician ever has. Months later, I’m still shocked by the electric-sitar molasses of “Ameriican Requiem,” the maximalist-minimalist bass lines of “Levii’s Jeans,” and the twisty-turny dance megamix that takes over late in the album. Even after the online arguments about Beyoncé’s influences and ideology became tiresome, Cowboy Carter’s music remained grander and weirder than discourse could capture.

Listen to: “Riiverdance”

4. Floating Points, Cascade

A neuroscientist who’s made acclaimed forays into jazz and scoring ballet, Sam Shepherd turned his brainpower to the dance floor for Cascade. The songs start as seemingly by-the-books techno beatscapes, but they bloom and morph in thrillingly exotic ways. On “Birth4000,” it sounds like the ghost of Donna Summer is cooing through a motorcycle engine; when “Afflecks Palace” explodes into live drumming, it’s as though a blueprint is suddenly becoming a building. Making body music this vivid takes intellect.

Listen to: “Vocoder (Club Mix)”

3. Kim Gordon, The Collective

A 71-year-old noise-rocker reciting spoken word over drum machines might scan as avant-garde impenetrability. But the miraculous truth is that Kim Gordon, Sonic Youth’s droll co-founder, made one of the most sheerly amusing albums of the year. Her free-associative lyrics are either mesmerizingly strange or plainly hilarious, and her tangled, clanging riffs have an oddly soothing effect. She’s describing modern brain rot so as to counter it with something better: creativity expressed at max volume.

Listen to: “BYE BYE”

2. Charli XCX, Brat and Brat and It’s Completely Different but Also Still Brat

It’s not just a color, not just a meme, not just a presidential campaign’s desperate bid for coolness—but it’s also not just a really fun album. Brat is an evolutionary branching moment for music, consolidating long-brewing cultural shifts and pointing the way ahead. The main leap forward is the approach to vocals, which connect rap’s penchant for melody in syncopation, Swift-pop’s flair for sassy specificity, and electronic music’s insistence that production—filters, beats, samples—is songwriting. This sneaky complexity transforms bubblegum pop into something more like an everlasting gobstopper: You can enjoy it endlessly, turning over flavors of gossip, escapism, and emotional revelation.

That Brat is a codex of fresh musical language was made clear on the remix album, which invited a wide cast of performers to try on Charli’s style of cyborg confessionalism. The results were revelations for each participant: Lorde had never sounded so down to earth; Robyn had never been this swagged out. The remixes aren’t better than the main album per se. But they do nudge Brat’s affective landscape—anxiety fighting recklessness, creating bittersweet estuaries and steep spires of ecstasy—to greater extremes, confirming the core belief of Charli’s career: Pop can still take new shapes.

Listen to: “Girl, so Confusing Featuring Lorde”

1. Mount Eerie, Night Palace

I’ve never been a 46-year-old indie-folk songwriter living on a rainy island in the Pacific Northwest, raising a young daughter, contemplating the impermanence of existence. But Phil Elverum’s astonishing, 26-track album makes me feel as though I know him and his life as well as I know my own. I can feel the clammy wind, smell the piney brushfire, and sense the precise blend of isolation, ennui, and wonder that seems to fill his days.

Like some charcoal scenebook, the album is a study in stark textures—guitar distortion, click-clacking drums, Elverum’s kindly, boyish voice. Stylistically, the songs wander from jaunty rock to roaring metal to solemn spoken word, but every verse is united by a low hum of feeling, the sense of great truth slowly rising into view. With forthrightness and flashes of humor, Elverum memorializes his individual life in the context of the land he occupies, the matter that he’s made of, and the void beyond all of that. If you fear our world is changing, here’s a staggering reminder that it always has been.

Listen to: “I Spoke With a Fish”

This is an edition of The Atlantic Daily, a newsletter that guides you through the biggest stories of the day, helps you discover new ideas, and recommends the best in culture. Sign up for it here.

The days of tech CEOs tussling with Donald Trump are fading. After distancing themselves from Trump during his first administration—and publicly rebuking him after the events of January 6, 2021—many Silicon Valley leaders are now taking a softer approach. Jeff Bezos, Mark Zuckerberg, and OpenAI CEO Sam Altman have each pledged, through their companies or their personal coffers, individual $1 million donations to Trump’s inauguration fund. The Google co-founder Sergey Brin, who protested Trump’s immigration policies in 2017, apparently dined at Mar-a-Lago with Trump and Google’s CEO, Sundar Pichai, this month; Bezos, along with the the heads of TikTok and Netflix, are reportedly on the schedule there this week too. As Trump put it in a press conference today: “In the first term, everybody was fighting me. In this term, everybody wants to be my friend.”

Friendship may not be exactly what these tech CEOs are after. Self-preservation seems to be playing a role—these companies don’t want to lose out on government contracts or face retribution from a man known for threatening to punish his critics. For years, Trump was no friend to tech, and vice versa: During his first term, he used Twitter to lob insults at Amazon and its then-CEO Bezos. And as recently as this past summer, Trump was hurling unfounded accusations at Zuckerberg. Ambition is likely part of the calculus too; CEOs hope that Trump will go easier on the industry than the Biden administration did, including on crypto and AI. Now, as Trump prepares to take office a second time, tech executives seem eager to please the president-elect—and to start a new chapter in their relationship that elides the past.

Throughout Trump’s 2024 campaign, tech executives were privately speaking with Trump about their interests and policy preferences; after the election, the public congratulations quickly rolled in. Business leaders attempting to get on good terms with an incoming administration is not unheard of. But the machinations here are happening out in the open. As my colleague Ali Breland wrote last week, “Until recently, elites and politicians who worked together feared the scandal of the sausage-making process being revealed, and the public backlash that could come with it.” Now, with Elon Musk setting a new standard for blatantly self-serving political participation—including attempts to influence the outcome of an election—his peers are operating more brazenly than they once did. Beyond the tech CEOs who are donating to and hobnobbing with Trump, several prominent venture capitalists who stumped for Trump are now advocating for their fellow tech leaders to be nominated for roles in the Trump administration. The venture capitalist David Sacks, an outspoken Trump supporter, has been named the “A.I. and Crypto Czar” for the incoming administration. And of course, the vice-president-elect was once a venture capitalist too.

Donating to any president-elect’s inauguration fund is a standard way for corporations to signal goodwill. Some tech companies, including Google and Amazon, quietly gave relatively small amounts to Trump’s first inauguration fund, according to data published by OpenSecrets. Firms such as Google and Microsoft donated to President Joe Biden’s inauguration fund. And a  seven-figure tech donation is not unprecedented—Microsoft gave more than $2 million to President Barack Obama for his 2013 inauguration. The flow of such large sums from multiple executives this year, Margaret O’Mara, a historian of Silicon Valley, told me in an email, is “both a reflection of the growth of inaugural spending generally” and “the surging profits and net worth of tech’s biggest names.” And the meetings with and warm statements from tech leaders who are donating this money signals a new chapter of cooperation between Big Tech and Trump.

The tech industry has always relied, to an extent, on the federal government, but its political allegiances have shifted. A free-market libertarian strain has long run through the region and industry, though in the 2010s, the industry cozied up to the Obama administration, a relationship that benefited both sides. During the first Trump term, the government-tech relationship became uneasy: Social platforms attempted damage control after blowback from employees and users who blamed them for Trump’s ascent to office (remember Zuckerberg’s national listening tour in 2017?). As Trump enters his second term having received close to half of the country’s vote, support for him may not risk that same level of public outrage: In many circles, the Trump taboo is over. As O’Mara put it, the social consequences of supporting Trump are lesser, and the business risks of crossing him are higher.

When Zuckerberg visited Mar-a-Lago on the evening before Thanksgiving, he and other guests reportedly stood with hands over hearts while listening to a recording of the national anthem sung by people accused of January 6–related crimes. Whether Zuckerberg knew who the singers were is unclear. But the scene was uncanny given that January 6, when it happened, was a bright-red line for the tech industry. Facebook, Twitter, Instagram, and Twitch banned or suspended Trump, and companies such as Amazon paused donations to election deniers. Now, with the arrival of Trump 2.0, that red line has been erased entirely.

Related:

Even the Koch brothers weren’t this brazen. What’s with all the Trumpy VCs?

Here are three new stories from The Atlantic:

Maybe Democrats didn’t do so badly after all. The technology that actually runs our world The push for puberty blockers got ahead of the research, Helen Lewis argues.

Today’s News

At least two people were killed and several others were injured by a shooter at Abundant Life Christian School, in Madison, Wisconsin. The suspect, who was found dead, was a student at the school, according to the Madison police chief. Canadian Finance Minister Chrystia Freeland resigned from Prime Minister Justin Trudeau’s cabinet. She wrote that Trudeau told her on Friday that he wanted to move her to another role in the cabinet, and that she disagreed with him on the best path forward in the face of Trump’s threat to implement new tariffs. German Chancellor Olaf Scholz lost a vote of confidence; the country will face snap elections next year to form a new coalition government.

Dispatches

The Wonder Reader: These books and movies will entertain even the most fleeting of attention spans, Isabel Fattal writes.

Explore all of our newsletters here.

Evening Read

Illustration by Pete Gamlen

What If You Just Skipped the Holidays?

By Faith Hill

For years, ahead of family holiday gatherings, Alicia Dudley would wake up anxious. Since she’d gotten married, her relatives and her husband’s had wanted them at multiple different celebrations for each occasion. Bundling up her small child and toting him about was a pain. Dudley, a creative director in Virginia, couldn’t believe that on her rare, precious days off, she was doing what she always did: running around.

Eventually, she made a simple but major decision—she quit the holidays.

Read the full article.

More From The Atlantic

Has your cat closed its rings today? Oliver Sacks’s lifelong search for recognition Conor Friedersdorf: How to move on from the worst of identity politics The health-care system isn’t hopeless. How Democrats lost their way on immigration

Culture Break

Rosalind O’Connor / NBC

Have a laugh. This week’s Saturday Night Live highlighted Chris Rock’s talent for trolling, Shirley Li writes.

Watch. Spend some time with six binge-worthy movie series recommended by The Atlantic’s writers and editors.

Play our daily crossword.

Stephanie Bai contributed to this newsletter.

When you buy a book using a link in this newsletter, we receive a commission. Thank you for supporting The Atlantic.