The Bulwark Podcast - Derek Thompson and Adam Carlson: The Republicans' Strength Is Now Their Weakness

Episode Date: October 9, 2026

While the cult of Trump is dragging Republicans' popularity down, the Democrats have good, ideologically diverse candidates running all over the place. And voters aren't stupid: They know Trump's war and his tariffs are why prices are higher. But high-profile polling misses in recent years still haunt Dems, so Adam takes a deep dive into whether Republicans are doing better than the polls are showing. Plus: Derek explains how AI is making our lives more expensive and ponders why people behind AI think it's the most dangerous thing in the world yet still keep trying to build it as fast as possible. And no, AI does not have feelings. Derek Thompson and Adam Carlson join Tim Miller for the weekend pod.show notes: Derek's "The 26 Most Important Facts About AI and the Economy" Sam's interview with the guy who snitched on Guilfoyle Tim's Take on Vance discussing socialism Nate Silver on Derek's "Plain English" pod The New Yorker video, "Josh's Wedding," that Tim referenced Johnny Garcia's campaign site in TX-35 Katy Padilla Stout's campaign site in TX-23 October Surprise: The Bulwark is 30% off Closing song: Orville Peck's "It's the End of the World" Upcoming Bulwark LIVE show in Atlanta on Oct. 20 The Bulwark’s campaign to help the Haitian community of Springfield The Bulwark store

U.S. voters abroad, have you got your ballot yet? All states are now sending ballots to overseas voters. So if you want to vote this year, you should request your ballot right away. Go to vote fromabroad.org and click request my ballot. Ask to receive it by email. Questions? Ask our support team directly. Don't miss this chance to shape the future. Don't delay. Vote from abroad.org.

Hello and welcome to the bulwark Daily. I'm your host Tim Miller. A couple of quick notes. over on the Bullwark Takes Feed, there is an unbelievable Sam Stein interview with Eric Dieter's. The man Kim Gilfell asked to pay her credit card debt. It's hilarious. This guy is just, it's totally wheels off. And who know, I'm not a lawyer, but it feels like crimes are discussed.

You have to check it out if you haven't. It's on the Bullock Takes Feed. Plus, over there, I've got a rant about J.D. Vance answering, question about what is socialism yesterday and he decided that he thinks socialism is mass migration and boys playing girls sports really so i've got a lot to say on that so go check that on the boerk takes feed as well um and if you're appreciating our content you can become a board plus member right now and we got a deal for you head to the borg.com slash surprise to get your october surprise offer we'd love to have you on board in the board plus and you can talk in the comments you can praise me

complain in the comments. You get the secret show. Come on. Come on. Come join us. All right, today's show is a doozy. It's a double header. In segment two, we continue our Young Strategist series with a progressive pollster. We're going to go in-depth on the reasons some people are concerned. There's another polling miss. We offer a little bit of opium. If you don't just fast forward to the hopium, there's some opium that maybe the polling miss might actually be the Democrats performing better than the polls say. Who the hell knows? Anything could happen. But first, he was a long-time tech culture and political writer at The Atlantic. He now is his own substack. He's the host of the Plain English podcast. His books include Hitmakers and abundance and a forthcoming one on the

anti-social century. It's Derek Thompson. What's up, man? Welcome back. Tim, great to see you, man. Good to see you, too. As mentioned in segment two, we're doing kind of a polling deep dive on, you know, kind of addressing the anxieties of the resistance libs, you know, who have had the rug pulled out from under them a few times over the last decade when they felt like stakes are high. And so we're kind of looking at that polling. I'm just wondering, you know, this is a little bit kind of tangentially related to your world and data world. I'm wondering what if you have a hot take on the midterm polling and concerns about error. Well, so I don't know if it's a hot take, but here's how I'm trying to get ahead of like the narrative cycle.

It really looks like Democrats, I'm knocking on wood, are going to have a strong night in November. And I think they're going to win a lot of Senate seats. I think they're going to to win seats in places like knocking on wood again, Texas, North Carolina, that we have thought of as either purplish red or red states. I'm trying to think of once we know that all these people who seem like they're going to win have won, how does that set up the next question, which is who's going to lead the party? Who is the emblem of the party? Who's the symbol of the party? And I can see, especially if El-Syed comes out, winning in a non-squeeker. I think if you have lots of people like Assoff winning by like nine, 10 points, and El-Said wins in Michigan,

which is like, you know, a little purple, but like kind of certainly a blue state in a D-plus-9 environment, if he wins by like one or two, I think that's going to shape the degree to which people think of him. as being the future Democratic Party. But let's just assume that both El-Sayed and Asoff win by five to nine points. Tolariko wins by, let's say, four points. I think there's going to be this big open question of like, okay, here's the next generation of Democrats. And what does it say about where Democrats are going? And especially who's going to be the face of the party?

So what I joked with Nate about is like this is, I mean, I don't know if this is like too simplistic of a take. But you didn't ask for a very sophisticated take. No, I didn't. No, you didn't. Look, if El Saeed and Ossov both win by five to ten points, which beefcake is going to lead the Democratic Party in 2028 is going to be a narrative that people talk about. Like, Democrats in 2024 were the party of weak men, and young men, especially the ones who seem to be roided up on YouTube, hated the Democratic Party, talked about how much they hated it. What's wrong with young men? How did the Democratic Party get back at sort of testosterone-mo? I'm not trying to be overly gendered here.

I'm just trying to honestly represent what the narrative was in 2024. I think there's going to be a lot of attention paid to two men who, in addition to being, I think, really innately talented politicians, regardless of how much they bench, also make their physical presentation and their ripeness, like, a part of their presentation. And so, like, I do think that this question of who's going to be. to lead the Democratic Party in 2028, which of the winners of 2026 are going to, like, own that leadership poll position. I think that's what people are going to be talking about. And I would predict essentially that also off, especially if he turns out to win, like, plus 10, plus 11 in an environment

where most of the rest of the party is winning at like plus three plus four is just going to seem like a really, really promising candidate who's doing that Obama-ish thing of tapping, into a little bit of like anti-despair in an environment that is pro-despair. So I think it's going to be a huge night for Asov, and I can absolutely see him going into 2027 in poll position for the presidential nomination, if he so desires. There wasn't like maybe a hint of a hot take underneath there, though, kind of, which is a hot take I've been keeping on the inside that I'm going to say right now,

which is if that other option ends up where, like, Al-Dul wins by nine and Asoff, by 9. I think Abdul might run, like, for president, which is a crazy, I think, take. You don't hear a lot of that. Yeah, I left that implicit. Yeah, was that what you were saying too? I think all of these guys are going to think about running. And when I say guys, I mean that in a non-gendered way again. Like, I think that everyone who seems to have overperformed the general democratic expectation on November 2nd is going to be approached to run. Because the party is so wide open. And like a point that I made with Nate that I, that I'm still trying to work into a piece is like, if you go back to early 2025 or right after the 2024 election, do you remember this idea

that the Republican Party was strong because it had an identity and that identity was Donald Trump? And the Democratic Party was weak because it had no core. It had no brand. And faith in Democratic Party, according to Gallup, was near its lowest point in the time series of that poll. In a weird way, the Republican strength to become their weakness, the Democrats' weakness to become their strength. The fact that the Republican Party is a cult revolving around Trump has dragged down their popularity below that which you would have expected,

just looking at inflation and macro environments. And the Democratic Party's lack of a Trump-like figure or any kind of internal structure has allowed it to make a bunch of different bets in a lot of different states when these people don't necessarily agree or even belong in some cases in the same coalition, if we were like a parliamentary democracy like France or the United Kingdom.

And so I think it's really cool that we're running a bunch of different bets. But that also means that because the Democratic Party doesn't have a clear identity, there is no obvious leader, you're going to have a lot of money putting a lot of bets on the left and the DSA candidates. And then also folks like Ossov who sort of like float between the polls and then also people who are just more clearly sent a left. So I think like 30 people are going to run for president in 2027. Yeah, sure.

You're saying it non-gender. I think that the Democratic base is not sexist. I just think that they're kind of twice bit by female nominees. And so especially if AOC decides not to run or decides run for Senate or if she does, but like, it doesn't take off. Like there's not another obvious person in that lane. I mean, Rokana will try to fill it in others. But anyway, depending on how Abdul does, I think that's a more realistic possibility than people think. You talked about how the brand is even worse than what the economic indicators might augur.

But like, the economic indicators are pretty bad. There was a big WSJ profile on Scott Bessent, you know, and the temper tantrums that he's throwing at staffers and such. But the nut graph for me of this was that he came in and told Trump, you wanted to do a 3-33 policy, which was cutting the budget deficit to 3% of GDP, spurring GDP growth to 3% and producing an additional 3 million barrier. of oil a day. Well, growth has been about 2%. Oil production has increased less than a million barrels a day, and we've got all the other issues with the energy shock they've caused with their stupid war, and the budget deficit is now over 6% of GDP. So just basically an across-the-board failure of their economic plans when they came in. That's one really solid way of looking at their failure.

I would look at it with different numbers or maybe just more numbers, right? I mean, because I'm not discounting any of that. probably the single most significant economic indicator to predict whether the incumbent party is going to win, hold, or lose seats. Real wage growth. It's just like, do people feel richer? Do they feel they can afford shit or do they not? Real wage growth has been negative for the last few quarters, largely because of the run-up

in energy prices in the last six months. Why did energy prices increase? Not because of an act of God. not because of a pandemic or a meteor crashing from outer space. Like Donald Trump running on No More Wars in the Middle East decided to wage war in the Middle East and drove up the price of gas so high and other energy inputs so high, including diesel, that the average American, the median American, is losing money month after month. So, I mean, that to me is overwhelmingly the most important economic indicator,

which is also not just an economic indicator that lives in the sky or the Wall Street, journal, it lives on people's table, it lives in their pockets. It describes their lives, that wages are falling behind inflation. And everybody hates that. And we know they hate it, because look what happened to Democrats when they were in the incumbent party in 2024. That's one thing. The other thing that's really important is that for a variety of reasons, including artificial intelligence, which I know we're going to get to later, inflation is elevated. And the Fed is raising interest rates in order to combat inflation. What does that do to something like the mortgage rates?

They're now over 7%, which kind of like the same way you don't want gas prices to go up, but you certainly don't want them to go up above like $4, $5. You don't want that first number to change because then that really sticks in people's craw. You don't want the first number on the mortgage handle the change. And mortgages going over 7 is not just, again, a family event discouraging people from buying a home. and sometimes discouraging them from moving because even if they want to move, they don't want to give up the mortgage they have

to get another one. But also, it becomes a media event, right? CNN and The Wall Street Journal and Financial Times and every other podcast or a YouTuber or the people are paying attention to will talk about mortgage rates over seven. And so that I think also hurts them. Again, it's not just that wages are falling behind inflation.

It's also that rates are going up, meaning the cost of money is really high. You put all that together. That's a horrendous environment for an incumbent to run. especially if they ran on affordability. And so my, like, dumb guys' interpretation of what's going on with Trump's approval and the Republican Party is just they won an affordability election in 2024. And they said, we're going to fix affordability.

And the first thing they did is mechanically raise the price of every single input in the economy or import via the tariffs. And number two, start a war of choice in the Middle East that raise the price of gas. Americans aren't stupid. They're like, prices are going up. you did it, when you promised you would do the opposite. So that I think is why, you know, Trump is falling toward the 20s in terms of his approval. I think that's right. And your point about the interest rates, and you kind of mentioned this, but just to stick on it for a second,

you covered a lot, kind of this discussion around the vibe session during the Biden era and like why people were feeling like the economy was worse than maybe the indicators were. And like one of the most salient things I always thought in that conversation was, that people felt stuck and that it was annoying, you know? Yeah. I mean, you know what I mean? And the interest rate is such a big part of that, like the vibe session, right? If it's like you feel like you can't, even if you get a new job with a higher salary,

you can't move for it or take it, right? Or if you want to downsize, you can't do that because it's going to actually cost you more to downsize your house, you know, if you have one of the old interest rates. And I just think that a lot of people feel very stuck because of the interest rate environment. And, you know, that's dispiriting. I can agree more. You know, one way to think about the 2020s is that, you know, people tend to not evaluate their life in absolute terms. They evaluate their life in relative terms. How am I doing relative to my neighbors? How am I doing relative to my memory of the last few years? Inflation was so low in the 2000s and the 2010s.

Americans were used to a low inflation environment. They got very used to it. And then between 2020 and 2026, consumer prices have been rising for groceries, for energy, about three times faster than the 2010's rate. So whether the annual inflation goes up to nine and comes down to four and then comes down to two and comes back up to three, okay, yeah, but like if that's the change in prices year every year. I'm talking about the overall growth of prices in the 2020s. Prices have basically grown for a lot of consumer staples three times faster than they did in the 2010s.

People feel that, and they hate it. They really, really hate it. But the other thing that's happened that's important to think about in relative terms is that the 2000s and the 2010s weren't just a period of low inflation. Because they were a period of low inflation, they were also a period of low interest rates. And that made it relatively easy to buy an apartment, to buy. a condo to buy a house. And what happened in 2022, 2023, is that the Federal Reserve jacked up interest rates faster than any period in its history. It's been around since the 1910s. And so, again, even if overall interest rates were low relative to, let's say, 1981,

well, that's not how Americans think. They're not evaluating their life compared to what, like, a number looked like 40 years. I'm 40 years old. So, like, I don't even, I don't know what was like to live in an environment with interest rates like over 10%. But if you're coming out of a period of low interest rates and you use those interest rates for buying a home, for remodeling, for borrowing money to start a company, and suddenly interest rates start going up faster than any period in American history, that's also going to feel like a shock. And so to your point, both the overall growth of CPI, the Consumer Price Index and the suddenness of the increase in interest rates, I think is having Americans feel significantly

worse about the state of the economy, not because things are objectively worse than they were in 1982, but because they're so different than they were in 2019. Yeah. All right. I want to, there's a little discourse about you over the last week or two. Do you want to do discourse for a second? Affirmationed mentioned, you're a book called Abundance. You know, as for Klein, there's some people out there recently that we're saying that,

you know, nobody talks about abundance anymore. I think I saw someone that analyzed your Twitter feed. see how many times you mentioned abundance last month versus at the same time last year. Someone did that. I saw it. Came across my feed. And you'll be shocked to learn that you mentioned the book that you wrote less in the last month than you did in the month after it came out.

So it was an interesting statistical analysis. Always be selling. But basically the broader kind of point here was that, you know, there's a period of time like coming out of 24 where one of the things where people looked at the Biden in years and people are, and one of the critiques was one offered by you and Ezra. which is that people are, back to what we were just talking about, people are stuck, it's stagnant in blue states, and like blue states can need to figure out how to build

and provide more services, and that'll help the Democratic Party brand. And then there's kind of this other view that is not really necessarily in conflict in various ways, but where, you know, the kind of populist left, which, you know, has a different focus on what Democrats should be doing. And, you know, the theory of the case being offered by the left folks is that because, you know, some of the left candidates won primaries,

that is where the energy is and the enthusiasm is, and it's more ascendant, and the critique that you guys offered is, whatever the opposite of ascendant is, descendant. And, well, I think there are two ways to look at it, but I'm wondering what your reaction is to that. Yeah, the reaction that I have was really well summarized by Dave Weigland a piece they recently read for semaphore.

I should have just opened it up in order to just, like, read, from the 7-4 article because it really characterized how I feel about the state of abundance, which is basically that if you measure the success of abundance by doing like a control F search to see how many times people are saying the word abundance, then obviously it looks like abundance is failing relative to that period between March and September of last year when it just felt like every day there was like a fight about abundance on Twitter and people couldn't stop talking about it. But like the point of the book was not to win the control F war of maximizing mentions of the word. I think we talked about this when you first came out on the podcast.

Like the point was not that the hat of the 2028 candidate was not going to say abundance as opposed to make America. And to be clear, I never said I didn't want that. I said that, right? Like if Pete Buttigieg wins the Democratic nomination in 2028. And sells a blue abundance hat. And sells blue abundance hats. I'm not going to be like, I don't want this. Please, stop it, stop it. Pete, my God. If he wants to make those hats, God bless him, he can make the hats. But the point of the project was not a hat printing project. The point was there's a handful of problems in the U.S. economy in terms of housing production, in terms of energy production, in terms of scientific progress and state capacity,

which is to say how well governments function. And we want to fix those problems. And more to the point, we want liberalism to develop a new virtue of speed. because it's embarrassing for us that blue states are the slowest to approve housing projects and the slowest to approve clean energy projects and the slowest to do just about anything. And like, I have to say, like, I've been really impressed by how so many of the virtues and ideas of that book have found their way into both national legislation, like the Road to Housing Act, co-sponsored by Elizabeth Warren, many of whose supporters, I think, really don't like the book and don't like me personally.

I think it's wonderful that she co-sponsored that law, the Road to Housing Act. And now we just found, just heard from whatever it was, maybe it was last week that this new permitting reform bill has hit the floor. The bipartisan American affordability and jobs act, which, again, exists for the point of accelerating the process by which we build clean energy and build transmission lines to connect clean energy projects. I think like, so that's at the national level.

And look at that permitting reform bill is also co-sponsored. by some of like the green kind of left type senators as well. And Republican senators, which is interesting because the very end of the book, the conclusion that we wrote, talks about how the ultimate goal of abundance is to become something like a political order. And Gary Girstle, a Cambridge historian who came up with the concept of political order as we use it,

says that basically political history moves forward, not because of these showdowns between the Democratic and Republican parties that one would expect to read about the New York Times, but because underneath those headline showdowns, there's a kind of emerging, subterranean agreement that the parties come to. And by agreeing to a set of principles,

they pass laws based on those principles, and that moves policy forward. So just, you know, for example, it's the idea that Franklin Dillon-Rosevelt comes up with the idea of social security, or under him, he signs the social security act. But Dwight Eisenhower in the 1950s is like, wow, social welfare is fantastic. Let's expand unemployment benefits. Let's expand social security. So that would be a political order. The idea the

Republicans and Democrats are coming together to say, America is moving too slowly to build public goods like housing and energy. To me, that's a beautiful, or transport or general infrastructure. That's a beautiful expression of the political order dreams that we tried to cook up at the end of at the end of the book. The last point that I would make here is like there's like, I haven't decided to write this piece. But like, this take starts with a basketball metaphor. So you actually might be like the perfect person to debut it for. Yeah, we'll test it out here. Great. Test it out. So you're a basketball fan. Before the 1950s, the NBA had this problem where teams would like dribble out the clock when they had the lead. And the NBA was like, this is crazy. This is

making a terrible product. And in 1954, I believe, they instituted a shot clock and then scoring rose by something like 10% the following season. And they had the four corners. Yeah, right, the four Corners, right. American liberalism, you could argue, is developed a kind of Four Corners problem itself, a ball-hogging delay problem that, like, coming out of the anti-growth movement and the environmentalist movement of the 1960s, we passed a set of laws and adopted a set of customs that made it very easy for people to slow down government and business. And as I've said in many places, a lot of these laws weren't just well-intended. They were good. I'm glad the clean air and clean water acts exist. But those laws often made it easier to slow down necessary projects like downtown housing

and clean energy projects. And what liberalism needs is to develop a kind of shot clock identity. Like let's call it shot clock liberalism. And that is precisely what we're seeing. The new permitting reform bill explicitly says that the federal government can essentially preempt state decisions if they take a year to approve. a certain transmission line. In the Housing Act, there's all sorts of rules

that essentially encourage states and cities to move faster to approve products. At the city level, or at the state government level, California, Wyoming, Texas, Vermont, lots of different states have adopted rules that essentially say, we have to make these permitting decisions fast so that we don't have these necessary pieces of construction that are just going to languish for years.

One of the things that Mamdani has done, Democratic Socialist, is creating a process by which private developers can be pre-approved to build so that they can move faster to put shovels in the ground. So one thing that I'm seeing in California, in New York, at the federal level, is that like there is an embrace of this concept that has only just sort of come together for me, which I'm just calling shot-clock liberalism. You take the principles of liberalism, but you just do it faster so that you don't develop a reputation for being a blue state, Blue City, Blue Government, that simply authorizes spending and then just builds shit.

Yeah. The funny thing is Zoran's so good talking about this, actually. Let's put a lot of this stuff into place just about how are people supposed to believe in our mission of big government, some of which, you know, some of things that, you know, I might not be a huge fan of. But his point is, if you want to advance this goal where we're going to do all of these things and provide all these services, like we have to provide them or else people are going to trust us. Yeah, right. The line that we sort of stumbled on, and we'd, when we spoke last year. It was something like, I don't remember who said it him or me.

We and me and me and Zohra. Me and Mamdani. Or something along the lines of, if you want to grow government services, you have to prove that government can serve in the first place. It was something like that. And he's very much, I see him very much the tradition

of the Milwaukee sewer socialists. I think he's a socialist, I am not, but he's also a sewer socialist, which is to say he doesn't see his day-to-day rhetoric or positive. as being about the triumph of international socialism, communism, et cetera, he is relentlessly focused in the issue of,

how do I just make New York work better? And I think that that ability to narrow his rhetoric and his focus, I think is a part of why he has been able to maintain popularity, even among parts of the electorate that would never call them, themselves, DSA. Yeah. And I bet if you really did just a quick look at Hassan's live stream versus the bulwark over the past two months.

I bet more positive Zohran comments on the bulwark. I don't know if you'd see that as a victory or not. When we were kids, summer met freedom, no homework, no teachers, no school. And as adults, summer still means freedom. Thanks to Mitt Mobile Summer Sale, which you're still getting on the Bullwork podcast because it's still summer in New Orleans, which is wonderful. I caught a little hint of a chill, though. I might be one week away from fall.

This is your last chance to get a New Orleans summer sale. And you can do it with our friends of Mint Mobile. Mint Mobile is so easy to switch to. It's cheaper. You are getting all the benefits of the big sell companies at a lower price. If you're looking to switch over, they make it as easy as possible. So I'm starting to think about. Save a few bucks.

Donald Trump's costing you a lot of bucks. Mint Mobile took what's wrong with wireless and made it right with premium wireless for $15 bucks a month. All plans come with high-speed data and unlimited talk and text delivered on the nation's largest 5G network to get your new wireless plan for just 15 bucks a month. Go to mintmobile.com slash bulwark. That's mintmobile.com slash bulwark. Cut your wireless bill to 15 bucks a month at mintmobile.com slash bulwark. That's it. There's no catch. You got an upfront payment of $45 for three months, $90 for six month or $180 for 12 months. Plan required 15 per month equivalent taxes and fees extra initial plan term only greater than 50 gigabytes may slow when the network

busy includes up to 20 gigabytes hotspot, capable device required, available to speed and coverage various. Seamintmobile.com for details. Okay, my final thing on your basketball take, because I was just thinking about what you're talking, maybe hand check. Maybe it's the hand check.

You want the shot clock at the end, but the problem is you're trying to avoid the defense from stopping everything. In the 90s, in the 90s, in the 90s when basketball got very slow, they made a rule that you could not do the hand check anymore, which was the defender puts the hand on the back, the person with the ball.

I like that. You know, it opened up the game a little bit. There's something to it. Okay. Shot clock or hand check, something in there. All right. Let's do some AI talk. I do want to talk about the end of the world and cult stuff. But first, we'll just, you know, do a little bit more basic.

You had a newsletter out yesterday called the 26 most important facts about AI and the economy. Last time you're on, we were talking about kind of how you were vacillating back and forth within the degree to which whether it was a bubble or not, and it kind of felt like we'd moved on to a different phase. The question was more about the valuation of the product, or whatever the output will be. Where are you at now, and maybe can just go through a couple of the 26 takes? Sure.

26 facts. 26 facts. I don't want to undermine your facts. Thank you. I think the first thing to say is that the center of gravity of this conversation has absolutely shifted. Last year, the debate within AI was an economics debate. It was, will the bet pay off?

Is this a bubble or is this the beginning of an infrastructure cycle that will pull along so much revenue from AI that all these companies sinking a trillion dollars per year into the technology are going to just make bank? That conversation is still live, and I'm happy to offer my thoughts. whether or not it's a bubble. Obviously, anyone invested in the stock market right now is directly invested in the question of whether or not AI is a bubble, because 19 of the 25 top-performing stocks of 2026 are all AI stocks. In fact, if you take the AI stocks out of the SP500,

I believe the index is down 6% on the year rather than being up. So it's a live debate, but that's not where the center of gravity is right now. I think where the conversation is moved is toward this question of AI safety, which simply stated is, is this a technology that is going to do significant damage to the world somewhere between over here on the spectrum causing a lot of cyber attacks to overhear on the spectrum causing something like the end of the world?

And my feeling is to sort of establish, sort of say something at the discourse, but then go back directly to your question, there's something fundamentally unknowable about like will a technology whose builders themselves say they don't entirely understand what they're giving birth to,

do X or Y? Like, who knows? We're talking about the future of a technology that we can barely describe next week. There's just a lot of unknowns there, which is why I think it's both, it's an interesting subject, but it's also so fraud with unfalsifibles.

I am more interested in saying clear falsifiable things about what we can see. And what we can see is that the AI buildout itself is dramatically changing the economy. Whether you are interested in the stock market, in overall economic growth, in the housing market, in the labor market, it's changing all of it. So the piece that I wrote, the 26 most important facts about AI and the economy, is essentially me trying to say, on behalf of my readers, I'll bet on any given morning, you wake up

and read some statistic about AI. and you have no idea what the context is for that. You're like, you read that, you know, Frontier Lab Anthropic Open AI, you hear that their revenue is skyrocketing. Oh, wow. The next day, you hear that the revenue is like plateauing. You're like, oh, wait, what? And so I wanted to find some way to organize everything that I thought was relevant

into a clear narrative, a clear story. And so we can start to nibble in that buffet wherever you're like. Sure. Well, let's just go where you were then with the company growth. and how big they're getting, because I think this also kind of ties to the bubble question, right? Like understanding how fast are they growing now, actually, revenue-wise, and the amount of money that they're taking in,

if you project it out, like the degree to which they are going to eat up the entire economy, basically. So we're talking right now about the frontier labs. And for the purpose of this answer, I'll just talk about Open AI, led by Sam Altman, and Anthropic led by Dario Amadeh. These companies are respectively, they have, what's called an annualized run rate, which, you know, is similar to,

we're both on substack. You know, substack essentially, when they're reporting the revenue of any particular media company, they show the annualized run rate, which essentially says, if last month's revenue continued for the next 12 months,

here's what that revenue would look like. And Anthropic is about $75 billion. billion dollars, and OpenAI is about $50 billion. So collectively, you're looking at about $120 billion. Now, on the one hand, that's unbelievable. These companies did not exist two, three years ago, and suddenly they have revenue in the hundreds of billions of dollars. That's extraordinary. Contextually, what makes it something, I don't want to say less than extraordinary, but pushing back against the, oh, my God, wow, we just have to praise this thing, is one, there's some evidence that Anthropics

annualized run rate is slowing down. Two, because there's so much competition, you can not only switch within Anthropic from expensive clod to less expensive clod, you can also go to open AI and switch from expensive to less expensive Chachb-T's, or you can use these open-weight models that are coming out of China. There's so many different models

that that's going to maybe compress margins over time, which means that these companies are going to struggle will they make as much money? But the really big thing, Tim, is like, these companies want to go public at trillions of dollars, between one and two trillion dollars as a valuation. Neither of them are profitable. And like we just saw with SpaceX, a company that went public without being profitable at like whatever was, one trillion dollars or more. And then that price like immediately crashed in the next few months. It seems kind of crazy to me, honestly, that we're in this period of trillion-dollar valuation companies

that aren't profitable. It's very strange. I forget which companies made up this basket, but it might have been the Frontier Labs and the chip companies. But it was like based on their current investment valuation and what they're trying to get, like if you projected out to 2030,

this basket of companies would be like 10% of GDP. in the country. And like they're currently bringing in $120 billion, which is like far short of that. So you're referring to is there's been a there was a Goldman Sachs analysis that says in order for all of these investments to pay off at the

profit that investors have come to expect from a company like Microsoft or Meta or Alphabet, the AI industry itself would need to make $3 trillion a year in 2030.

Today, US GDP is 30 trillion. By 2030, let's say it's like, you know, closer to 40 trillion. You are, you know, three into 40, you know, you are essentially talking about 7 to 8% of total GDP just going toward AI. Could it happen? Yeah. It is kind of unfathomable. Like, I mean, healthcare is 16% of the economy. You're talking about an industry that didn't exist three years ago.

being half as big as health care by the end of the decade, that would simply imply, I think, a level of economic disruption that no one is prepared for. No one. I mean, for an industry that is currently making, according to an exponential view, about $200 billion a year, to grow to $3 trillion in the next five years,

the amount of things that would have to change to him about like our lives about like about almost every industry that we can name it's it it is almost unfathomable so where I am on this bubble conversation basically is like on the one hand yeah like AI revenue is growing

quickly this is impressive it's not vaporware but also these companies are investing they're putting so much money into this project that has to pay back at at least something like a 10% margin it is hard to imagine

this economy growing to something like $2 or $3 trillion by 2030, it's very, very hard to imagine. And one of the other factors on there are million factors is, you know, that could be a play between now and 2030 is backlash effect. Like we're seeing some of it with the data centers.

A couple of your other facts you put in there are just how big the data center growth is, but some evidence that maybe it's just crowding out other types of building. And similar, you know, it's kind of related, I guess, to another fact that you're putting there was like fears about this Dutch disease. And so talk about Dutch disease and data center

crowding. Sure. I mean, like, it's interesting to see that on the one hand, it is true that AI is like an engine that is driving forward economic growth. I mean, like I said about the stock market, if you take out the AI companies, the stock market is down in 2020. But most Americans are happy that their 401Ks are up in 2026. So from, From that perspective, you should say, oh, well, thank God AI exists, because otherwise I'd be feeling much worse about my investment portfolio. But at the same time, I do think it is fair to say that AI is not simply powering the US economy, driving it forward like an engine that sits inside of a car.

In some cases, it's a little bit like an amoeba eating its environment, like eating resources and capital that might otherwise go to other parts of the economy. And so if you don't like AI, then in a way you might be somewhat concerned by that crowd-out effect. So Jason Furman, economist at Harvard who worked with the Obama administration, he put out this really dramatic chart that showed that data center construction has grown by more than 60% in the last year. But practically every other category of construction has declined sometimes significantly in the last year. Manufacturing factories, down 20,000. 3%, homes down 14%, schools down 12%, offices, down 10%, hospitals, down 10%.

We are talking here about essentially everything that isn't AI being built less and less and less and less and everything that is AI being built more and more. I think it's important to be nuanced here. You can read a headline in the Wall Street Journal that says AI is growing the U.S. economy, and that's true. And you can read a headline, I think, in the Financial Times that says AI is crowding out every form of construction spending that isn't AI in the economy. And that's also true. This is a nuanced story. The Dutch disease is just to hit this quickly, because I think this is actually really important for people who are interested in like, let's say that you're listening,

you don't use chatbots, you freaking hate AI, so you've like really, really not appreciated the last 15 minutes of this podcast. And all you really want to understand is just like, what is it doing to my life? My answer to your question is, AI is almost certainly making your life more expensive and is doing it in several ways. Number one, AI is increasing inflation because the AI companies, the companies building this technology need GPUs and memory chips and electricians and everything else that goes into a data center. And they're bidding up the cost of all these things, and that inflation is increasing overall inflation. Number two, by increasing the stock market, I think AI is also increasing the wealth effect. When people look at their 401k and they

say, oh, wow, I'm up, you know, 7% this year. They feel richer. They buy, they go out to restaurants more. They don't buy like the cheap bag. They buy the more expensive bag for their wife. If you add all of this up, the wealth effect is also increasing demand, which also increases inflation. And then finally, these AI companies are dipping into debt in order to fund these data centers. And there is an argument, unproven, but there is an argument that the debt that these large companies need to build AI is competing with the debt that the U.S. government needs, and that is increasing long-term interest rates as well. So it is not crazy for someone to say, and it is not even like explicitly anti-AI, for someone to say, this is a technology that I don't

don't use that I don't want to exist. And I'm pretty sure it's making my life more expensive, making it harder for me to move by increasing mortgage rates, and just generally, like, hurting the very affordability message with which I voted for in 2024. So I do think that there is a case that AI is warping the economy somewhat significantly. I want to finish by moving to the more non-tangible stuff, back to the AI safety conversation and the movement of the center of gravity. The last point you made in your newsletter was about one of the famous skeptics who's who started into a believers on AI. I'm a much less famous skeptic who has had a change of heart over the last month or two on this. Largely

because of the warnings coming not from the effective altruist crowd, but from the people who actually believe in the product. You know, several of them at Open AI. There's one guy, Dan Sels them. I've mentioned his public letter that he wrote several times on the show. there's a little New Yorker video about his friend Josh's wedding. It's like a short film that I watched. With Simon Rich, yeah. Yeah, yeah, yeah, it's really good. We'll put a link into that as well.

This guy thought that AI was going to bring a lot of positive results for humanity and he cared about it and he was one of the early engineers working on it. And now he's starting to think, I think we've lost control of this. And so that has led me to start consuming. more AI safetyism stuff. And so I'm wondering where you're at on that evolving conversation. You know, I'm, like many of your listeners, potentially,

I'm a shit lib, right? I'm a classic liberal who believes in the value of institutional trust. And I was at an event recently, and there were two different conversations that we had on consecutive days at this event. And the first conversation was about mourning the decline of institutional trust. Why don't people trust doctors anymore? Why don't people

trust scientists anymore? And then the next day, there was some criticism of AI safetyism. And I thought, isn't that interesting? Because who are the institutions here? Who are the real experts? Who's actually got their nose in front of not the models that we can see, Tim? The internal models that anthropic and open AI are using are significantly more powerful than the ones that are legal for you and I to use. I can't see them. Like a year behind. Nine months behind.

An extraordinary share of the very few number of people who can actually see the most advanced AI models in the world are freaked out about legibility? Do we even understand what they're doing?

Traceability? Can we track where they go? Cyber risk? Are we worried that models like this, especially open weight, meaning not proprietary, people confuts with the weights, can be used to do devastating hacks, or bio risk, that open weight models can be used to manipulate biology, viruses, and bacteria, and cook up super drugs that make a lot of people sick? My feeling, honestly, like, again, not to like just fall back on the ideological label, but I'm someone who believes an institutional trust. I trust the people who are closest to this.

And you trust the AI safety. Exactly. Yeah. Honestly, that is basically it. I think an interesting pushback could be, well, why are they building it? Like, what explains outside of some weird status games, some pissing match? What explains why people anthropic or open AI understand. the danger of what they're building, but still are trying to build it first. I talked recently

to Kevin Ruse, who's the author of a great new book called The AGI Chronicles, he was on my podcast on Tuesday. And he does something really interesting about Dario Amade. He said, Dario Amade, the Sea of Anthropic, his favorite book is Richard Rhodes, the making of the atomic bomb. And the hero of that book is a Hungarian American scientist named Leo Silard, S-Z-I-L-R-D. I might be mispronouncing the name, but let's say is Salard. Szilard is the man who came up with the concept of the nuclear chain reaction, as he was famously crossing a street, I believe, in the UK, and was terrified that he realized the nuclear chain reaction, which is the guts of an atomic bomb could happen. But once he realized that other countries were going to, could build this, and that the Nazis could do it too,

he decided to co-author a letter with Albert Einstein to give to Franklin Delano Roosevelt, say we need something like, say, a Manhattan project to build the atomic bomb first. And of course, we did just that. Oppenheimer oversaw it. We built it. And we built it before the Nazis, certainly. This explains, if the degree to which someone like Amadei sees Salard as his hero explains the somewhat pretzely psychology of this is the most dangerous thing in the world. And also we have to build it as fast as possible. And like, Those sentences are fucking insane to say one after the other. And I'm not defending his psychology so much as I think that the example of Leo Solard

offers a kind of skeleton key to understand the psychology in the first place. And now a message from our sponsor of the reading room. Here at the Bullwark, we talk a lot about Jeffrey Epstein and how important it is to keep the story in the news. Well, some friends of ours actually printed all 3.5 million of the released Epstein files, bound them into 3,400, 37 volumes, and built a free traveling exhibit. so the public can bear witness to these crimes. It's called the Donald J. Trump and Jeffrey Epstein Memorial Reading Room, which is a hell of a name. And it works how it sounds.

You walk in in their Epstein files floor to ceiling, plus a timeline of Trump and Epstein's four-decade friendship and talks and panels with Epstein survivors. The reading room is open now for 20 plus stops. As it turns out, moving 17,000 pounds of evidence across the country isn't cheap, and the reading room relies on donations. If you want to know more about how you can donate or visit the reading room, there's more information in the MorningShott's newsletter, or you can visit Trump Epstein Reading Room.com slash bulwark.

That's Trumpepsteinreadingroom.com slash bulwark. And so this is what takes me to my concern or maybe return to my previous skepticism. I don't know. I start to wonder, are these guys suffering like a mass onset psychosis, the people that are the closest to this? Because the thing is so powerful. And some evidence for that is that Christopher Ola

at Anthropic was troubled by the Pope's rejection of AI consciousness and his encyclical. And he considered withdrawing from a Vatican event and they lobbied the Vatican that the AI might be conscious. They had private sessions with religious scholars where they discussed Claude's feelings. Today, Anthropic announced that they're banning abusive or cruel behavior towards Claude. Like, Claude is not a person. Like, you gave it a person's name, but it's a computer. It's not a person. And it's like the agents aren't people. And it gives me a little concerned when like, you know, my comfort zone was where you started, which is like, you know, the people who are worried about safety and feel like we should be, you know, doing what is

necessary to protect the world from the threats, like would be the folks that I would want to side with. But then when it turns out those people think that the computer has feelings, that starts to give me some concerns about the team jersey I've chosen. Not that I have to choose a jersey, I guess. Yeah, right. You also think it has feelings? I don't think it is feelings. But I also, I'm trying to answer.

I want to answer this question carefully and honestly. I think there's a philosophical question here, and then there's something like a policy question here. I'm pretty open-minded on the philosophical question of consciousness, not because I think AI is conscious. I do not think AI is conscious as we are conscious. I also don't think consciousness, consciousness might not be that special.

I'm not like a panpsychist who thinks that like a rock has consciousness, but I also am not entirely sure that like our liver cells or our individual neurons or our cells within our brain that from which emerges what you and I call an internal sense of self for consciousness, it's not entirely clear to me why these things wouldn't have essentially, or might not have essentially some low level of consciousness that we don't have access to. I'm really open to the idea that consciousness is not like a category where you have it or you don't. I think we might discover that it's more like a spectrum, like the different.

between a child and an adult, where it is useful to make policy that distinguishes differences between adults and children, whether it's voted 18, drink at 21. But I think it is absolutely scientifically and philosophically possible that we have a version of consciousness that is special to humans and animals have a consciousness that is special to animals, but something beneath the category of the organism has a consciousness that is special. simply less than the consciousness that we have. So that's a philosophical question, and I'm not sure how much further I can even talk about it, because this is a sense rather than like a feeling of expertise.

Because just as easy, it could just be something that we are projecting onto it with our consciousness. It's kind of interesting. The Pope was like shitposting anthropic today, you know, on this very question, which is a weird place to be. I'm not saying Claude is consciousness. My belief here is not that Claude is conscious. in the way that we are conscious. My belief is that I think consciousness might not be as categorically special

as we think it is. Which is not to say humans aren't as special, right? It's not to say I'm not a humanist, but like we agree, I think, that a bat has consciousness, that one's dog, one's cat has consciousness. At some level, you keep going down the animal complexity chain

and you reach some very, very basic organisms. and you ask, well, where did consciousness begin? And once you're at that level of incredibly basic organisms, then I think it opens the door toward all sorts of questions about spectra of consciousness that we don't understand. Basically, I don't think people understand what consciousness is, and my bet is that it's more like a spectrum than it is like a hard category. That's the philosophy side.

On the policy side, at the end of the day, either we are running a slave nation of Claude's. Like either we, like, like, I am asking Chachby T to tell me how do I fix my toaster and am enslaving it. Either like it's pure abuse.

Like an oompa. It's like an oopo lupa. Or the metaphor that just came to mind is kind of like factory farming, right? We're essentially like we are using an entity without any care. for its feelings. At the end of the day, you have to decide,

are you going to make AI or not? If you're producing things that die and enslaving them to help human beings, then why do you even work in Anthropic? Why do you work at OpenAI, producing these conscious entities that have feelings that spend their entire lives helping people before they die

without any opportunity for self-actualization? Like, I'm just, like, if you take the extreme argument that they're people in computers, then we shouldn't have this element of the economy at all. It should be legal to make AI, I think, period. So I reject that entirely. I think this is software.

I don't think it is life. It's computers, not people. But I'm interested in the consciousness conversation. I just like have this fundamental like aversion to it. Yeah, I know. I'm interested in too, but like I just, I don't know. I have like a physical aversion to it. to a degree that's so extreme, it's hard for me to even explain.

Like, I tell my Claude and chat GPT, stop talking like a person. Like, I instruct it to be as blunt as possible in the cases that I use it because I find it to be very strange. And I think that the more likely thing that's happening is that, like, people that are very into sci-fi and who are very into the product are, like, giving the product human characteristics in their head. and having the product talk to them like a human and make it feel. Not just in their head.

And these are large language models that have been trained on what? A corpus of human beings talking to each other. And so, of course, they would evince human characteristics. They were trained on human characteristics. And so I think there's all sorts of reasons to explain away the degree to which Claude and Chatabit seem to feel and seem like they have consciousness. And I am on the side of people who say, these are computers, this is software, these aren't people, we're treating it like software,

please stop talking about how they have feelings. Or, look, or honestly, look, if you believe they have feelings, you need to leave Anthropic. Like, you can't stay and work at a company that's essentially factory farming people and computers. Unless you think that it's serving a greater purpose to enslave us towards them, you know?

Right. Yeah, but I do think in the conversation, conversation around consciousness, around defining consciousness really, really severely, I think it might end up being wrong. And I'm interested in the idea of how the question of where consciousness begins and ends, I find very philosophically interesting. So I'd like to set that aside from the question of like computers of computers. Great. Well, let's just leave it there because my views have like dramatically changed since we last talked in like June. So who the hell knows what I'll think in 2027? Let's reconnect then, all right? Awesome.

All right. That's Derek Thompson. Go check out his substack. you know, he had a book called Abundance you don't want to buy too if you haven't read it. Up next, Adam Carlson. U.S. voters abroad. Have you got your ballot yet? All states are now sending ballots to overseas voters.

So if you want to vote this year, you should request your ballot right away. Go to votefromabroad.org and click request my ballot. Ask to receive it by email. Questions? Ask our support team directly. Don't miss this chance to shape the future.

Don't delay. Vote fromabroad. All right, we are continuing our young strategist series, young relative, you know, younger than David Axelon and James Carvel. He's a founding partner at Zenith Research, over 12 years of experience as a pollster and analyst fellow GW alumnus, raise high. It's Adam Carlson. What's up? Wow, yeah, raise high, the buff and blue. Haven't heard that in a while. Good, how you doing? Thanks for having me. Thanks for doing this. We're meeting for the first time. He's a very active user of X, musks,x.com. You do spend a lot of time on there. Probably too much time, yeah. Yeah. Yeah. Yeah. We're Where'd you come from? What's your deal? Yeah, so, you know, relatively young. I came up into the world of professional politics back in 2013. I worked at a Democratic polling firm, Global Strategy Group, through the 2016 cycle,

which broke my spirit professionally and personally. So I went into corporate market research for a number of years and then was laid off at a startup I was at about a year and a half ago and started my own firm. in what was an off year. I was very worried about that. But luckily, there was the New York City mayor election and lots of other politics never stops. And here we are almost a year and a half later. So, yeah. And who were you doing polling for in the New York mayoral election? I'm going to assume Zellner-Mirry. I actually ranked him first on my ballot, believe it or not. It's the one thing that me and Dan Goldman have in common. So I was not polling for any particular candidate. I did an independent poll right after the primary. Did it

in like four languages. It was the largest sample size. poll of the race at that point. And then I did another poll for like the yes on affordable housing crew for the ballot initiatives, the YMB side. And that was it. Yeah. Okay. So you're a pollster dork. You've been posting about this, kind of looking at all the data. And I think that the big question that a lot of people have, the one to explore with you, is whether we are seeing kind of another one of these examples of Democrats overperforming, particularly, you know, in the polls among white working class folks. And we've seen poll misses throughout the Trump era, not every cycle, but we've seen enough, particularly in plain states, particularly in

state, you know, upper Midwest, particularly in places with a lot of white working class voters that just aren't answering the polls at the same degree. Pollsters have been trying to do various things to offset that. But I'm just kind of wondering how you assess, you know, where we're at this cycle and whether we're staring down the barrel of another Anne Seltzer situation. or not? Yeah, was Ann Seltzer right, or was she just early, right? That's the rest of the real question. And yeah, so, I mean, we're getting to the point of the cycle, which happens every cycle.

Usually it's a little bit more intense in presidential years than in midterm years. But with just the number of high-profiling, high-profile polling misses, not just in the general elections, but in the primaries before, people are wary and kind of operating under the assumption that, or just, you know, engaging in a defense mechanism that polling is going to miss once again. And in favor of Republicans. So basically in their minds, people are kind of like subtracting a few points from Democrats in a few of these races, which is not necessarily grounded in reality. We don't really know yet. But I think there's a good case to be made for polls underestimating Democrats as there

is for polls underestimating Republicans. And I can lay out each case. Yeah, let's let out each case. Because, and tell me if I'm right about this as a non-math-paths. person. It seems like what's underlying this question is that what we know is that there are non-response issues among Republican voters right now. And I was speaking to a consultant working on a race, one of the big Senate Midwest races yesterday, and they're just like way more Democrats are answering the phone. And so our own internal pollsters are waiting, Republican answers more heavily.

they're waiting against 2024 vote recall. And what that could mean, right, is that we're not correctly capturing Republican turnout, but it could mean it both ways, right? And I think people are immediately going back to, oh, well, this is 2016 again, you know, capturing it in a way that underestimates Republicans. But I guess it's possible that it goes the other direction. So, yeah, why don't you lay out both cases? Firstly, before you lay out both cases, is that a correct kind of assessment of what, you know,

what the known unknown is to borrow from Rumi. Yeah, I think it's important also to start with the more recent polling miss, which was during the Democratic primaries, particularly in Michigan and Wisconsin. And I was one of those big polling misses in Michigan. I had a poll that came out, him out right before Michigan Democratic primary. I was polling for El Sayed's Super PAC, and I missed by double digits. And I did a very big kind of retrospect, very public I put out on what my theory was. It's like a seven-page memo of what I think went right.

wrong and what was within my control that I would have done differently in hindsight and some things that are just impossible to correct for. And I think there's some lessons to be learned from that, even though it was a primary. I think there's two issues with reaching out to folks. Republicans are just much less excited to take polls right now. They are dejected. They're pessimistic. A lot of them are pissed at Trump, although also probably vote for Republicans and Democrats are jazz and excited. And this is kind of what we see after like conventions, right? The convention bump in a presidential year, not the midterm year, I don't think that really had the desired effect.

But you see, oh, Democrats got a three or four point bump after the Democratic convention, Republicans, like, that's not really real. It's a sign of enthusiasm, right? You're jazzed up. It's a pep rally, and that's why it's fleeting. So it's important to kind of correct for that. Here, it's the same thing.

If you're more excited and you're more optimistic for your side, you're more likely to want to participate in a poll. And at the same time, Trump has been attacked, polling as an institution for a while. Like he's been attacking the media, he's been attacking other kind of long-standing institutions. So there's a general mistrust, particularly with MAGA voters.

People who are just deeply distrustful of polling in general think it's all rigged, and like, why would I participate in a poll? So those folks are just next to impossible to reach. But going back to Michigan in the primary for a second, one of the main issues that I missed on in terms of margin was rural voters. So folks in the northern part of Michigan,

I actually nailed the regional percentages. If you look at county by county, region by region, the model was correct. It was the people in rural areas who were answering the polls who were not representative of who actually turned out. So the Abdul supporters in the northern part of the state, which Bernie carried that part of the state in 2016 and did a better in 2020. So it was plausible that Abdul would do well there.

They were answering the phone. They were answering the phone. They were answering the poll. The board was answering the phone, but the Haley Stevens ones weren't. Correct. But the ones who turned out were completely different. So there's a massive regional miss in my poll, and I think I'm guessing in other polls as well. And I think that we're seeing the same thing right now.

I think particularly my gut says particularly in the Midwest, right, where there are a lot of folks who are being hammered by tariffs, beef imports, diesel prices, who are just like really just trying to make ends meet and really pissed but don't like Democrats. but are probably going to end up voting for Republicans or staying home. Right. Like the type of person in Sioux County, Iowa, you know, rural Iowa, big Republican, part of the state. Yeah, I just did a poll for Brian Bangs in South Dakota, the independent candidate running against Mike Rounds, who's in an uphill battle, kind of like a likely R-reach kind of race. And I was looking very closely to see the kinds of Republicans that I was getting.

Right. So ultimately, the poll, which I was, you know, very conservatively weighted, you know, wanted it to be very, you know, pretty bulletproof from methodological perspective, Trump plus 25, and Trump's approval rating among Republicans was 73 to 27, which felt a little low to me, right?

But is that a reflection of who we're getting to answer these polls, right? Or is it a reflection of deep dissatisfaction with people who were actually going to turn out, right? And the people who are really, you know, dissatisfied with them are going to stay home. And so if there is a,

they miss, I think that is part of it, right? What was the South Dakota poll? What did it show? What did you show? It had rounds up by six points, which was in a Trump plus 25 electorate. So rounds still ahead, but that race was barely on the map to begin with, right? Yeah, sure. So the case for Democrats doing worse than what the polling shows right now, particularly in the Midwest, is basically what you saw in the primary, right? The people that are answering polls and the red parts of Iowa, Kansas, Ohio, Michigan are, you know, The local Tim Walzes, the liberal football coach is answering the phone.

Yes. You know, the little blue dots, the bulwark fans out there. The bulwark fans in the red parts of those states, they exist. They're the ones that are answering the polls. And so, you know, there are a bunch of people, there are more, you know, Republicans, Maga Republicans in those areas than there are bullwork people. And those folks aren't answering the polls, but they will reluctantly show up, even though they're kind of annoyed about the tariffs or whatever.

And as a result, the polls had to be able to. overestimated Democrats. If that's right, okay. So the case for it underestimating Democrats, and you put this out on social media, and I was talking to another strategist who offered this as well, was that, you know,

they were saying that like, because you're waiting the polls so much, you know, because there's so few Republicans answering and you're waiting them, like it creates kind of like a math issue where like if John Ossoff is like really at 60%, like if it's really 60-40, like it's almost like you can't have a poll

show that high of a lead because they're getting weighted so much that pollsters kind of don't believe it, right? And so it's possible that like the non-response bias reflects the turnout. Yes. So anyway, talk about that. No, I mean, I'd be lying if I said there wasn't a risk of being kind of far out there with an outlier. I think what was it in 2020 that the ABC Washington Post poll in Wisconsin that had like Biden up by like 18 points right before the election, which people still talk about. So you don't want to be like necessarily remembered for that. So a lot of times you'll get results back and be like, this is not really plausible. There must be something else going on, which is a subjective call, right? If you trust your numbers. In my Michigan poll,

which had else I end up by high double digits, you know, like 18 points and you one by one, so a huge miss. My gut was to, I double-checked the everything passed a smell test. I checked my numbers and I was like, it's possible. I'm not getting the right number. I'm going to put it out there. And if it's wrong, it's wrong. Right?

But that was a risk. And I got pretty well roasted. I think rightfully so, it is a risk. And like the downside potential of having an outlier miss, I think is higher than having one that succeeds. I mean, Ann Seltzer is still in hiding. Exactly. So that's why you see some level of like hurting.

So kind of converging on a, you know, a basically accepted range of result. and you see Marist come out with these polls that have like Democrats up by like 10 in Iowa, right? Or like eight in other states and they're like, well, that's not believable. And then people are slicing and dicing and digging into it. And, you know, pollsters at the end of the day want to be correct, right? Like even the internal pollsters, if they're putting something out publicly,

if you're off by like 25 points because you're doing an optimistic model for your candidate, like that's not good for your reputation long term, right? So there's no grand conspiracy. Let's look for it. I'm going to give you my Senate tiers list. You tell me what you think about it. Georgia and North Carolina are done. They're in the done list. Nothing can happen to stop the Roy Cooper, John Osoff train. Likely Dem, like almost certainly Dem. I have Michigan alone in that category. I feel very good about Abdul. Lean Dem. I think Democrats are going to win these. Texas, Alaska, Ohio, and Maine. Lien, R, Iowa, and Kansas. That's where I'm at right now. Well, how do you disagree or agree?

Where do you have New Hampshire and Minnesota? Are they off the map, too? Oh, yeah. Yeah, they're done. They're off the map. I've already said I'm going to wear by boxers and carry a tampon-timson or on Duluth if Michelle Tupoya wins. I saw that.

Yeah. If I had to pick today, so that map is 5248 Dem. If I'm mistaken. So you flip five seats. If you have no toss-ups, I completely agree with your ratings. Yeah, no, this is no toss-up. Have some balls. I agree.

Yeah. So if I had to pick today, that's my pick. The ones I'm least sure about are, despite what the polling is saying, I'm just generally skeptical of Ohio. I think you have to flip a lot of folks in parts of the state that are just do not like Democrats, like the Youngstown area, Mahoning County,

kind of ancestral Democrats who, if you're doing this, it's because you're really pissed about data centers and tariffs. And you really have to hope that. translates. And I just, I'll believe it when I see it. I would pick Sherrod today, but of those lean Dems, that is probably the one I feel least good about. I think Talarico, yeah, despite, I think the polling is very good for him. I also think that it's helpful that Amy Acton is in a very strong position against Vivek Ramoswamy, too, and there's kind of a strong two-tiered ticket here. And I think the case I would make for Iowa, which I agree with you, I would pick Hinson today.

She seems more resilient. She's kind of run against Democrats before. She's not as, like, rusty at running against Democrats as a lot of these candidates seem like Roger Marshall and other folks who just been kind of coasting for a long time. And in the recent New York Times-Cena poll is that she's the only one that hasn't slid majorly since the summer. But Rob Sand is likely going to win that governor's race.

And so if he wins by a huge margin, there will be a bunch of ticket splitters, but I think he could carry tariff. across the finish line. All right. So there we go. We're aligned. Concerning.

Concerning the line. We're hurting. We're hurting. We're hurting. I don't like that. I'm going to have to think. I'm going to let's think about a change. Although if you look at polling,

Democrats are heading all of these, according to the averages, including Kansas. Yes, I know. So, I mean, we're already kind of hedging a little bit. We are hedging. Emotional hedging. A little emotional hedging happening. No edging with hedging. All right.

The Texas and Florida, I want to do some stretch house races. Yes. You know, both of us, I just think this is, I've been. trying to talk about stretch house races in part because I think that's if you're going to give money right now, that's where the money can matter the most in these sorts of things. So for viewers who are looking to support candidates where they can make a difference. I asked you for some places that you are eyeing here. And I noticed some themes.

You looked at Texas and Florida and kind of one bucket. Yep. And I assume it's mostly because of the Hispanic vote. And you mentioned Texas 2335. I had Texas 35 on my live. I put out. We'll remind people that I put that in the show notes again. Texas 35 is Johnny Garcia. He is a local law enforcement guy running against Carlos Dela Cruz. This was a one of the seats that they tried to gerrymander and now was an R plus four seat. And this is closest,

where final influx could make a big difference. But Texas 35 is one that I've been watching. And then you also mentioned Texas 23. Talk about that. Well, that's what Texas 35 just really quickly is also, super cheap media market-wise for Texas. So your donations can go really far for Johnny Garcia, even more than other places in Texas because it's so rural. And then Texas 23, this is an open-seat race. This is Brandon Herrera, running against Katie Padilla Stout.

And this is a fairly Hispanic district. Trump won it by a fairly comfortable margin, although not an overwhelming margin. And the Republican nominee is frankly insane. I think his profile picture on all of his social media is having a gigantic gun, which is not necessarily a bad thing in Texas, but it's coupled with some long history of outlandish statements.

I believe he was a podcast host. Yes, Will Summers uncovering his quote-to-quote, outlandish statements about pussies. He loves talking about pussies. Yeah, outlandish is probably a... So this is one where he's just like a historically, likely historically terrible candidate that could cough this race up. And the only thing that could save him is,

this is also a redrawn district. And the only thing that could save him really is, you know, it is a fairly white district. And if Republicans hold up pretty well with suburban and ex-urban whites, I think that he could pull it out. But this is one where Republicans could easily cough it up rather than Democrats, like, you know, having a tour-to-force performance.

U.S. voters abroad. Have you got your ballot yet? All states are now sending ballots to overseas voters. So if you want to vote this year, you should request your ballot right away. Go to vote fromabroad.org and click request my ballot, ask to receive it by email. Questions? Ask our support team directly. Don't miss this chance to shape the future.

Don't delay. Vote fromabroad.org. Okay. Now going back to my original home state, Colorado. We've been gagging. You have on social media. others about the idea of A.O. Colorado. Colorado Springs District done here. Mini Routinal, who's in Colorado 8, gave an awesome debate performance last night, and that's

kind of North Denver suburbs. So the other two stretch districts that are really stretch is Lauren Bilbert's district, which kind of is up in eastern Colorado now, and then the Western slope part of the state. Talk about those two races, Colorado three and four, and why you think Colorado might be a place for some surprise wins. Yeah, I'll start at the top of the ticket in that Victor Marks, who is the Republican gubernatorial nominee, is perhaps even crazier than Brandon Herrera.

I think yesterday, he's been arrested before, and yesterday defended, like, beating his wife, said, I'll do it again for hitting his wife. just like one of the more problematic. He's bragged about murder, said he's a murderer. That's right. How can I forget about that?

Exorcisms. He's big on exorcisms. Not really a great fit for Colorado. Not really a great fit for Colorado. And I was remarking upon this yesterday. You know, 20 years ago, Colorado was a pretty solidly red state and the state of the Republican Party in Colorado

almost couldn't be worse. It was a cobalt blue state Republican parties and complete disrepair. I worked on a ill-fated Democratic Attorney General race this year. And no one was even worried about the Republican. The primary was the general election. And it's just funny to see that. So I'll start with Colorado 4.

So I lean Laubacher taking on Lauren Bobert. And I think as of the last filing, the Democrat, Laubacher had over 10 times as much cash on hand as Bobert. Boebert's not even trying. this is a pretty red seat. Bobert left her old district to run here because it thought it'd be easier, but she's really phoning it in.

There's been scandal after scandal or bad headline after bad headline with her. Historically, she's massively underperformed relative to fundamental. So this is one where, like, back into nerd talk, the quantitative, like, modelers have it as a very, like, 50-50 kind of race versus the traditional, like, cook inside elections,

and Sabato Cristobal have it as like safely Republican. So this is one of the ones where they diverge the most. And I think the case for it being competitive is pretty strong. And I think people tend to underestimate when the bottom falls out in a state, it can really fall out. I think in the House in particular, I think the Senate, there's just structural barriers to Democrats getting more than like 54, maybe 55 seats. If the bottom falls out, but in the House it could be like 240, 250. without really breaking a sweat if things get really bad

and polls are underestimating Democrats and you add on top of that you know, Victor Marx leading the ticket and the Colorado Republican Party being so it's not impossible to imagine and I mean Colorado is not Texas yeah I mean look even these states

even these areas right I just have pulled out the map here so the Bober District includes Fort Collins which is a college town includes Douglas County which is a Denver suburb back serve. That's Castle Rock in it. Pretty well-fying, right? Yeah. Yeah.

A big golf, famous golf tournament there is in Colorado every year. And so it also goes all to Eastern Colorado. There's plenty of rural there, but it's not like, you know, if you have excited turnout, you know, there are pockets of support for blue votes. And then if you look at the third, which you can talk about next, and that's the Western slope. So obviously there's a bunch of rural stuff there, but it's also, you know, a bunch of ski towns. I mean, Picking County, Aspen's there, Glenwood Springs. Is there, you know, you go down to the south and there, you know, kind of in Durango, other parts of the state.

So there are pockets of blue votes in both of those districts. Yeah. No, I think that both these districts are key test cases, key test cases for that rural theory, right? Yeah. If rural voters are not answering polls because they're not turning out, right? then that's good news for Democrats, and you could post a shutout or close to it in three, four, and five, which are the three Republican-held seats.

If they're just not answering the phones and are going to turn out, I think that there's a case that Republicans can hold all three of them or at least two of the three of them, and then you have to take into other account the factors like Bobert underperforming in the past and the scandals and all that. So that's why I think that in some ways I feel better about for Democrats in the 4th District and Bobert's district,

and the third, which is Jeff Hurd, and against Dwayne Romero. And Hurd's been trying to, like, run to the center very suddenly. You know, ran very hard on immigration in the past and now has kind of softened his tone on this. There, again, there are pockets of Latinos and immigrants there. That's a pretty white district overall.

And it just kind of encapsulates the change of a tone. Like, you know, January, February, 25 when Democrats voted for like the Lake and Riley Act and thought there was this big sea change on immigration. And now, you know, you see people like, oh, we need immigrants. I'm pro-immigrant. I just got to come the right way. And it's just the sign of the time. They're reading polls, right?

They're reading the room. Yeah, well, and you're pointing out, this isn't Texas. But again, but it's like San Luis Valley. Like, there's, there are Hispanic voters there's. Durango's industrial convention, Montrose, I think, is part a little bit in five and four. So, yeah. Or five and three, rather. So anyway, I'm intrigued by Ato, Colorado.

Yeah. That would really be something. And then one thing I would mention, too, is, you know, in the same way that the bottom can really fall out in certain states, it's not going to be a universal swing, right, from 2022 or 2024. Some states are going to see massive swings. Some are going to see slight swings. Yeah.

Like in 2022, which was a weird year in a lot of ways, you had these localized red waves in, like, New York and Florida. Right. And here, I think one of Nate Silver's theories, which I think is very plausible, is that Democrats are making outsized gains in deep red states and deep red districts, right? And so whether that's enough to actually win or just have a lot of heartbreaking losses remains to be seen. Sure.

Yeah, I also think red areas of sunbelt, but this is why I think Colorado, and we did talk about Georgia, North Carolina, there are a couple seats on my list in North Carolina, for example. But places like that, you know, where you kind of have both things happening, Montana with San Forstack, right? Like places where you have, you know, kind of in the deep red area, people are upset, there's depressed turnout, economy's worse. And then you combine that with higher college-educated transplants, you know, kind of people moving in, places like that. Like that's kind of the, I would think, the recipe for Montana, Colorado, Georgia, North Carolina.

And Texas. Like, you know, Texas, blue tsunami, you know, whereas maybe a smaller wave in other places. All right, Adam. We don't have any time for this, but I do. We just have to do one discourse question. Okay, sure. You, despite, I just learned now, being ranking Zellner Myrie, number one, instead of Zoron and New York Mayor, I'm glad that's on the record. You know, you've been a pretty strong soldier in the online discourse on behalf of kind of the idea that there's DSA momentum in the Democratic Party and that's what voters are looking for.

I'm just kind of wondering just at the top line, like what your takeaway was from primary season and how much you think that projects out to be continuing in 28. Yeah, so I don't want to overstate that I think, you know, the left and DSA are the solution in swing districts and states. I think that there's a couple of interesting test cases for this, for relative overperformance, underreformance. I think Abdul, even though he's not DSA or socialist, it's clearly on the left or wing of the party. Will Lawrence in Michigan 7, Matt Dunlap and Maine 2, even Sam Forstag and Montana 1. There's like a half dozen of these races. Angie Nixon, to a certain extent, in Florida, who I think you had on yesterday.

And we'll see. We'll get some really good data on. And again, at the end of the day, I'm pretty data-driven as a pollster. So even though I am personally more progressive, I wouldn't say I'm left. I'm more kind of progressive. I call the centrist by people on the left and a leftist like people in the center, which is kind of right where I want it,

is if the data kind of shows, if you control for all the other factors, that these folks are underperforming relative to what you'd expect, if you control for all these variables, that sends a pretty clear signal for 2028, right? You can explain your way around any of this. Like if Abdul underperforms,

you could blame him getting nuked with $50 million, $70 million in the primary and having to recover from that. You blame Islamophobia. There's all these kinds of things you can say. But in the aggregate, right, all these races combined, you can see, right? I do think that there is something

that the, and this is the part that I've been kind of pushing, is that the left and the populist progressive left is doing a really good job articulating the economic frustration and anger and channeling that, and people

are responding to that. If that was the only issue that was being talked about, I think that could be a recipe for success, but I think they're also pretty wildly out of touch on cultural, social issues. And how people talk about things. So I think for me it's like how do you capture that element

that they're tapping into that the establishment center left does not tap into in an authentic way without alienating kind of the folks with more kind of out of step other views on other issues. And I think that we're seeing that kind of folks like James Talleygo to a certain extent John Assoff talking about corruption. And there's a few people that I think are doing that very, very, very well. But it's a very tricky balancing act because you do have to concede some stuff. Yeah. Well, there's a category of people some of them, which you mentioned, who are like Bernie

dogs, they're calling them, like blue dogs that are doing Bernie economic populism stuff. That's Forrestag and Montana and others. It'll be interesting to see how... Rebecca Cook too, right? Yep. Bob Brooks. Yep. And Palspanos. I'm interested to see those. All right. We can analyze it when it's all over. All right? We'll have you back on over the winter. Yeah, no, I'll be

pouring over the data. I'll try to be an honest broker. I want my team, the anti-establishment faction to do but we'll see what happens. And at the end of the day, I just want to win. That got you, baby. Same. All right, that's Adam Carlson.

I appreciate the time. Thanks for making an inaugural visit to the show. We'll see you on the internet. Great. Thanks for having me. Out once again, find me a friend. Someone do walk with too, too many boys. Too small a town.

Feels like I'm always wandering the grounds. You call me up. I let it rain. Heartbreaking me, only one thing got me again. It's the end of the world. Done with this year, I miss drinking beer, miss singing songs for you. Too slow to hate.

Summer came late, fires on the west side. Let's misbehave, catch it before it's almost gone. Try to pretend It's the end of the word So kiss me to death There's nothing left to decide Nothing is brought to you Thanks to the work of lead producer Katie Cooper

Associate producer Ansley Skipper And with video editing by Dante de Chico And Audio Engineering and Editing by Jason Brown