I'm practicing my lean. Yeah. The podcast wraparound. The wraparound. Nate, the podcast wraparound. The reach, babe. No, not the former, not the latter. We don't need h on our case on our second official episode. Okay. Okay. Okay. Here we go.
Welcome to Still Counting. I'm Gayen Duke.
And I'm Nate Silver. On today's show, new polling suggests Democrats are in a stronger position than ever for the midterms. So, have the midterms broken their way? And what would that even entail? Then, concerns about AI safety have broken containment and entered the political debate. We're going to discuss what Silicon Valley and now Washington are fighting about, what Americans think, and how it's changing politics.
But first, the midterms. So, since our last episode, there has been quite a bunch of new polling, including one from the New York Times/Siena College. In fact, Nate, you talked with Dan Pfeiffer of Pod Save America yesterday about that polling, and folks can go check out that special on YouTube, but we're going to talk about it more here and about what the model shows us more broadly. So, and usually we suggest that we don't write an article or do a segment based on one poll, which we're not going to. We were going to talk about the whole sort of universe of new polling, but this is one the New York Times is sort of...
It's real at this point. It like drives the political conversation. It's a very high quality poll.
Look, ordinarily if our wonderful producers had asked us to do a segment on one poll, I would have said, "No, sorry. I'm going for a walk or something instead or going to write about how, you know, where the AI is going to kill us all." Um, the reason I chose to do this, and you can go to the YouTube feed for more on this, is because it's actually not an outlier, right? It's, um, it's consistent with almost all the polling we've seen.
Well, let's say what it says, which is Democrats leading in the generic ballot, which is when pollsters ask, do you support a Democrat or Republican... audience knows what generic ballot is?
Okay, Democrats leading the generic ballot by about eight and a half points according to the latest poll.
On a likely voter basis. Um, yeah, we talked about last week how after Labor Day, pollsters traditionally switch over from using registered voters to likely voters. Um, and our model flipper...
That's never going to get old. That's always going to be creepy.
I thought it was a weaker... the week one dolphin noise was better. That dolphin feels like...
I know that was a throwaway.
Dolphin kind of more palatable, honestly.
Um, but you know, our model had detected that likely voter polls were likely to be better for Democrats. So, it was kind of priced in to some extent, but still to see Democrats up by eight and a half points — it can be eight or nine depending on how you round it — um, is consistent with what we thought. But, um, but any notion of like, oh, people will come back, kids come back to school and then we'll see Republicans rebound because, you know, their voters will get fired up by Democrats being too left-wing or be loyal to President Trump. I mean, there's not very many signs of that. Instead, you see in the New York Times poll a pretty big gap between likeliness to vote between Harris 2024 voters and Trump 2024 voters. And you also see independents, um, really leaning toward Democrats in their congressional races.
Right? Something like a little over 60% of Harris 2024 voters saying they're almost certain to vote,
which is why JD Vance is telling all of the Republican voters to, um, go out and vote, right?
Yeah. And it was like, it was something like 56, 58% of Trump 2024 voters. We got this... I'm suggesting here that we got a lot of data here beyond just that generic ballot topline. Claire, what stuck out to you in the latest poll?
Honestly, the immigration and economy numbers. Um, Democrats on immigration, people trust them 52% to Republicans 45%, and on the economy, they trust Democrats 51% to Republicans 45%. The immigration number in particular, when I was first scanning through, I was like, "Oh, that's really interesting." Because that's obviously where Trump was kind of doing his bread-and-butter work since 2016, but particularly in the last year. Obviously, there's been a lot of immigration actions by the administration and the economy. It's like all we can talk about ever is the affordability crisis. I mean, I was texting a group of friends yesterday about the bond market and...
...and interest rates for home loans because, you know, that's what you worry about when you're almost 40. Um, and so I think a lot of people, you know, just those are really terrible numbers for a Republican that basically made his campaign and his party sort of coalesce around those two issues.
Yeah, it's driven, as you might expect, by college-educated voters, voters with an advanced degree or a degree. Um, are 59-37 in favor of preferring Democratic policy on immigration. Non-college voters, who are the majority of the electorate, lean toward Republicans by only four points there... five points, excuse me. Um, so yeah, you see, I think the aftermath of, you know, um, of Minneapolis. Also, I mean, ironically, uh, you can argue that in certain ways Trump is being penalized for his own success. I mean, we have had a lot less immigration on the southern border, right? So, like it might be a less visible problem and so now Trump looks more extreme with some of the ICE actions and things like that. Um, but yeah, this was supposed to be his strength. It's not as bad as like some of his other numbers. Um, but it's no longer a strength. It's a lesser weakness, I guess you'd say.
Yeah. I mean, folks will have heard me say on a previous podcast that voters still prefer Republicans over Democrats on immigration. Now, this is likely voters, which sort of over-represents college-educated voters, high-propensity voters, people who are very locked in. The electorate will look different come 2028, but this is a change, right? This is Trump's biggest strength becoming a vulnerability that is on par with the economy and foreign wars according to this. Right, underwater Republicans are by seven points on the economy, immigration, foreign wars.
Well, I don't... I just... I don't think...
I don't think immigration is going to motivate nearly as many votes.
But in terms of preferences between the two parties, I mean, it's obvious that affordability is the number one issue in this election. Are we doing the affordability thing? I call it inflation, man. Affordability feels too euphemistic to me. I have an objection to that term.
Okay. Yeah. Now, I want to...
Inflation. That's what we've always called it. I mean, affordability might be kind of like technically more accurate, but it feels like a little bit of...
What about it? What about it doesn't... I mean...
It just feels like a neo... how do you say, neologism or whatever, right? But like it... I feel like it's a word that like is a little bit... contains a little spin. I mean, there are a lot of political terms, you know, that I have...
But doesn't it have to also do with wages not keeping up with inflation or cost of living? Isn't it more encompassing than simply just inflation itself?
Isn't it... like we spend a lot of time...
I think that's what it is.
...in the Biden era denying that inflation was so high or that voters had realistic reasons to be concerned about it and then it kind of transformed. I'm just interested in political language. Probably you think I'm making an annoying pedantic...
...objection. That's why... that's why I'm Nate.
Well, but the thing is...
I don't think that what people are pissed off about is just the fact that we're experiencing sort of 3.4% annualized inflation. I think people are feeling, well, the accumulation of inflation over a six-year span, but also the sense that... and there are certain parts of the economy that have to do also with interest rates. You're just talking about home costs, which are not directly inflation but cost of living. If you want to call it cost of living, great. Let me make a point though on exactly that, which is that in your Silver Bulletin model, when you look at Trump's approval rating on the cost of living, inflation, whatever the [__] you want to call it, he's 40 percentage points underwater. Like it's really... it's really [__] bad. When you look at this though, when you look at, you know, the economy or cost of living in this, Democrats only lead Republicans by seven or eight points, right? Republicans are still getting 43 to 45% of the electorate on this issue. That means that there are a lot of people who are not happy with what Trump is doing, but are not going to vote for a Democrat. They're still going to vote for a Republican when push comes to shove. And there are a lot of people who still remember the Biden years.
The communist thing still works. And I do... I mean, to go back to your thing about the Biden years and inflation, that is true, right? Like there was a little bit of, um...
Yeah. It's like, oh, it's misinformation... like a little gas. Yeah. A little gas, $5 a gallon gas. It's not misinformation. That's like, you know, that's actually what you have to pay if you actually try to get your tank filled. Um, but look, there's this famous, I think, Bill Clinton debate answer in 1992 where, um, where a woman asked a question about the national debt and President Bush answers it very literally and then Clinton realizes this woman just has concerns about the economy in general and answers it less literally and talks about inflation and things like that. And so, you know, I'm just sensitive to like political speech because people were not calling it affordability in the previous presidency. I feel like it's a piece of framing that's actually effective. Like if I were John Ossoff or Cory Booker or something, I'd use that term, but I'm just sensitive to like the speech that...
I do. I get it. I also think it is more... setting aside the Biden or the politicized use of it, I do think it's more descriptive of how people feel. Like the reason why I was texting about the bond market is because shout out to Ben Castleman, former 538 staffer, who is now the chief economics correspondent at The Times. He was on The Daily yesterday and he said this one thing that really freaked me out, which was basically like, um, actually maybe the last 20 years of, um, low interest rates were the aberration itself and that we will just have like forever high interest rates. And I think that coupled with the fact that like the first home that people buy in America, the price of it has gone up astronomically. In New York, I mean, let's not even talk about it, it feels sort of a filthy thing to say. And those things coupled with like 30... what, grocery prices are 30% higher than they were in 2020. It just feels more encompassing and atmospheric than to say like interest rates or inflation high, right?
I agree. Maybe my point is too pedantic, but like, look, people don't cleanly differentiate the different economic anxieties that they have, right? I mean, it's always been based on a mix of kind of vibes and objective data. I mean, that was kind of one of my big critiques about, like, um, about the Biden years. People would say, "Well, the economy is actually doing fairly well by jobs-related measures and it's just bad vibes." It's like, well, the vibes are always bad because press coverage is usually negative. And like a lot of people in a lot of westernized countries feel as though they're getting the wrong end of the deal. And, you know, we now have 1 or 2% growth per year instead of two or three or 4% and that compounds over periods of time. So, let me say retroactively, we should have called it an affordability issue under Biden as opposed to not calling it an affordability issue under Trump. And I say that because when you look at the data, Americans were becoming literally poorer for the first basically three years of Biden's presidency, which is to say prices were growing faster than wages, and by the summer of 2022 by a real significant clip. And so Americans, as much as you could say, oh, but like look at employment levels, Americans were getting poorer and you couldn't tell them, don't believe your lying pocketbook or whatever.
Let's zoom back out to the midterms overall. When we sat down here last week for our first episode, your Senate model over at Silver Bulletin was roughly 50/50. And I made a joke about you amping up the drama, um, as we returned from our summer vacations. Now, now Democrats are back up to about 60% in your Senate model. Um, I think that's the highest they have been since you launched. I imagine the New York Times/Siena poll is a piece of this but not the whole.
Yeah. One individual... I mean, the model is designed so that any one individual poll should not affect it all that much. Right. It's the overall constellation of polling. Um, you know, Quinnipiac had a poll showing Democrats up 11 points on the generic ballot. YouGov has them up like 10 or 11, and among likely voters they're doing much better than among registered voters. Um, Marquette University, which is a strong pollster, had them up some gargantuan number like 12 or 13 points. It's not even in our model yet, that came out this morning. Um, so it's been pretty consistent. It's also been like, it's really hard now to find any Texas polls where James Talarico is not ahead. Um, he's ahead within the margin of error in any individual poll, but collectively that presents a fair amount of evidence. Or Sherrod Brown's been ahead in basically every poll of Ohio. There are some exceptions. I mean, New Hampshire polling has been maybe surprisingly close. We talked a little bit about New Hampshire last week. Um, but, you know, our flipper is actually pretty steady and stubborn. Um, for a presidential model, it moves faster because you have more of a focal point on just this one race that has 50 different manifestations in 50 states. Despite that, um, we are gathering a fairly consistent story about Democratic performance in these polls has been quite strong post-Labor Day.
And does this, what we're looking at right now, take into account the fact that Cook just moved its rating for New Hampshire?
Uh, does this account for the Cook race move?
It does, except we have to debias the expert ratings because they are completely miscalibrated to what any reasonable person, probably including them, thinks what the national environment is going to be.
Yeah, we got a listener question on this which I was debating whether or not we should dive into. Let's do it really quickly and then we'll move on to whether AI is going to kill us. So, Coro Queen on X asked about the gap between the quantitative forecast models like yours and the model at Split Ticket, which is the argument our friend Lasha Jane runs, that VoteHub, which is Zachary Denini, where Democrats are projected to win in your model like 30-plus seats, the modal outcome is like 30-plus seats in the House, or 27 seats over at Split Ticket, etc. Then when you look at the expert ratings like Sabato's Crystal Ball, Cook Political Report, Inside Elections, the modal outcome is, you know, Sabato's Crystal Ball, it's Democrats winning four House seats, Inside Elections, it's Republicans winning five. Uh, this may have changed since this listener asked this question. How do you make sense of the vast discrepancy there?
Extremely... I'm trying to figure out how un-PC I can be.
Just, just go with like media.
No, they have no [__] balls. They have no balls. They're terrified to put themselves on the line and say the Democrats are going to have a good year. And then what will happen is, is, you know, on like October 25th, they'll say, "Oh, now the blue wave has arrived." The blue wave is already [__] here in the polls. The polls might be wrong. The polls probably are not going to change that much between now and November. So, you know, grow a pair of balls. You can aggregate that. I don't care. Like, it's gutless to call something like New Hampshire a tossup.
Were you a [__] talker in baseball? Like, in sports?
I wasn't very good at sports, Claire. Maybe up over a lifetime. Can I ask you, is there anywhere on the map where you're like, I believe the race raters over Flipper? Well, it might be something like, like in Texas, the fact that they rate it as tossup or maybe even tilt R or something, right? Like that's useful as a relative signal, right? So like if you can sort out, you know, I might disagree with them on like New Hampshire or something, but like it's directionally helpful to have like an expert signal. But it's like, why on earth are you telling us this is going to be like a D+1 environment? It's not even a good baseline assumption in a presidential year. Yeah, the House tends to be fairly neutral, but like in midterm years, particularly in second terms, it almost always favors the out-party anyway. And like it just... it's just kind of hedging in this very strange way that as a forecaster who's been more criticized than anyone else, being frankly often more right than anyone else — not always, but like being criticized with like legibility in the public of how we present probability — is like they're making my job kind of harder and it's pretty [__] annoying.
Okay. Well, we'll see. Maybe they'll respond. We'll bring it up next time we gather here. But...
Aggregate those clip. We'll bring them on the show next time.
Still Counting is brought to you by ZipRecruiter. Hiring these days is very time-consuming. I mean, it always was, right? It took a while to hire you guys and look how that turned out. I mean, it couldn't have been worse. Am I right, Claire?
I'm a rigorous hirer. I'll tell you that much. Um, but ZipRecruiter can help. The future of hiring looks much brighter because ZipRecruiter's latest tools and features help speed up finding the right people for your roles. So, you save valuable time.
Valuable time. You can be modeling, Nate. Get back to it.
Build a model, man. Get that model running. Need some model juice.
And now you can try ZipRecruiter for free at ziprecruiter.com/counting. ziprecruiter.com/counting.
See a candidate that you're really interested in? You can unlock their contact info instantly. Over 320,000 new résumés are added to ZipRecruiter monthly. That's a lot, which means you can reach more potential hires and fill roles sooner. Use ZipRecruiter and save time hiring. Four out of five employers who post on ZipRecruiter get a quality candidate within the first day. And if you go to ziprecruiter.com/counting right now, you can try it for free. Again, that's ziprecruiter.com/counting. ZipRecruiter, the smartest way to hire.
All right, I'm sorry. We're going to talk about the midterms more as we discuss AI. But let's dive into the AI apocalypse. So over a month ago, 1,300 staffers at frontier AI companies wrote an open letter calling for AI regulation. There was barely a ripple in the news. A week ago, a researcher at Anthropic named Jacob Coxon posted about his resignation on X, warning that, quote, "The people building AI earnestly believe that it could kill us all by the end of the decade." Those tweets set off a fervor that is still in progress as we record, rippling through Silicon Valley and Washington alike. Claire, why did Jacob Coxon's post break through in a way that, I mean, months but also like years of warnings have not?
It is kind of a mystery to me. I can't remember actually who the first politician was who amplified it. I think it helped that the other, um, AI leaders amplified it, but it was a little bit odd to me because like honestly I've been... as the person on this podcast who doesn't really use AI, I have been, um, like morally disturbed by AI for a long time. And...
Is this a Catholic thing? I'm kind of...
I mean, yeah, like I think it is. Yeah. I mean, listen, I do think that I have like a very... that there is something to me that is really disturbing about, you know, when we're all talking... when AI people will talk about, um, the new world that we can live in with the benefits of AI. I, you know, like many people, the idea of curing cancers or things like medical breakthroughs, great, that sounds good. But when they talk about things like, um, we can all just like not work and pursue our hobbies. I mean, that stuff actually, um, you know, you can look at a state like West Virginia which has been not hollowed out by AI but hollowed out by industries that don't exist like coal. And people, you know, there is something to the dignity of work and people feeling a purpose in their life. I mean, I do have pulled up in front of me on my computer screen the papal encyclical by, uh, Leo the 14th.
All right, we don't need to turn this into a Catholic podcast just yet.
Oh, but I'm saying... I'm saying this is... listen, this is a long preamble, but I actually think it's important because the reason why people are so existentially disturbed by this is it is like the Industrial Revolution, like the Luddites before them, like it is this real feeling of threat to humans' way of life. And the fact that it no longer feels like total weirdos talking that way. It feels like people are sort of saying like, "Okay, wait."
Well, that's part of what you're seeing now is that like I've known these weirdos for a long time. I'm weirdo adjacent, right? One might say, right? There's a lot of coverage of AI in my book where I talked to Sam Altman and a lot of the leading figures, um, in all the labs, um, talked to all the effective altruists, all the rationalists. I'm happy to do a little glossary if people want or not. It's always a mystery as people who cover the news cycle but, you know, in some ways are critics of the news cycle too. It's always a mystery why any given story blows up, right? Um, why some mass shootings, for example, dominate the front pages for days and others are just ignored. Sometimes it has to do with the privilege of where these are taking place, right? Why some police killings become major stories and some don't. Proximity to, you know, New York and DC and things like that probably helps too. But yeah, we were, you know, people are experiencing a lot of understandable anxiety about AI. And objectively speaking, when you have, um, when you have leaders of these companies saying, "Oh, by the way, I agree with him, right? I agree with him that this could cause an existential risk." Um, I think it's overdue. A lot of sudden shifts in news coverage have to do with stories that were overlooked before. And I'm happy that this is penetrating from just the weird nerds who I support — I'm a weird nerd myself — into mainstream discourse. And it feels like timing's about what I would have expected even if Coxon is not a particularly remarkable... I mean, what he's saying is what a lot of people have been saying for years.
Yeah. I think it's sort of... it is like the weirdos in Silicon Valley who have been doing this thing. Like if you're an AI researcher, you have existed for the past 15 years in a subculture, right? You got a PhD at one of these places. You went to become a mathematician maybe at one of these Silicon Valley companies. You lived in a group house. You were segmented within others like you talking to each other for years and years. There was a groupthink that sort of became, you know, that turned into the different sort of contagions of doomerism or the effective altruists, all those things. But you were all talking to each other. And frankly, like the fact that it has broken contagion... like all respect to weirdos and, you know, smart weirdos and people building technology, but I don't want to live in their world. Like I don't... that does not... I do not... I and a lot of other people do not ascribe to that being like a good future. And I think the fact that so many people are sort of realizing like, oh, this is... the president's not going to stop it, the government doesn't know how to regulate it, these guys sort of are also telling us that it's bad, but they don't have a plan to stop it. It is kind of like, oh, maybe we should freak out.
Can I... I want to validate... like given that we try to direct our lens towards the broad public and analyzing these things, I want to maybe add some numbers to this. I did a whole... it was like if you want to listen to every poll out there about AI, we talked about it on GD Politics this week, but like to summarize a bit, what you see is a lot of skepticism. And I think this is quickly shifting terrain. And so we're going to have to see how the past week and a half gets priced into American concerns going forward. Like even in this New York Times/Siena College poll, you don't see AI as a particularly big concern and you also don't see a... like there's...
Republicans actually had more trust... what to make of that, but...
I'm assuming this poll happened before...
Not exclusively. It ended over the weekend, but it was the biggest issue on which people didn't have an opinion. 20% of Americans couldn't choose a party whereas like 97% of Americans can choose a party on immigration.
But let me say up until this point at least, when you... and actually YouGov did a poll that after Coxon was in the press a lot and Dario Amodei etc. started reacting, you saw still in the sort of fast, quick response YouGov poll, by far the bigger concern is still jobs over will AI kill... I mean, they literally asked, are you concerned that AI is going to kill us all? Americans are concerned about that by a 12-point margin. Are you concerned that AI is going to take all of our jobs? Americans are concerned about that by more than a 50-point margin. And so to your point about maybe a, you know, a Catholic work ethic or keeping a human in the loop or the role of humans in determining the outcomes of their own lives, a lot of Americans are there. And then when you ask them what do you want the government to do, you can propose almost anything and it'll only be like 15% of Americans who are like, no, I don't want to do that. Whether it's like AI safety for children, whether it's stopping AI from becoming super intelligent, whether it's protecting artists, etc. Like there's been this growing concern. There's been this growing local issue of data centers too, which is also, sorry about affordability, ultimately water and electricity prices, and it's sort of burst open in this way that's about safety. But still the biggest concerns are about humans' role in working for...
It is also... I will also say just, it is actually a big thing in education and how it's affecting children. Like I think, you know, we live in New York City, like I'm sort of in that discourse, right? Like New York City, um, just banned, I think, the use of AI products, I think, in the schools. I think a lot of other school systems are sort of starting to think about that. That sort of goes with the pairing of like phones. People are... in addition to the work aspect, and I wouldn't frame it as Catholic work ethic, I would say like the dignity of the human and how work and purpose is connected to that.
I was just trying... I was just trying to play into your...
I was giving you an opportunity to read the encyclical is really what I was doing.
Put that out as a bonus episode.
Yeah. Um, my dad and I can do it together. Um, but I think it's the idea of, um, not just work, right? I mean, this is about work, but it's about people envisioning a future and thinking about their kids and thinking, "Wait a second, is the education that I'm giving my kid going to even be useful?" Like, it's people really starting to turn on to not just the Andrew Yang thing of like, hey, when he said what, four years ago...
