전사 🔗
00:00:02
Speaker 1: Bloomberg Audio Studios.
00:00:04
Speaker 2: Podcasts.
00:00:05
Speaker 3: Radio.
00:00:06
Speaker 2: News.
00:00:18
Speaker 4: Hello and welcome to another episode of the Odd Laws Podcast. I'm Joe Wiesenthal.
00:00:23
Speaker 1: And I'm Tracy Alloway.
00:00:24
Speaker 4: Tracy, I often say, I'm really glad we're not political reporters.
00:00:29
Speaker 1: Oh, I know.
00:00:29
Speaker 2: I know.
00:00:32
Speaker 1: I'm glad that I wake up every morning and thank God that we didn't.
00:00:35
Speaker 4: Go down that path. But here's the problem. I've said this many times. Here's the problem. And we've been writing about this.
00:00:42
Speaker 3: We didn't.
00:00:42
Speaker 4: We were in Chicago. We had that conversation with James Robinson and stuff like that. I sort of think politics, unfortunately, does matter. And actually, even if we wanted to say like, no, look, I just cover markets. I just cover the economy. Unfortunately, I'm coming to the conclusion that not only do politics matter, they're probably going to keep mattering more and more.
00:01:02
Speaker 1: Everything is politics now. You can't get away from it.
00:01:04
Speaker 3: It feels like it, yeah.
00:01:05
Speaker 1: And certainly this week. So we're recording this on October 29th. The episode is going to come out on November 3rd.
00:01:12
Speaker 4: It's Election Day special.
00:01:13
Speaker 1: That's right.
00:01:14
Speaker 4: You're listening to our Election Day special.
00:01:16
Speaker 1: It's a pretty big week for politics. So we should talk, to put it mildly. To put it mildly. I know, I know.
00:01:22
Speaker 4: Anyway, let's just jump right to it. Joining us here on our Election Day special, Nate Silver. Everyone knows him. Founder of the Silver Bulletin newsletter, famous for 538, modeling elections, knowing a lot about polling and data and all that stuff. Nate, thank you for joining our Election Day special.
00:01:39
Speaker 3: Of course. Thanks, Joe. Thanks, Tracy.
00:01:40
Speaker 4: Let's just talk about like how unpopular is President Trump right now as of October 29th or Election Day and to what degree? I mean, midterms historically pretty bad for the party and power, but just get a big picture. How historically unpopular is President Trump and what is the knock-on effects for various Republican candidates in competitive races?
00:02:00
Speaker 3: Yeah. So Trump's approval rating is at about 37 percent. By some measures, that's the worst approval rating for a president at a midterm since Harry Truman, I think, in 1946. I'm getting that right. And remember we had a midterm in 2018 where Trump and Republicans lost 41 House seats when he was somewhat less unpopular and where gas didn't cost $ 4. 30 a gallon or whatever it is. So look. I kind of call this the rock versus the hard place election, right? The background conditions couldn't be better for Democrats, right? The historical trends in midterms, voter anxiety about the economy, the Iran war, Trump's a lame duck and 80 years old. They've done really well. And with turnout, they're doing well in early voting so far. They had a very good year in 2025 with New Jersey and Virginia. It's really very consistent indicators. I mean, they might have a couple of problems. Some of their candidates are seen as too far left, right? So that's kind of the rock. And the hard place is, yeah, they'll probably win the House. They'll probably pick up some governorships and state legislatures, but they do have to win in some really red states, just how the math works, to win the Senate, right? States as red as Texas, Kansas, Alaska, South Carolina is in play, if you believe the polls. And so it's kind of this big, I mean, this is why I like, I would say I'm a little bit anxious, not just for the implication for the country, but because like, it's kind of become a tricky race for like models and polls and forecasts, right? There's no... There's no gimme. You either have to lie and say it's a toss-up. That's not an honest answer at this point, right? That's hedging.
00:03:30
Speaker 4: But you have to take a kind of leap of faith that the Democrat could win in Texas.
00:03:37
Speaker 3: You have to kind of take a leap of faith on some of this stuff. And a probabilistic leap of faith. But at this point, it would be a really bad night for pollsters if Democrats don't win the Senate.
00:03:50
Speaker 1: Just to sort of test our priors a little bit. Everyone's treating this as a referendum on Trump, which kind of like makes me nervous when everyone's doing the same thing. But in your view, is that an accurate way to measure it? Like, do we have hardcore evidence that people are voting on Trump rather than voting on a local representative?
00:04:09
Speaker 3: I think the best evidence is that Trump is no longer featured in Republican campaign commercials, right? Trump is not being talked about anymore on Republican podcasts. However, Trump himself is trying to make himself like the main character of this story when I don't think very many Republicans want that, right? Ironically, they had this Republican convention, was it in Dallas just after Labor Day? And like, that's when Republican polls kind of went from bad to worse. I don't think that's the main cause or necessarily even the secondary cause. But yeah, I mean, the behavior you'd expect if people are being kind of game theory optimal is that Trump is taking a back seat and trying to let his more moderate candidates distance themselves a little bit. And instead, we've seen the opposite, right? I mean, if Trump hadn't endorsed Ken Paxton in Texas, I'm not sure that James Tallarico would have a slight lead there. If Trump hadn't chosen to go to war with Iran, I tend to think the Senate would be a toss up and not lean D. So no, he's the central character. And it's also kind of like, It's kind of like the last time Trump is on the ballot, barring extra constitutional shenanigans.
00:05:15
Speaker 4: Yeah, which
Load full transcript
who knows? Now, this gets to the question of like, could it be a bad night for pollsters? And there is this seems to this view has congealed in the general public that starting at least since 2016, pollsters always overestimate Democrats.
00:05:32
Speaker 3: Yeah.
00:05:33
Speaker 4: I think your view is that that's actually just factually incorrect. But why don't you tell us about what we know about polling error?
00:05:40
Speaker 3: So what we know, number one, is that polling errors are correlated. That if a poll misses high on the Democrat in Iowa, probably in a similar state like Kansas or Nebraska, all three states are competitive this year, good chance you'll have a correlated error. So therefore, you cannot look at individual races as a Monte Carlo random. But A, they're less correlated in midterms than in presidential years because you have different candidates on the ballot in every state. That's important. In 2022, we saw this kind of really big red wave in like New York, New Jersey, tri-state area. And Republicans did really well in Florida. The rest of the map was relatively good for Democrats. John Fetterman won and people like that, given that Biden had like a 39% approval rating. You know, number two, we have had systematic polling errors that underestimated Trump in 2016, 2020, 2024, a little bit more modest, but still direction the same, right? In every election in the Trump era where Trump is not on the ballot, You have not had that error. And in fact, in like New Jersey and Virginia last year, you had a massive underestimation of Democrats and governors races. And so like, first of all, it could all be noise, right? If you like code up a, you know, have Claude or whatever, or GPT code up a SIM to just show you like what random polling error looks like in the past 20 elections, it looks kind of like, you know, it's impossible to distinguish it.
00:06:57
Speaker 2: Right.
00:06:57
Speaker 3: But one story might be that like Trump has this kind of magical power to attract voters low turnout voters that also happen to be hard for pollsters to reach and where they get screened out as being unlikely voters. Those voters don't seem to want to show up for like a Roger Marshall in Kansas or a Ken Paxton in Texas, right? It seems very, you know, Trump got a lot of young men last year who flipped to 2024, who flipped to him. Young men don't usually turn out in terms of watching the NFL or whatever, right? And like there's a consistent story where like conveniently this problem of like these missing voters aren't an issue if they don't vote.
00:07:38
Speaker 1: This was actually going to be my next question. But like what are you seeing in terms of early signs of turnout? And then what are the chances that a former Trump voter actually votes Democrat rather than just decides not to show up?
00:07:53
Speaker 3: As compared to general elections or presidential elections, turnout is relatively speaking more important. in midterms. Differential turnout is almost always a factor in some of these bigger waves. I'm not an early voting expert, but there are people who are pretty good at it, right? It can be very ad hoc if you don't kind of set clear rules ahead of time. And they seem to be seeing good numbers for Democrats, better numbers in the polls. I don't know, because the polls are really good for Democrats, right? But it provides some more consensus evidence for it, particularly high turnout among Black voters who may fear being disenfranchised. high turnout or not high turnout, but comparatively high turnout among unlikely voters, people who in the voter files have voted no more than one in the past four elections. Democrats are getting a lot of those, but they're getting more than Republicans, right? I think some pollsters are finding there's very little D to R crossover. If you're any type of Democrat at all, this is not the year to indulge in a moderate Republican, right? Maybe a little bit of the reverse. But like, I mean, things tell a consistent story. It's just kind of this existential thing about whether people believe the story or not.
00:08:57
Speaker 4: Why is Trump so unpopular? Like, is it just gas and diesel prices? Like, break down the drivers of his unpopularity.
00:09:05
Speaker 3: Trump is kind of, well, people are bored of him and tired of him, right? I think people are just like, he's been president for, I guess, six years, right? But it feels like 10 years because he was still kind of the dominant figure of that kind of Biden interregnum. You know, he's 80 years old. But no, look, I think the basic story is that he has broken promises. most prominently on cheaper gas and on no more stupid wars. We can talk about how much the Epstein stuff affects him too. But yeah, one of the things about Trump's reporting number is like, of that 37% approval, like only 19% of it is like strong approval. I think there are permission structures that have opened up for Republicans to kind of distance themselves from Trump and incentives, right? You know, Trump, his endorsement will still matter after 2026, but like, you know, J.D. Vance and Marco Rubio, who are nominally leading candidates, have a big problem in that they kind of are, when Bush was this unpopular in 2026, and it got worse, by the way, after the 06 midterm, you didn't have Dick Cheney trying to run for president, right? You had a rebellious Republican in John McCain who ultimately got nominated instead. And so, you know, when you're a lame duck, the bottom can drop out potentially.
00:10:16
Speaker 2: Yeah.
00:10:16
Speaker 1: So at the moment when we're recording this, if you look at, I guess, the polling consensus and prediction markets, it seems like the House is pretty much a lock. Senate is a toss up at the moment. What's the actual range of outcomes for a Democratic win here? Like, what are the extremes that you could possibly see on both sides?
00:10:37
Speaker 3: I would push back on the Senate
being a toss up. I think at Pauline Marquette and Kelshi, Democrats are up to like 68 percent or something as when we're recording this. And in our model, they're at 70, high 70s now?
00:10:48
Speaker 4: To win the Senate.
00:10:49
Speaker 1: For the Senate?
00:10:50
Speaker 2: Okay.
00:10:50
Speaker 3: Oh, wow. Well, there's this kind of, both a blessing and a curse is to actually have to put out a forecast in the form of a probability, right? Because it'd be easy to say, oh, things are looking good for Democrats. but don't get your hopes up too much because polls can be wrong. And yeah, I guess it leans down, right? Like that's chickening out though.
00:11:10
Speaker 4: Like the thing is, it's like, we could all do that. Like this is, I guess, like part of what's interesting about modelers and what's interesting about prediction markets is We could all say, yeah, it looks like Democrats, like if you're going to provide value as the creator of a model, you got to be able to say something beyond it looks like a good night for Democrats, but who knows?
00:11:31
Speaker 3: And look, in 2024, like our model literally was 50-50. Like literally we ran 80,000 simulations and it took like the 71,000 simulation for one candidate to go ahead, right? So there are times when the correct modeling position is like, oh, people are too sure of themselves. I'm not sure if this is that election, right? If you go and look at Republican-sponsored polls or go look at where Republicans are spending their time and their resources, and they're bracing for a landslide, right? They are doing telehauls and the Louisiana Senate race and things like that, right? And seem concerned about South Carolina, let alone Texas, who people kind of knew would be a competitive race. But, you know, Democrats, I think, felt like they have been betrayed.
00:12:17
Speaker 4: Is there a Senate race in Louisiana this year?
00:12:19
Speaker 3: There is. I don't think there's been a single poll of it. But prediction markets say there's a 10% chance that Democrats will win that race, right? I think it's a little ridiculous.
00:12:26
Speaker 4: But that's not nothing. The point is that's far from trivial.
00:12:29
Speaker 3: No, we've had to learn these details about these very red states. Like, oh, South Carolina was a Trump plus 18 state. That's what I'm a little bit more skeptical.
00:12:36
Speaker 2: About, right?
00:12:37
Speaker 3: But the vicissitudes between what differentiates Kansas from Nebraska and things like that, if the national environment... Is as polls say it is. Yeah. Democrats are ahead by about nine points on the generic ballot, which are national polls, essentially.
00:12:51
Speaker 2: Right.
00:12:52
Speaker 3: At nine points, Republicans, there's too much of a headwind for Republicans.
00:12:57
Speaker 2: Right.
00:12:57
Speaker 3: And like, you know, you can predict exactly which winning tickets Democrats would draw. But the GOP has to be that spread.
00:13:05
Speaker 2:
Right.
00:13:05
Speaker 3: Or it just gets too difficult. difficult and like i think part of the problem with mental models versus actual models right people say well i can easily see this case where there's a bust again in the polls in the midwest right and i can see this case where sure the polls are basically right and polls are a little cautious and democrats have a sweep right and people kind of like wait those 50 50 but first of all there are the messy middle in between stories and like if you have a lead in the polls then you kind of muddle through half the time right it's like an nfl game where you have a field goal lead at halftime and then lots of penalties and turnovers, right? And you wind up winning by a field goal, right? So it's ugly, but Democrats tend to win the ugly races. And like, there is more evidence, including non-polling evidence consistent with the Democrats have a big night story. Like if you just talked about the things we already talked about in the first 15 minutes about gas and Trump approval ratings and Iran and all that stuff, right? You know, it's not a case where you have this clash really between indicators.
00:14:18
Speaker 1: I feel like we've been talking about this for many, many years now, but there's this persistent tension between all the consumer sentiment surveys where people report how they're feeling about the economy. Those have been very bad for many years now versus the actual numbers in the economy on consumer spending and other things that look pretty good. Do you have a story you tell yourself as a forecaster, a story to explain that discrepancy?
00:14:46
Speaker 3: Slightly hot take. I think some of the consumer confidence polling is not very good. You know, the good thing about election polls— I don't think.
00:14:53
Speaker 1: That's a hot take, by the way. Like, we've been talking about it for a while.
00:14:55
Speaker 4: The pollsters are not good. They're not doing best practices at polling.
00:14:59
Speaker 3: We had an article that's been published by a guest author named Joel Wertheimer, who.
00:15:03
Speaker 4: Oh, Joel's been on our podcast.
00:15:04
Speaker 3: He's great. Yeah, and clearly, I think the University of Michigan survey changed the composition to online in a way that just has, like, way too many Democrats in the sample. So, like, ironically, like— Ironically, like— The complaint that people make about election pollsters, which is more or less false, we can talk about that more, is kind of true of these consumer sentiment surveys that like, for better or worse, election polls get a dose of reality every two years. Election polls are made in an environment where pollsters are actively fearful of missing, particularly missing high on Democrats, right? So that's why like, that's why people talk about like, what's the difference between like polls and prediction markets? Polls kind of are prediction markets, right? or polling averages are at least, right?
00:15:47
Speaker 2:
Okay.
00:15:47
Speaker 3: A poll is a data-informed opinion at this point, right?
00:15:55
Speaker 1: Because you have to weight it.
00:15:56
Speaker 4: The pollster has to do so much modeling, right? Is this what you're getting at?
00:16:00
Speaker 3: So like Marist released a poll of Maine last week, and the top line number, let me see if I'm getting this right in my head, said Troy Jackson, the Democrat in Maine, is up nine points, which would be a very good result for Democrats. The average, our average, had been like D plus three in that state, right? That's their registered voter poll. They applied a likely voter model that brought that lead down to plus two.
00:16:21
Speaker 1: People self-correct for the Democratic.
00:16:24
Speaker 3: Bias, right? Yeah, there's no way. First of all, the math, just I don't want to get too deep in the weeds, even though it's odd lots. Like the math doesn't, the math doesn't make sense, right? They decided what result they felt comfortable with and reverse engineered.
00:16:36
Speaker 4: You see, we just, okay, we don't have to get into the math. But what you're saying is they had a poll that said Troy Jackson was up 9%. But in our likely voter model, it's 2%. And what Nate Silver is saying is like, I don't know how you got from nine to two.
00:16:51
Speaker 3: No. And the governor's race went from 11 to zero. Like the math literally doesn't add up. And like, so it's like, okay, so this is your poll and this is your opinion.
00:16:59
Speaker 1: Yeah.
00:16:59
Speaker 3: Is the way I see it. And like, and we're aggregating opinions. And so like on the one hand, when you see a poll like that, I'm like, I have existential doubt about the accuracy of polling for sure.
00:17:08
Speaker 2: Right.
00:17:08
Speaker 3: On the other hand, if you have a model that's like a consensus model, thermometer that looks at the correlations and translates poles into probabilities. Like, that's really what Flipper, my model, is doing. And, like, so, like, oh, actually, I kind of like that from a modeling standpoint because it proves the kind of underlying hypothesis that, like, polsters are rational actors and, therefore, although poles are imprecise instruments, there is a lot of incentive not to systematically miss. I got it.
00:17:39
Speaker 4: So what you're saying is because you can look at the math and say, look, You can't really justify turning a plus nine into a plus two as someone who aggregates polls and then creates a model that can give you some sort of confidence that like.
00:17:56
Speaker 3: Confidence that the errors in the long run, if you have elections every two or four years, will never have data to prove it. But confident that the errors are symmetric in the forecast.
00:18:05
Speaker 4: Yeah, I get exactly what you're saying. So, OK, people don't pick up the phone like they used to.
00:18:10
Speaker 2: Yeah.
00:18:10
Speaker 4: You know, certainly they don't recognize it. And if someone calls me and asks for my political opinion, no way. So then there's this vision. Well, why don't we just create an AI replica of Tracy Alloway and an AI replica of Joe Weisenthal and an AI replica of a high school educated oil worker in Texas and an AI replicated version of a HR professional who works for General Motors in the Detroit suburbs. And we just pull them. And then we don't know. And they're always going to pick up the phone. What's wrong with this?
00:18:42
Speaker 1: This is the digital twin idea.
00:18:43
Speaker 4: Why don't we just create a digital twin of all America?
00:18:46
Speaker 3: So the problem is like, and by the way, I think these companies are smart. I know some people that work for them, right? But like polling cannot be outsourced to AI agents, right? Because it's one of the few sources of ground truth that we have anymore, number one. And also the models are kind of, so if you ask a poll to create like a synthetic Hispanic respondent, right? 99% of their synthetic responses Hispanic respondents love soccer. It's a bit like if you ask like Claude to pick a number from zero to 10, like always pick seven. It's a bit like, I mean, this gets into the whole existential thing about how like large language models work and like they're kind of, it's kind of the ideal type or something, but like, but no, we need original data, right? Models like Flipper or prediction markets can aggregate that data, but, but, you know, because like voters aren't all that predictable. You know, I don't even like the kind of slicing and dicing that you see. Oh, the Hispanic vote, right? Well, the Hispanic vote is younger Hispanics and older Hispanics, people who are relatively new to the country and people who have been here for a long time, right? Cuban Americans and Puerto Ricans and Mexican Americans and Dominicans and everything else. And like those groups don't always agree with one another. Sure. And people miss it. Like the Hispanic vote is one of the most swingy demographics in the United States. And so like, yeah, it just gets overly reductive and kind of almost like a snake eating its tail a little bit.
00:20:08
Speaker 1: I feel like turnout in particular would be difficult for a digital twin to model? Like maybe you can come up with political preference based on demographics, but like predicting whether someone is actually going to leave the House and vote seems particularly difficult to me. But tell me if I'm wrong.
00:20:23
Speaker 3: Yeah, we did a little simulation in South Carolina, which is one of the more surprising races to be polling closely. And, you know, and one more, frankly, be a bit more surprised Democrats won.
00:20:32
Speaker 2: Right.
00:20:32
Speaker 3: But like, you know, what if black voters in South Carolina turned out at 2024 rates and white voters turn out at 2022 rates, like that shifts the top lines by like 10 points, right? It's a deliberately extreme example. And in South Carolina, the white vote is very conservative. But,
you know, that's an example of like, you know, because you have had some midterms where 2018 in particular, you had some states getting up to presidential type turnout. But if it's differential turnout, that would you need that for a state like South Carolina to flip.
00:21:04
Speaker 4: The Democrats have 14 percent in Mississippi. on polymarket.
00:21:07
Speaker 3: Yeah, that one's a little weird.
00:21:09
Speaker 4: These are funny states. These southern states are pretty funny, right? Because the top line votes are actually offered frequently fairly close. It's just that every white is a Republican and all the non-whites are not Republicans. And so while it looks really close, it's just like, it's just purely demographics.
00:21:28
Speaker 3: No, so we have a concept in our model that we call elasticity, which is how sensitive is a state to changes in the national race. And that's based on voter-level microdata where we see people clustered at the tails or more in the middle. And, you know, South Carolina, but especially like Louisiana, Mississippi, have very low elasticity. So that's why. So we are lower than polymarket and Democrats in those states.
00:21:50
Speaker 1: Actually, since you mentioned demographics, can we talk about gerrymandering for a second? So to what degree do you see that, I guess, backfiring on the Republicans at this point?
00:22:00
Speaker 3: So, yeah, a gerrymander where you wind up with fewer seats than you would have under the original maps is a dummymander. Is a term for it?
00:22:06
Speaker 1: This is an actual term.
00:22:07
Speaker 3: It's a term. We have terms for everything in politics. Only poker players have more terms in politics. In Florida, Ron DeSantis drew a map for kind of a presidential year baseline. In presidential years, I expect Florida to remain a pretty red state, right? It's not robust for like a D plus nine climate where Florida goes from R plus 13 to R plus four, right? or something, right? It wasn't designed for that. And if Democrats kind of run the table in Florida, then they could actually wind up with more seats than they would have under the old map, right? In Texas, Republicans hope to make a lot of their gains from, I guess, this maybe frankly arrogant presumption that, oh, now Hispanics in South Texas are now our voters instead of being swingy. And so, like, that probably won't be a dummy mander, but could be a case where the places they hope to maintain gains, you'll have Democratic incumbents holding on.
00:22:57
Speaker 4: Intuitively, if you're a mapmaker... And your only goal is to maximize the number of seats that your party wins. You have to take some risks, right? Is this the key idea here?
00:23:08
Speaker 2: Correct.
00:23:09
Speaker 3: I don't even think they necessarily played it wrong, right? So, you know, again, the House is not quite a sure thing. There is a 5% chance. The way the 5% comes up is that if there is a systematic polling error and the map winds up being a
little bit less blue, then the gerrymanders become more effective again, right? So there's like, you know, it's probably not enough for Republicans, but there's like this exponential, there's inflection point at which Joe Mandarin goes from not very helpful to helpful for Republicans.
00:23:35
Speaker 4: There's definitely some analogy here to like the options market and convexity.
00:23:40
Speaker 2: Oh, probably.
00:23:41
Speaker 4: I can't quite. I'm not smart enough to articulate it, but there's clear someone smart enough will be able to come up with some what the what we're really what would we be talking about in the financial version of this conversation?
00:23:53
Speaker 1: Yeah. Is there a potential upset that you're particularly interested in at the moment or that you think people aren't paying enough attention to?
00:24:01
Speaker 3: I think governors' races don't get enough attention in general. According to our modeling, we could wind up with almost two-thirds of the country being represented by a Democrat, but there are lots of big gubernatorial races where, unlike in the Senate, in the Senate you have lots of these one- or two-point Democratic leads. In governorships, it's the opposite. Abbott is ahead in Texas, but only by a A couple of points. I don't think I'd bet against Abbott given the odds that Flipper gives. But like I think that governor race is really important in part because Abbott is the most prolifically pro data center governor, I think, in the country. And one of the big things I think.
00:24:39
Speaker 4: And even he's not that pro data center anymore.
00:24:41
Speaker 3: No. So let's say you have at least one Democratic chamber of Congress. Maybe, you know, even if Democrats only got to 50 Senate seats, you don't have the Senate. But like you're going to have a lot of gridlock. I think the one policy area where you could see action before 28 is on AI, right? I mean, if the timelines are even a third as aggressive as the labs think, then they'll kind of be forced to take action. But I'll be looking for, is there evidence that AI had a big effect in this race and that people were seeing this through pro-AI, got punished for it?
00:25:26
Speaker 4: My son the other day told me he wants me to bring him up more on the podcast.
00:25:32
Speaker 2: Oh.
00:25:32
Speaker 4: He wants me to tell more.
00:25:33
Speaker 1: Is he listening?
00:25:34
Speaker 4: No, but he's like, he's like, he's like.
00:25:36
Speaker 1: Just tell me you mention him every day.
00:25:38
Speaker 4: Can you make me famous by telling, mentioning me more on the podcast? So I will tell a story. I've mentioned before, he's very into money. We were watching a University of Texas football game the other day and there was this, there was this ad that, for the Abbott campaign. I don't know if it was the official Abbott campaign or one of these outside groups. And it said, Gina Hinojosa's income tax plan will cost your family thousands of dollars. And my son turns to me, he's like, Dad, we're not voting for her, right? Oh, no. It's so funny. And I was like, this is like, you are exactly who this ad was targeted towards. My seven-year-old, you know, it's like, he's just like, Dad, we're not going to vote for her, right? This candidate that costs thousands of dollars.
00:26:20
Speaker 3: No, I was watching the World Series, and there were like three straight ads in the Mike Lawler, the Westchester County district, right? I'm like, no, even in New York, you can't really escape.
00:26:28
Speaker 2: Oh, my God. Yeah.
00:26:30
Speaker 1: Yeah.
00:26:31
Speaker 4: At the end of the day, do you have a view, you know, there's the view of, well, should campaigns invest more on like things like turnout and knocking on doors, et cetera, or should they just spend a bunch of money on ads? Do you have in your, any of your research, do you have views on some of these questions?
00:26:50
Speaker 3: I think with ads, you guys, I.
00:26:51
Speaker 4: Think like David Shores says, just, just buy ads.
00:26:54
Speaker 3: I think you get pretty sharply diminishing returns with ads. I like David a lot.
00:26:57
Speaker 1: Yeah.
00:26:58
Speaker 3: But everyone in the system has an incentive to say that advertising really works. I think it's better to buy ads to improve voter awareness and expand the map potentially. But look, there's a famous story about Sam Beckman-Fried, talked about this in my book, right? Where for some reason, Sam Beckman-Fried wanted to back an effective altruist candidate in an Oregon house primary a couple of years ago, right? And gave him like, or not didn't give him, right? Spent $ 10 million- or something without his permission on his behalf. And that candidate who at one point was leading the polls wound up losing the election badly because I talked to him for my book and he's like, well, voters just got so much mail and who is this odd crypto billionaire funding, my campaign, right? It wound up backfiring a little bit.
00:27:47
Speaker 1: This is the flaw in all of advertising, right? Like, you're just annoying me with your product placements and your ads and your intrusion into my life, and it doesn't make me want to buy things from you.
00:27:58
Speaker 3: That's why you need, you know, subscription businesses like terminals and.
00:28:01
Speaker 4: Yeah, that's correct.
00:28:02
Speaker 2: Newsletters.
00:28:03
Speaker 4: All right, here's another important question. Let's just say the Democrats end up having a very good night. What will we watch... within that, to know what the future of the Democratic Party is, whether the future is going to be a more DSA-inflected one, et cetera. You know, give us what we might learn about what works for successful Democratic Party politicians.
00:28:28
Speaker 3: For the internal politics of the Democratic Party, the most important race is probably Michigan, where Abdul El-Sayed beat the more moderate Haley Stevens in the primary. But not by much.
Not by much. That was not a great race for polls, by the way. If he wins by enough, of a margin that the race is called on election night, then that will be like a feather in progressives cap. Maybe people like me will say, well, he won by four and the baseline is nine. So, oh my gosh, he underperformed his work. Very bad candidate.
00:28:56
Speaker 4: This is your impression of yourself.
00:29:00
Speaker 1: It's made silver squared, right?
00:29:02
Speaker 3: On the other hand, I kind of think if Democrats have a really, really, really good night and it's like, well, we can dominate all these moderate white guys. Adam Hamilton in Kansas is a pastor. Tallarico is kind of woke progressive, right?
00:29:15
Speaker 1: Yeah.
00:29:15
Speaker 3: But like he kind of has managed to brand himself differently. It might have helped him to have Jasmine Crockett in that primary. He got to define himself as like the sensible middle ground.
00:29:25
Speaker 4: But you talk about the Hamilton race in Kansas and what that would, yeah.
00:29:29
Speaker 3: So Kansas, I have family in Overland Park, right? And for years it's been like, you know, Roger Marshall could be in trouble and I kind of never really believed them. Like, yo, it's Kansas. Kansas has low-key gotten a bit more democratic. It voted democratic on this abortion referendum. It elected Laura Kelly as governor twice. You know, Roger Marshall has these medical billing practices that are very unflattering. And like Adam Hamilton would be like, oh, we just want cheaper gas and integrity in California. In Kansas, right? And like that pitch might catch kind of the opposite. It's a relatively elastic state, right? There are opinions that can change. It's pretty well educated. So like, yeah. And, you know, last week, CNN had a poll showing Hamilton up four points in Kansas. But like, yeah, Kansas is one. Like, I have more trouble with believing the South Carolina polling than the Kansas.
00:30:16
Speaker 4: But basically, there is certainly a possibility that the day after election day, You're going to be like, look, you and Matt Iglesias are going to be like, look at Kansas. Democrats are right to nominate the middle ground white guy. And the DSA people are going to say, look, we are right to nominate. AES wins by nine points. He has a great night, like the polls say. And you just keep fighting.
00:30:40
Speaker 3: No, this is like it's a fairly clean national experiment.
00:30:44
Speaker 2: Yeah.
00:30:45
Speaker 3: Natural experiment, Michigan. I went to there. You know, there's never been a Muslim American elected to the U.S. Senate. So there are. reasons apart from his progressivism, why Elsa could struggle or be good. I mean, you know, he's bro-y, right? He doesn't come across as like, he's bro-y in a very Michigan way. He doesn't come across as like- I was going.
00:30:59
Speaker 1: To say, also, it's Michigan versus Kansas, right? It might be that you like fit the candidate onto the state, right?
00:31:06
Speaker 3: And Michigan's a, uh, You know, they have a strong gubernatorial nominee in our favorite in that race. I don't know if there's some degree of coattails or something. But, yeah, all we're going to, you know, from November 4th onward, the story of the next year and a half is going to be party infighting.
00:31:20
Speaker 4: What about this? Like, isn't there something going on like Florida AG or there's some race in Florida, right, where some socialist seems to be polling well? Do you know anything about that?
00:31:29
Speaker 3: Well, I mean, in the Senate race.
00:31:31
Speaker 4: Is it governor?
00:31:32
Speaker 2: Is it Senate?
00:31:33
Speaker 4: What's happening in Florida?
00:31:34
Speaker 3: In the Senate race, you had a DSA... Democrat Angie Nixon.
00:31:37
Speaker 4: Angie Nixon. Oh, she's running for Senate.
00:31:39
Speaker 3: She's running for Senate, and she's not polling that well. I mean, she's, yeah, Florida, one thing we should probably account for in our models is net domestic migration, right?
00:31:50
Speaker 2: Okay.
00:31:50
Speaker 3: The models keep missing low on Florida because you have a steady flow of red voters into the state, and we're using benchmarks based on the last election, but Florida's a moving target for that.
00:32:01
Speaker 4: It's interesting, if you look at polymarkets, because I keep pulling them up as we mentioned them. And everyone has shown a little bit of a bump for the Democrat in the last several weeks. So Mississippi, South Carolina, quite a big one. Annie Andrews at 28%, which is way higher than she was just a few weeks ago when it was at 13%. No bump at all in Florida. Angie Nixon is at 10%. And that's actually lower than where she was during much of the summer.
00:32:28
Speaker 3: Yeah, and this follows polling, right? The New York Times had some polling that was very strong for Democrats, but not in Florida. They had Ashley Moody ahead by six or seven or something.
00:32:34
Speaker 1: To what degree are prediction markets just like reflecting polling now?
00:32:37
Speaker 4: Yeah, talk to us about that.
00:32:39
Speaker 3: I think they're reflecting polling plus sentiment in kind of a, it's somewhere between smart and midwit.
00:32:51
Speaker 2: Right?
00:32:52
Speaker 3: So again, people in prediction markets do have to quantify ultimately their price, right? And I think that exercise is helpful. When they go play poker, they can be like, oh, I think that guy might be bluffing. I'm saying, I say, well, I'm getting two to one from the pot, right? Last decision I have to make, do I think is a greater than a 33% chance or not? Like, I think the prediction markets have a slightly Republican-leaning clientele, right? Like Democrats are a little preachy about them sometimes. Republicans have more money, right?
00:33:21
Speaker 4: Well, we saw the famous, like, huge whiff in the L.A. mayor race where, like, Spencer Pratt, like, who in their right mind thought a Republican was
going to win a mayorship at a major U.S.
00:33:30
Speaker 3: City? Yeah, but, like, they have, like, these, like, the recognition that polling error is correlated, like we talked about at the top of the show. That's an important recognition to have. The markets were actually overconfident relative to my model in 2016 on Hillary Clinton, right? So everyone knows that, but, like, I think The belief that, oh, polls always underestimate Republicans is not the most sophisticated version of that necessarily.
00:33:55
Speaker 4: Is there anyone systematically making money? I mean, I'm going to put you on the spot here or the Silver Bulletin on the spot. But do you feel that someone could make money? betting on prediction markets anytime there's a deviation between silver bulletin model for a given race and the prediction market?
00:34:14
Speaker 3: Well, look, I would say when our model came out on August 11th, we moved the markets by several points, right? And so like, if you have markets that can look at our model and are influenced by our model, also look at other models, right? And on top of that, incorporate some degree of vibes. I mean, I helpful. They can also just get to things faster, right? If there's a scandal, the New York times reports, on Thursday morning, then it will take several days to show up in the polling, right? And so like, yeah, I mean, I went through recently and looked at every race where you have a gap between Flipper, my model, and Polymarket. And like, I agree with Polymarket on half of those, you know, although ironically, some of the ones I agreed with, I'm like, oh yeah, Flipper's way too high in South Carolina. And then there's this good poll from South Carolina that brings, you know, so maybe you're kind of arguing it to yourself a little bit, but like, yeah. So the test would be like, you have to create like some type of laboratory environment where prediction markets, aren't allowed to look at polls. And my expectation is that prediction markets would be extremely terrible.
00:35:10
Speaker 4: Well, it seems like actually the test would be, you would want to have prediction markets that aren't allowed to look at models, right? Because of course they're going to look at- Fair enough. Right? Because ideally you wouldn't want the day the silver bulletin numbers drop to influence the prediction markets. Ideally, you're looking at polls and creating a model and prediction market traders are looking at polls and creating a price- And the question is, which is the better way to derive a probability?
00:35:38
Speaker 3: Yeah, look, and like building a simulation with all these regional correlations is tricky, right? We are like, as of when we're recording this, we are like literally like 0.995 correlated with polymarket on individual states. We have the overall democratic win probability a bit higher. You know, I don't know if that's because it's attracting a different type of trader, right? Like that, like you get maybe more
of the you know, the whales, right? Not like the French whale from 2024, but like whales, like degenerate rich guys with a lot of money to burn, right?
00:36:11
Speaker 4: The dream.
00:36:12
Speaker 3: Whereas they're maybe less likely to invest in individual races. But there's really not that much of a gap, right?
00:36:20
Speaker 4: You've been building models for a really long time. And of course, you first sort of became a household name in 2008. So that's 18 years ago.
00:36:28
Speaker 3: That's crazy.
00:36:30
Speaker 4: So that means there are people who are going to be voting in this election that were, like, born during the Obama administration.
00:36:35
Speaker 3: Oh, my gosh. Oh, my God.
00:36:37
Speaker 4: Some of these things freak me out a little bit. Can you talk objectively about, like, is it unambiguous in your mind that AI has made it easier for you to build models and to do your job?
00:36:49
Speaker 3: Yeah, for sure. Look, I get to hire two world-class programmers, coders, for programming. whatever I pay for the max plans, right? A hundred or 200 bucks a month. So I have these complex projects that you definitely couldn't one shot or even come close to it. But I know this really well and I've quoted these things myself. And I know when you feed GPT or Claude a 10,000 word program and say, carefully audit this, it's quite sharp about it. Now I would still say on like two thirds of prompts, I have to provide quite a bit of guidance, right? It's not like, oh, cool, man. Next step.
00:37:31
Speaker 2: Right.
00:37:32
Speaker 3: But like, it's great. It's a great teammate, you know, to take something in a more obscure example. So we have an NFL model. The NFL has all these complex playoff tiebreakers, right? They're purely algorithmic. And last year when I built our model, it took me two and a half days to code them up. And like some of the code was buggy. We skipped a couple of tiebreakers. It was very slow.
00:37:54
Speaker 2: Right.
00:37:55
Speaker 3: You know, now you have GPT or cloud do it and they do it in 15 minutes while you're at the gym, right? And it's code is 10 times faster. Now, again, there are dangers, like they can introduce bugs and kind of hide them from you. We built a soccer model too. And like in one version of soccer model, it had the ratings like looping back around. So like the 2026 rating would inform like the 1900 rating and it got, it would do things that were, it was quite weird. But you know, the combination of having like a lot of initiative and entrepreneurial spirit and expertise. Cause there's a certain type of person I meet. You guys are probably in that category, right? Where it's like, it's like people are saying me, I feel like I'm the most productive I've ever been.
00:38:36
Speaker 2: Right.
00:38:36
Speaker 3: I feel like I'm getting two days of work done in, in a 10 hour day sometimes. And I feel like that some of the time, but, but again, I'm probably in the 99.9th percentile of like,
these models are helpful with the exact, I'm not really a coder, right? I understand model architecture. I'm an empiricist. I understand probabilities, but like, my coding was always the weak point. So it's like, this exact thing they're really good at and it's really valuable to me is the exact thing that they do very well.
00:39:04
Speaker 4: Your FIFA World Cup model did have Spain versus Argentina in the finals.
00:39:06
Speaker 3: We had a really good, yeah. And that's the thing about, so NFL, you know, you have good weeks and bad weeks and like, that's why, you know, maybe if Republicans win the House, with the Senate, we'd say, well, 25% things happen, right? It'd be great. But like, If they win the house, maybe we'll just make Silver Bulletin AI in a sports blog. Just give up.
00:39:24
Speaker 4: Just give up. Nate Silver, thank you so much for joining us on our Election Day special.
00:39:31
Speaker 2: Thank you, Joe.
00:39:45
Speaker 4: That was a fun conversation, Tracy. I think my favorite part of the conversation... was Nate preemptively parodying how he will sound if Abdul El-Sayed slightly underperforms.
00:39:57
Speaker 1: It was a good impression of himself. And Nate Silver.
00:39:59
Speaker 4: Doing the impression of an annoying white moderate guy was good. But that was fun. I'm glad. Happy Election Day, people. I'm glad Nate joined us for today. Yeah.
00:40:09
Speaker 1: By the time this comes out, I guess maybe some of the polls will have changed.
00:40:12
Speaker 4: Maybe some of the polls.
00:40:13
Speaker 1: We'll have some additional.
00:40:14
Speaker 4: Data, maybe. We'll have some more early voting data.
00:40:16
Speaker 1: Yeah, exactly. One thing I thought was really interesting was this idea of polls basically all becoming models now.
00:40:23
Speaker 4: Yeah.
00:40:23
Speaker 2: Right.
00:40:24
Speaker 1: And like after years and years and years of being accused of, I guess, overestimating Democrats in the polls, everyone now is self-correcting. And so the question is, like, how big is that self-correction?
00:40:37
Speaker 4: Yeah. And it's interesting. And I guess we didn't go into the math specifically. But if you can empirically establish that your self-correction here makes no sense.
00:40:47
Speaker 2: Yeah.
00:40:47
Speaker 4: Then what you do learn is, as Nate said, pollsters are self-interested players. Because who is it? What was the famous pollster in Iowa? Seltzer, Ann Seltzer, who actually, you know, right before the 2024 election actually had Kamala Harris winning Iowa. Yeah.
00:41:04
Speaker 1: No one wants to poll a Seltzer, right?
00:41:06
Speaker 4: It was so off that she retired from the game. And that's kind of, I guess, what Nate was saying, too, which is that if Republicans win the House, it's like, you know what, I think we're done. I think we're done trying to go into sports. These crazy American voters. We're done trying to understand anything about them because we have nothing. We have no usable information on them. Go ask your digital twin.
00:41:27
Speaker 2: All right.
00:41:27
Speaker 1: Shall we leave it there?
00:41:28
Speaker 2: Yeah, let's leave it there.
00:41:29
Speaker 1: This has been another episode of the All Thoughts Podcast. I'm Traci Allaway. You can follow me at Traci Allaway.
00:41:35
Speaker 4: And I'm Jill Wiesenthal. You can follow me at The Stalwart. Follow our guest, Nate Silver. He's at natesilver538. Follow our producers, Carmen Rodriguez at Carmen Arman, Dashiell Bennett at Dashbot, Kale Brooks at Kale Brooks, and Kevin Lozano at Kevin Lloyd Lozano. And of course, you can chat about all of these topics 24-7 in our Discord, discord.gg slash allthoughts. And for more OddLots content, you can check out our daily newsletter, which you can find at Bloomberg.com slash OddLots.
00:42:03
Speaker 1: And if you enjoyed this conversation, if you like it when we do our Election Day specials, then please leave us a positive review on your favorite podcast platform. And if you're watching on video, then please leave a comment or like or better yet, subscribe.
00:42:17
Speaker 4: And remember, if you're an OddLots subscriber, you can listen to all of our episodes absolutely ad-free. All you need to do is find the Bloomberg channel on Apple Podcasts and follow the instructions there.
00:42:27
Speaker 1: Thanks for joining.
00:42:28
Speaker 2: So