And it feels like everything is kind of changing again, because we've just passed the bank holiday weekend they call labor day here in america, and that is always a signifier in an election year that it's now gearing up properly. There's a lot of people who, frankly, don't have the time or interest to pay any attention to politics over the summer holidays. But now that you know, there's an autumn chill in the air, that back—to—school feeling everywhere, this is typically when people start tuning in and paying some attention to politics. So it's almost as though the election actually starts this week, marianna, if you can believe it. It started months ago for us. But for many voters, it starts now. Yeah, i know. We slightly envy the people for whom it starts now, or the people for whom it starts even later. And actually, speaking about big stuff that's coming up, we've obviously got this presidential debate that's happening next week, isn't it, sarah? i guess that will feel like a moment in this final stretch of the campaign, particularly because it's kamala harris and donald trump coming up against one another for the first time. And you know they've never met before? this is actually going to be the first time that they ever speak to each other. They've never met?! wow! she was in congress a couple of times when he gave state of the union addresses when she was a senator, but other than that, no, they've never looked each other in the eye. And they will on this debate stage. I think it's going to be a huge, pivotal moment for a lot of people really, really interested to see what the dynamic�*s like between them and, principally, whether or not she can stand up to what's almost certainly going to be an attempt at some kind of bullying or intimidation from him. And how she handles that could have a huge say, i would think, in whether or not she wins this election. And what kinds of clips get picked up and all that kind of stuff, what goes viral, what doesn't. And i guess afterwards, sarah as well, there are a lot of people who are going to be talking about the polls and whether the polls move or not, whether it looks like one candidate comes out on top or not. Are the polls to be trusted? do the polls have a track record of being trusted in the states? yes and no. It depends which ones you're talking about. And that's something obviously we can get into with our guest, nate silver, when we speak to him in a minute or two. I think the key thing to say about the polls is that they're still incredibly close. There is momentum in kamala harris�*s favour. She has gone. . . Joe biden was, like, maybe on average, three or four points behind donald trump in national polling. Kamala harris is now slightly ahead. But it's fine detail here, and it's so close, in such a divided country, it's really difficult to know when things are in the margin of error, whether they're correct or not. And just, you know, a few hundred thousand votes could make the difference. So whether polls can pick that up is difficult to say. One thing that i have noticed in the world i investigate is that when the polls don't get it right, that can have quite a big impact on, i guess, people's trust and confidence in the systems, because quite a few of the conspiracy networks or groups that i spend a fair bit of time hanging out in will seize on when a poll is wrong and use it as an example of where they believe the establishment, or what they'd call the kind of mainstream media, or anything of that variety is trying to manipulate the outcome of the election by saying, oh, well, this is what's going to happen, or, this isn't what's going to happen. And so it feels as though actually, this time around, particularly with everything that happened back in 2020, there's more pressure than ever, really, for the polls to get it right. Yes, and of course, people don't always read polls correctly. I mean, any poll that's published now is not saying this is what's necessarily going to happen in november, because all sorts of events could take place between now and then. What they're saying is this is a snapshot, as best we can capture it, of what people are thinking now. So that can be one of the reasons why people will say, oh, but you know, so—and—so was 20 points ahead all the way through the campaign. How could they have possibly lost? well, the answer is things changed and voters changed their minds. But this is all going to be very fertile territory for our guest who's coming up next. Now, this is rather exciting because we're going to welcome to americast now someone who is a genuine celebrity in the world of political polling and predictions. There aren't very many of them, but at the top of the tree is nate silver, who shot to prominence in the 2008 election when he got 49 out of the 50 states correct in how they voted in the 0bama—mccain election. He got all 50 correct in 2012. And so he is a big name in the world of political polling who has now written a book about risk and gambling. And yeah, that book is called on the edge — the art of risking everything. And it's brilliant to have nate sitting here next to me in the americast studio. Hi, nate. Thanks to both of you. Happy to be here. Thank you so much. It's always a bit awkward when everyone has to talk about you when you're sitting here. I'm used to it by this point. Good, good. Right, ok, let's start by introducing you a bit. So you started in poker and then went into political polling. Yeah. There's this line — you talk about feeling more at home in a casino than at a political convention. Tell me about that. What's that like? yeah, i mean, the reason i got into elections and politics is because the us government basically passed a law to take my livelihood away as an online poker player in 2006. They basically made it illegal to deposit money in offshore online poker sites. So i started following the us congressional election that year, hoping the people that had passed that law would get voted out of office — which they did, by the way, thanks in part to contributions from poker players. And then in 2008, i'm, gosh, 29, 30 years old, living in chicago. A guy named barack 0bama is running for president. I went to university of chicago. He's at the law school there. A much more exciting politician than the kind ofjohn kerry, george bush—era politicians that we were getting. And meanwhile, in the us, there was this obsession with data everything. In sports, we'd seen moneyball and the data revolution, and so, you know, someone with a statistical background and a background in forecasting and gambling, right? i made forecasts of how major league baseball players would do, for example. So the idea was to turn that onto politics and issue forecasts for upcoming elections. Typically after a convention, basically, no real news value. It's four days of free advertising. You have a bunch of advertising that airs. You'd expect to be at a high watermark. So our forecast thinks that her numbers might decline a little bit. And so basically that gets us to a 50/50 toss—up. And, you know, our model, if anything, maybe leans more into that latter case, because she has had a period of momentum. But like, you know, you have a debate coming up in less than a week now that might reset that momentum potentially. And again, pennsylvania, that's the state that might be a little bit of an achilles' heel for her. A cnn poll came out this morning that had pennsylvania, as a literal dead heat — tied. So, look, we are in unprecedented circumstances. You've never had a candidate take over injuly before. It's not quite a uk snap election, but it's a different rhythm, a different timing. And so everyone's making making a best guess. Look, in the past two elections, the democrats won the popular vote by two points and four and a half points, respectively. So if she's at three and a half points, then that's right in between clinton and biden. Of course, clinton lost the electoral college and biden won it. So maybe that's another way to get to 50/50, basically. You talk a lot in your book, on the edge, about risk and about risk—taking. Just to stick with this election before we broaden it out a little bit, how would you consider the risk that the white house took or didn't take? i'm really interested in that risk, which was, right, do we decide whetherjoe biden keeps going and it looks like we might lose, or do we gamble on someone else, and then we gamble on kamala harris? like, how do you assess that risk? i mean, i don't think it was really a gamble. It was obviously the right play, right? but it kind of felt like a gamble, like, they made it feel like a gamble at the time. People mistake change forfor gambling, right? if you're going down the road and the bridge is incomplete and you're about to fall into the grand canyon, or you can take a nice little off—ramp and exit and slow down and then, you know, come to a resting stop, then it's not risky to actually make the turn, right? like, this was obvious. Biden should have, a year ago, recognised that 80% of americans thought a year ago that, no, this guy cannot be president until he's 86, and should have, you know, stood down and made plans to have maybe a more smooth transition, maybe have a competitive primary, potentially. So, look, it took biden a long time to see the writing on the wall. But the democratic party is ultimately very effective. I mean, nancy pelosi and jim clyburn and party leaders, they are not a cult of personality in the same way that republicans are under trump. And ultimately they were able to persuade biden and use leverage. But i gave 100% of the credit to the democratic party and and 0% to biden for stepping aside at such a late hour. I don't know that it damaged harris. I mean, in some ways, having this abbreviated campaign, maybe americans don't need these, like, two—year—long campaigns. But, you know, he should have seen the writing on the wall sooner. And what's happened this year with the election because of all of that, because of the focus on biden and his age and then the speculation about whether or not he would pull out, it's made this campaign, even more than usual in america, focus very much on the personalities of the candidates. Yes. And people now looking at kamala harris as well as a personality, maybe, rather than a policy platform. How much does it matter, especially when donald trump is so vague often about policy? are voters making these decisions based on who they think will be a better steward of the economy? 0n who they think will be a better steward of the economy, how they feel about abortion and reproductive rights, how much they care about immigration? or are theyjust looking at two people, and as we've talked about in americast before, just reading the vibes of them? well, look, you mentioned the three issues that probably does matter. I think the economy and abortion and immigration do penetrate through. But for the most part, for 70% of people, it's probably mostly about the vibes, the personalities. Actually, for most people, it doesn't matter. It'sjust about the, do you have the d or the r name label by you? but yeah, americans are not, like, reading party platforms and scrutinising in detail the messages of the campaigns. You know, even for things like abortion, there are there are some people who think that joe biden was responsible for overturning roe the wade, for example. So i think the harris campaign has been smart to lead with personality. I think they may need to pivot or maybe need to pivot a little bit sooner. They haven't done, like, a lot of press availability, which i thought was kind of kind of silly and, you know, you're the vice president as well as the democratic presidential nominee. Go ahead and do a bunch of interviews. I think their excuses for doing that are poor — maybe leftover trauma from when biden gave interviews and they often didn't go well, but she's more capable than he is. But yeah, i'm not going to pretend that it's the most substantive election, but of course, it's still incredibly important. The result is very important. And so i think the harris people have figured out, ironically, that given the stakes of the election, going around like biden did, saying, democracy, democracy — democracy is on the line, people are a little bit tired of that message after being told that. Message after being told that — even if it's correct — being told that for eight, 12 years now, it's the most important election of your lifetime. And to have a little bit more fun with things, i think is not a bad idea. It's really interesting you say that. When it was the caucuses earlier this year, i was in very, very cold iowa. And i was hearing that from some of the people i was chatting to there, and then also in atlanta, that idea of, well, actually, i'm not really motivated to go out and vote because i'm worried about the future of democracy. I'm motivated to go out and vote because of what's actually affecting my life. Yeah, it's awfully abstract, right? i mean, america hadjanuary sixth, which is, you know, a terrifying event. It's ultimately, though, a near miss, right? if something had gone differently, maybe you had had members of congress injured or killed, for example, or you would have disrupted the electoral vote count somehow. But it's a near—miss, and people don't tend to learn lessons from near—misses, right? if you drank too much and you drink and drive, right, and just narrowly avoiding an accident, you might even say, oh, look how great a driver lam, right? and so that's a little bit where it feels like america is. But it was a hard pitch for biden to make, given that he was the president, he became president, and you had a fair midterm in 2020, 2022, where democrats actually did pretty well for the most part, and they still control the senate. And so, to say democracy is on the line may be accurate. I mean, look, we could debate that. But you know, i think it's kind of not as obvious a message to voters as the biden white house seemed to think it was. And on the issue of polling and the accuracy of polling, i mean, you mentioned january the sixth there. I spend way too much of my time investigating social media, disinformation, all that kind of stuff, and something i've noticed is how when polls get it wrong, that can then contribute to people not trusting perhaps the public institutions they could. Do you think the polls do often get it wrong? i think the media has to be more responsible when it reports on polls. You know, to bring things a little bit closer to home, you know, the media massively misreported what the polls said about brexit in 2016. The polls said that this was, you know, remain and leave were roughly 50/50. And i think the kind of london—based media preferred remain and tended not to emphasise the uncertainty in the polls. As you both know, there are more ways that polls can go wrong than just the official margin of error in the polls. The main issue being that, like, it's hard to have an equal likelihood of reaching different groups of people, that if you just kind of call people in the phone book, you'll get a bunch of old white women, basically, answering the phone, right? they're the ones who still answer landline calls. You have more have more trouble getting young voters or voters of colour, or certain types of male voters, for example, or in america, we have, you know, hispanic, spanish—speaking population that actually will vote and are citizens in some cases. So understanding the different sources of polling error, understanding that if we go into election day and kamala harris or donald trump is up three points in pennsylvania and wisconsin, etc. , that's still a very close race. Ithink, you know, that's what we try to do at my newsletter, silver bulletin, with the probabilities that we list. But, you know, i have seen cases where, you know, the media blame the polls for their own, failures of context, i suppose. We're the bad guys — not me and sarah, though. Well, look. . . I'm joking. The other thing, too, polling is important in a democracy because, except in states like california where you put everything up to a referendum, we don't have direct democracy. And so to give voters a way to weigh in, and to make sure that elites like us, the bbc or the nbc or new york times or whatever else aren't imposing our views, and polling serves as a check to give the common person a voice. So if polling is bad when we actually test it on elections, then we should be worried about losing that device, that's, ithink, you know, an important lever we have in democracy. So what should we be being careful of at the moment? should it be that because democrats are quite excited and enthused about kamala harris at the moment, they're more likely to answer calls from pollsters? is it difficult still to find trump supporters, or are they less likely to respond to pollsters? where do you think there might be some sort of in—built bias in the numbers we're looking at now? yeah, the enthusiasm thing is always a little bit of a concern because you might think, oh, it's great in the abstract for a party to be enthusiastic, but plenty of unenthusiastic people vote too. The concern is that when you have a party that's very enthusiastic, you have what's called partisan non—response bias, meaning that one party is more likely to respond or not respond to polls than others. There were other wrinkles. In 2020, one issue some pollsters believe was that covid. . . So, democrats were more cautious over covid. They are sitting at home, the pollster calls, you're like, shoot, this is exciting. A pollster called. I've got nothing to do. I'm going to answer this pollster's phone call. I'm going to vote for joe biden and kamala harris. Meanwhile, you know, republicans are, like, going out to the local dive bar and things like that, and, you know, partying, whatever else, don't care as much about covid. Missed the pollster's phone call, don't get recorded in the polls. So there are lots of ways that polls can go wrong. Moving on to your book a little bit more, and some of the themes that you explore, ai is something that you talk about and very specifically is something that you talk about and very specifically, the risks of ai and how we balance up the positives versus the risks. What do you decide in the end? where do you land in terms of how we should be feeling right now about al? i suppose i don't. I mean, there's a lot of reporting in the book where i'm trying to present the case for both sides from, like, the most, you know, thoughtful and qualified people on both sides. Look, i don't think that we're in imminent danger right now of some type of, like, fast ai take—off scenario. But these models have come farther, faster than most people expected. If you use chatgpt or claude or another ai. . . Not quite chat. . . They're large language models. You know, the fact that you can have this kind of talking computer that comes close to passing the turing test, depending on who you talk to, would have been considered miraculous by all but a handful of people five years ago. And it's not perfect — it hallucinates, it spreads misinformation sometimes. It's very human—like, often, actually, in its ability to kind of lie to you and spin its way around things and sometimes even flatter you a little bit. But still, it's kind of a miracle that it advanced this far this soon. If there's another miracle and it develops superhuman capabilities, then all bets are off, potentially. Would those superhuman ais behave the way we want them to is an open question, potentially. And that's the reason why some people are concerned. And by the way, we don't know that much about the inner workings of ai. It's not like the industrial revolution where you design