The Ultimate Guide to Risky Decisions: Risky Business with Maria Konnikova and Nate Silver: Maria Konnikova and Nate Silver

24:25Steven, Brandon, and Bruno, the business owners of SandCloud, reinvested their 2 % cash back to help build the company's retail presence. Capital One, what's in your wallet? Find out more at CapitalOne.com slash Spark Cash Plus Terms Apply. What can 160 years of experience teach you about the future? A lot, especially when it comes to protecting what matters. At Pacific Life, we've helped people and businesses confidently embrace the future with life insurance, retirement income, and employee benefits. With strategies built on strength and trust and a promise to be here for you today and tomorrow.

25:03Ask a financial professional how Pacific Life can help you today. Pacific Life Insurance Company, Omaha, Nebraska, and in New York, Pacific Life and Annuity, Phoenix, Arizona. And I think also maybe the use of AI now, we're kind of on a trend because AI can basically create any kind of content. We're kind of on a trend where if you see any piece of data, news, information, whatever, you should probably just assume it's not real. Eventually, all data will have equal probability of not being real. Yeah. Actually, it's quite scary because if you look at the psychology of how the brain processes information, the model that kind of is the established one in the sense of this is probably true is that our brain has a default true switch, Right.

25:55So we assume everything is true when we first encounter it, because that's the way we understand reality. Right. Like I can't question all the time, like, is this really here? So like you see it and you're like, okay, water bottle, um, you know, we're true. And then that's effortless, right? So like everything that comes in is effortlessly tagged as true. And then there's always a second step where we say, okay, is it actually true? Right. Like where you say, wait, No, like false, right? Like that's not actually something that exists. This is false. However, when there's cognitive load, when there's anything going on, that second step sometimes doesn't happen.

26:34And so we incorrectly remember things as true, even when they're false. This is why like when you're on a jury, if some piece of evidence is introduced and then the judge says, ignore that, like do not use it. I mean, that's not going to work, right? And it never works. Which is why they always, it's why the lawyers always say they know that it's going to be thrown out of the court. Exactly, but you do it anyway. And that initial thing will then just stay in the back of your mind and cloud the information later on. And before, like, before AI and all these things, like, I think in our world it used to be easier to realize, okay, because it is effortful.

27:14And so we have limited bandwidth, limited cognitive resources. So it used to be easier to figure out, okay, what are the things that I need to kind of verify? And some things, obviously, if I said, you know, Nate, yesterday I saw a pink elephant walking down Park Avenue, you'd be like, okay, no. For one second, your brain had to believe it to process what I was saying. But then you're like, okay, pink elephants don't exist, right? And even if someone dyed an elephant pink, it's not going to be walking in the middle of Manhattan right now. And so false. But when the internet is full of statements that look like there's a gray elephant that just escaped from the zoo, which seems plausible enough, right?

27:57We're not flagging it as pink elephant walking down the street, but it might be. And now we can't distinguish between the two. And so the load on our brains is such that I think a lot of times shit's just going to get tagged as true because you can't verify every single time. And actually, it's interesting because when I was writing about con artists, I got exhausted at some point because you realize that you have to fact check literally every single sentence out of their mouth. And so you start having to pick and choose. You're like, okay, I don't have time to fact check a five-hour long interview.

28:32So I'm going to fact check like the most important things. And then I ended up figuring out that this guy made up that he had a sister. And it went into my book because it didn't even occur to me that he would make up having a sister. And I didn't fact check it. And Nate, I'm guessing that you probably encountered this when you interviewed SBF. It's exhausting. And when we have to do it literally all the time, I don't think our brains can't do it. what's the axiom where like it takes 10 times more energy to like disprove bullshit than to create bullshit? I mean, maybe it's like 50 X more, right?

29:08And look, this is probably why I feel sympathetic to people that I have differences of political opinion with, right? Like if it's my job to follow politics and policy, then I can make these like fine distinctions about, yeah, this thing isn't precisely true, but it's directionally right or this and that. You know, If you have another job apart from politics, then it's hard to distinguish these things sometimes. And even things like, you're talking about what if you saw a pink elephant. The other day I saw a Clydesdale horse on Fifth Avenue in Manhattan, which makes sense because for some reason they still allow horse-drawn carriage rides.

29:46But it's utterly bizarre in some sense that you have a Cybertruck and then a horse-drawn carriage and then regular Honda Accords. It doesn't really make a lot of taxis. doesn't make a lot of sense. Yeah. Then you have stories of like different farm animals getting onto the subway tracks in New York. And so, yeah. So, so like weird, weird shit does happen. And I think, but it's even like, I don't think our brains are like, we're literally not built for this because for AI and all of these reality manipulations, because at least before you had to parse political statements and that kind of stuff.

30:22Now we have to parse like what we're hearing, wait, it sounds like Nate's voice, right? But what if it's a deep fake? What we're seeing, like, wait, it looks like Nate actually was here and was involved in this bar fight, but that doesn't seem like Nate. So like, let me, is it a deep fake, right? So like all of our senses are constantly being manipulated and it's happening all the time and the technology is getting so damn good. And then there's the next level of this, which is that, okay, we're all talking right now with devices near us. So presumably one of these devices is listening and selling the data.

30:57So the next time we all go on YouTube, we're just in this echo chamber where every video is pink elephants going down main street or different cities. And those are really made videos, but now we're seeing it over and over. Oh, I guess really, Maria really did see it. There's so many people who are seeing pink elephants these days. I guess this is a thing now. You start to believe it because of the repetition. Yeah. No, the number of times when there's like a suggestion of who to follow on Twitter in ways that I believe originates from some data trail I leave apart from Twitter, I don't think it's a coincidence.

31:32No, it's not a coincidence. But what device is actually listening to us and selling data? I can't figure that out yet. But I know it's happening. Like I talked to my daughter the other day about, I don't know, how the universe began. And then suddenly on my news feed, it's all these articles. Well, they discovered another way that the universe began. Like, who was listening to me? Yep. No, and I mean, this is paranoia. But like at this point, because we've had so many scandals where companies have said, no, absolutely not. You know, Alexa is not listening if you don't use the wake up word. And then it turns out that, oh, yes, it absolutely fucking is.

32:09and we have these and you have whistleblowers come forward with all these transcripts of a device that was asleep and was never listening, but actually was listening all the time. So yeah, someone is listening. I don't know who. I don't have any smart devices in my house, but people still listen. Well, I have a phone, right? And I have a computer. You're screwed. I'm screwed. That's why I don't want to give any company my DNA. Yeah. That just seems like a bad idea. I don't know. No, I have never gone on 23andMe, Ancestry.com, any of these things. And to our listeners, my PSA is don't do it because you don't want your DNA data out there in these databases.

32:49I mean, how many times has like your data been stolen? So I will not give blood or body samples unless I have to for medical reasons. My DNA is all over the place on every one of these sites. I remember when I first got my DNA results, it said I had the AP40E gene, which is the one where there's a slight chance of early onset Alzheimer's. And so I wrote to the CEO of 23andMe and I said, I'm going to make a cookbook, the AP40E diet. So all the foods that supposedly help cognition and so you don't get Alzheimer's. And she wrote back actually and said, oh, that'll be a bestseller. And I wrote and I asked, well, why do you say so?

33:35And she said, because everyone's going to forget if they bought it and then they'll buy it again. That's a great joke. That was the CEO's response. So now, Nate, you went from poker to election probabilities and you've done extremely well with it. What's the similarities? I feel like there's more, it's a little harder to model. Like poker, there are rules, and there's kind of like a closed world, so you could come up with game theory solutions. But there's a lot more unknown data with elections. I mean, look, in poker, even playing a lifetime of live poker tournaments, you never really reach the long run.

34:14But in elections, it's crazy, right? I mean, you're not going to learn very much from a sample of getting one trial every four years. look we try to turn this open-ended problem into a closed problem by just focusing on the polls more or less making a really good analysis of how the polls in different states are correlated and empirically how polls really are accurate or inaccurate and the answer is they're not always very accurate and so um so you know people are like what factors are you accounting for i'm like oh just the polls right it's a really good way to account for the polls but and there's there is a prior based on the economy.

34:50But it's basically just do a really good job of accounting for the polls and not try to have it be too open-ended because politics does change a lot. But with that said, I mean, you're always thinking about the structure of the model and how elections change. Things are much more polarized now than they once were. That has relevance to how you might design a polling-based forecast. But yeah, I don't know. It's kind of inherently crazy to like... I try not to like stake my reputation on it. It's just weird when you have like one thing that's like 10x more popular than the other things you do, even the other things are better.

35:25It's just kind of weird. When Nate and I, a few years back, first started talking about doing a podcast together, we were kind of figuring out what direction we wanted to go in. And Nate was like, you know, I really just, I don't know, like I'm tired of politics. I think I might want to like stay away from politics. And I was like, Nate, I love you and I respect you and I respect your kind of your mental sanity. However, let me tell you, you're not stepping away from politics, at least for the next year. Well, you got a lot of kind of backlash when people didn't really understand. People didn't really understand the difference between a poll, which is somewhat predictive, and the way you do your models, which is more probabilistic.

36:11so that, okay, in 2016, if there's, and you talk about this in the book, if there's a 70 % chance Hillary's going to win and a 30 % chance Trump's going to win, like you were sort of predicting in 2016, the 30 % is a pretty high probability that someone could win. It's 30%, it's three out of 10 times. Well, I'm even more DJ. I'm like, the market's 15%, right? So we told you to be a long Trump, right? Therefore, it's a great bet. The EV is very positive from the model. So I'm right from an arbitrage point of view. It was great. Someone could have made a lot of money following your prediction models in 2016, probably more than in 2008 when you predicted every single county correctly or whatever election it was.

36:57No, I think it's like the highest ROI, the one that people hate me for is like the highest ROI forecast relative to the markets. Oh, for sure. And as I always say, when these numbers come up and as I wrote in The Biggest Bluff, because I was so like, it was so revelatory to me. Nate's numbers for the election are the exact chance, because it was like 29 or 28%. It's the exact chance of hitting a pair on the flop in No Limit Hold 'em. And for anyone who plays poker, like how often have you hit a pair? Is it like never? Oh my God, I never bake a pair. No, like all the time, you are always hitting pairs because 28, 29, 30 % is a hell of a lot of time.

37:40And it makes it very, I think that's why poker players actually understand probabilities better in a lot of respects because it's visceral. You know that, you know what it feels like. If I tell you that the chance of something is the chance of making a pair on the flop, you get it, right? Because you've played thousands of hands, you understand what that feels like. So it means something to you. Whereas to some random person who doesn't kind of understand probability that much, like 70 gets rounded up to a hundred, right? And even 30 gets rounded down to zero because our brains love to round, like we love those absolutes.

38:18And I even saw, you know, someone was criticizing, Nate, I can't believe this. I'm so sorry to break the news to you. Someone was criticizing you on Twitter. I know that this must come as a huge shock because, you know, normally you don't get any criticism at all. But I saw this back and forth on Nate's Twitter feed about someone being like, I don't want, like, this is, probabilities are bad. Like, forecasts should not be, I'm like, wait, what? Like, what in the world are, you know what I'm talking about, Nate? Well, because people get so biased based on their opinions too. Like, if you're trying to make money, it doesn't matter who's going to win the election.

38:57You want to just make the right bet. And you see people get upset at, I'm sure you get this all the time, Nate, like if you predict, let's say Kamala is going to win, all the Trump people are going to hate you. If you predict Trump's going to win, all the Kamala people are going to hate you because you're not saying what they want to hear. One of my favorite games, sorry, Nate, one of my favorite games is looking at your Twitter feed because different people get mad at you depending on what your data is saying at that exact day. Please, I didn't mean to interrupt you. No, it's just so linear as a function of who is ahead in the forecast and who gets mad.

39:33No one has a good bluff. No one ever bluffs and gets mad at me when Harris, no Democrat gets mad when Harris is 60-40, right? No Republican gets, I don't know. People are very transparent in their motivations. Now, Nate has people like Donald Trump saying he's a very smart guy when he has Trump ahead. And then the Kamala Harris camp loves him when Kamala's ahead. And like I said, it's a very fun parlor game to play. You're like, oh, if I haven't looked at Nate's model yet today, Let me just quickly look at a few Twitter replies. And right away, I know exactly who's ahead in his polls. Well, like you're having this argument, I guess, recently or right now with Alan Lichtman, who he has this model based on who's the incumbent, what's the economy doing.

40:15And it's like a 13-factor model of questions that then says, okay, either the incumbent's going to win or the opponent's going to win. And then he trashed you because your model is different from his. and his model doesn't even apply this year because we had this weird stuff with harris replacing biden but then he then he takes this total like ad hominem attack on you all you don't have a phd so shut up and like what what the heck like doesn't didn't is that what is happening in academia now like if you're not an academic you're not allowed to comment on anything Look, I mean, this is part of the expert class kind of weaponizing their expertise for like part of some reasons.

41:00I mean, look, the Lichtman model, the 13 keys, I could point out some things that I like about it. I mean, for one thing, he is like actually putting himself out there and making predictions. You know, in some ways to have a multi-factor model, you know, I'd rather have 13 keys than like one magic key. I think that's at least directionally right. The problem is like about eight or nine and the 13 keys are quite arbitrary. When I go through, I actually think, using his keys, by the way, he's applied them in the past. I think it actually predicts a Trump win and not a Harris win. But yeah, it's a fun rivalry because you don't have the power to turn the keys, Nate.

41:42I like to create a little mythology around it. Also, again, I think he never has made a model where the presidential candidate shifts mid-election. So it's like throws out the window his whole system. Whereas, like, let me ask you this, how good would a system be different from your system if all I did was take, let's say, the top 15 most used polls and just average them in each state? So that's a simpler version of your model. How different would that be from your actual model? So to just have a polling average is not that hard. What's the Charles Barkley quote? It takes brains to rebound. The rebounding part is actually building the correlation matrix, right?

42:32Figuring out, because if it was just a popular vote, then making election model is relatively straightforward. But what's the correlation of the vote in Michigan and Wisconsin and Michigan and Wisconsin to Arizona and North Carolina and so forth? That's the tricky part that that requires some horse sense and expertise. Even though, let's say I'm just taking all the polls in Michigan, I'm averaging them together. All the polls in Wisconsin, I'm averaging them together. So I don't really need to figure out how they're correlated, or do I? You do because of the Electoral College. So people, you know, some of the reasons there were models in 2016 that had Clinton with a 99 or 99.9 % chance of victory because they treated each state as independent and the errors as independent, When in fact, if Trump beats his polls in Wisconsin, he's very likely to also do so in Michigan, Pennsylvania, and so forth.

43:23And that's exactly what happened. He had an overperformance systematically in the Midwest, and therefore this lead that other models misdescribed as robust for Clinton proved to be fragile. And so now I know you do take into account, for instance, the bias of different polls. So like if Rasmussen reports, it tends to be, you know, overstate the Republican numbers by 3 % in every single poll, you adjust accordingly. From what I understand, am I sort of correct? That's right. I mean, yeah, polls that consistently lean, you can call it a bias or whatever you want to call it, polls that consistently lean toward Republicans or Democrats, the model detects that and adjusts those polls.

44:01But like you were saying earlier, though, we've only had like two other elections with Trump, so it's hard to really know. So like, for instance, right now, we have this election where some Democrats like Tulsi Gabbard or RFK Jr. have endorsed Trump. Maybe that makes it people are more, you know, I think one of the reasons polls are sometimes biased is people are afraid to say who they're going to vote for. And maybe people are less likely to be afraid. Now, you just won't know that until after the election. I mean, look, in 2016, Hillary Clinton won the popular vote by 2.1 or something points and lost the Electoral College in 2020.

44:37Biden won the popular vote by four and a half points and won the electoral college narrowly. Polls this time are saying Harris is going to win by 3.3 points in the popular vote or whatever, and the electoral college is a toss-up. So they're not exactly going out on a limb here. They're making a very safe prediction for the time being. It's very close to being a toss-up right now. And by the way, pollsters are aware of the fact that everyone shat on them after 2016 in 2020. So they've changed their methods, whether they've corrected adequately or overcorrected, I don't know. But they have incentives to get the answer right.

45:29You guys talk a lot about poker, obviously, and a lot about, Nate, in your book, On the Edge, You talk about sports betting a lot. But the one thing I worry about with poker is, and I'm curious, Maria, like you're in there every day playing poker and you've been, you secretly just want to be a professional poker player and nothing else, I know. But poker's a crowded sport now. I mean, you know, back in like 1999, it wasn't as popular. It hadn't yet had the TV phase. It hadn't had Chris Moneymaker bring a surge in popularity. And so maybe it was a little bit more inefficient. where you could win without knowing everything.

46:06But now everybody, like every kid is memorizing all the statistics, everything. Has it become harder? Obviously it has. Like how do you deal with the fact that it's like a hard game now? It's so funny. You're not the first person to ask me this. And also to say like - I'm so unoriginal. I can't. No, no, no, no. Okay, forget it. I'm sorry. And to like, different version is like, you know, you picked a really bad time to get into poker. So my answer is like, I didn't know, you know, poker pre 99. Like, first of all, you know, like I wouldn't have been old enough to play. But even like I didn't start, I didn't even play after the moneymaker boom where I would actually argue based on what I know, it got easier because so many people got into it thinking that they could win.

46:51And so there was like this period, like this golden age of online poker where everyone thought they could be the next Chris Moneymaker. And people are like, but it used to be so much easier. Like you're getting in at such a hard time. It's like, I don't know. Like when I started, it was the dawn of the solver era, right? Like my first year playing poker was the first year that people were using PO solver, like in any sort of a concentrated way. And so this is all I know. Like for me, the game has always been like this. And I love it. Like I actually love the fact that you have to think through the game theory, that you have to think through all of these different things.

47:29And that's so many people, like I actually don't know, like obviously it's gotten harder, but so many people don't know how to use solvers correctly, right? It's giving also this false air of precision where it shouldn't exist. and people think that they know the solution, even though, first of all, poker has not been solved. GTO game theory is optimal. It's not actually game theory optimal because the game hasn't been solved. It all depends on the inputs. It depends on how you build the model. It depends on the assumptions. It depends on so many different things. And because I play live poker, there's like this other host of variables that enter into it.

48:08And if you have a false sense of precision, then it comes often with hubris and with, you know, letting a lot of other things, not realizing what a lot of other things that you need to be paying attention to are. And of course, the best players in the world understand all of this, right? And they're able to use the models correctly and to kind of extrapolate from them and adjust, et cetera, etc but on the whole like sure maybe it's harder than 99 or 2009 i don't know um but like i love all of the challenges that this is um that this is bringing and i don't think we're in any danger of solvers ruining the game if we're in danger of anything ruining the game it's ai and bots and online cheating.

49:02Yeah, online is messy, to say the least. Online is messy, yeah. Yeah, because you can use the solver sitting right next to you and the physical stuff then of being in person, of like the various tells or whatever you use in the physical games, it doesn't come into play. But Nate, you're someone who you started playing poker kind of before this sort of computer solver AI type of era. And then you kind of lost your way for a while to this election stuff. And now you're playing poker more. And have you noticed the difference in your own results and play as a result? No, look, I think on net, it probably helps somebody like me because you can essentially substitute a lot of, I think Maria, you said this before, you can substitute a lot of hours on the felt, on the poker field by studying solvers a little bit more.

50:01You know, on the flip side, like I kind of acknowledge that like, I'm never going to be one of these 20 year olds who like studies Pio solver sims or GeoWizard. What's it called? GTO Wizard. GTO Wizard every day, Geo Wizard. So like, I kind of am becoming more of like a field player, right? I understand what the theory says pretty well and when to deviate from it. I think people in general, I think they fuss a little bit too much about these things that when you like one one hundredth of a big blind or something, where if you can read people well and play exploitively effectively, then that can be very big advantages.

50:43So look, I think it raises the floor a lot. there are clearly people that like are casual and just through osmosis pick up better GTO play, right? There are fewer truly soft spots at the poker table than there once were. I'm not sure it raises the ceiling as much. I mean, I... Yeah. And there was a, I actually listened to a really, I thought, interesting interview a month or so ago with Ike Haxton, who's, I think, one of the best poker players in the world. And he said something that made me go, So he's like, oh, like people, he's like, honestly, bet sizing really doesn't matter that much. He's like, you shouldn't really be worried about like if you're, you know, betting a third pot or a quarter pot or, or less.

51:28He's like, you just need to like, you need to understand like what you're doing, why you're doing. And he's like, I don't think bet sizes. He's like, what are you leaving on, you know, what, what are you leaving if you get the bet sizing a tiny bit wrong? And people will pounce on you and be like, oh, this is such a fish because you're never supposed to bet. you know, 60 % on a board texture like this. Everyone knows that you only bet X percent. And that's just bullshit. And there might be like a EV difference of 0.4 big blinds. You know, it was just funny to hear Ike say that. And I was like, oh, that makes me feel better.

52:05Let's take a specific example from poker. So like, let's say you get your hand and it's before the flop and you're the last one to bet and you make a raise and everyone calls. And now the flop comes out, everybody checks to you and you're supposed to make this so-called continuation bet, like bet out there. And it's considered bad form to not to somehow. Like it's almost like a rule, like you have to do this. But clearly there's exceptions to every rule. So for instance, if you have a decent hand, you could not make the bet and everyone thinks you have a bad hand because you broke the rules. I mean, that's just a simple example.

52:41But it seems like for every rule, anytime someone says you have to do this, that's a perfect opportunity for great players to bend the rules. Absolutely. And what solvers are actually teaching us is that most things are at a mixed frequency, right? Most of the times there's not 100%. And I think that's really important to know. That's really interesting. And some of the strategies that used to be seen as bad, solvers use them. And it turns out they're not bad. So like in the situation that you described, James, if someone actually breaks that and decides to lead into the preflop raiser, right, it's called a donk bet, donk from donkey, because it's bad.

53:23And anyone doing it is a donkey. And if you are someone who doesn't check, doesn't respect the flow on things and bets into the preflop raiser, you must be a donkey. Well, it turns out that that's absolutely not true. And there are absolutely positions and boards where you're supposed to bet. And it's donk bets are actually quite good. And you're supposed to do it a good percentage of the time. And so solvers are showing us strategies that people used to think were really dumb can be actually quite smart. So I think that they're much more interesting than people realize. And it's the same thing now happening on Polymarket, which Nate, you're an advisor to Polymarket.

53:59It's the biggest betting market in the world right now. It's amazing. I mean, You know, a billion dollars is bet on the one question of who's going to win this election. But still, hundreds of thousands or millions of dollars was bet the other day on whether or not Tim Walls was going to say tampon in the debate. So, like, it's fascinating in these betting markets. There's hundreds of possible things to bet on. And because these betting markets are so new, I imagine as opposed to sports betting, for instance, there must be inefficiencies that one could take advantage of. I mean, there are probably inefficiencies in sports betting, too.

54:32It's kind of a barbell distribution where, you know, if you're betting on Kazakhstani ping pong, you can probably have an edge. And the things that everybody bets on, like the Super Bowl, there isn't necessarily enough sharp money to clear out all the dumb money. The in-between the day-to-day NFL regular season game or NBA regular season game is much tougher. Yeah, there are probably edges to be had. Although in politics or in sports, you get banned or restricted if you're any good. So that's the problem, right? Yeah, they're probably... But Polymarket, you don't. No, you don't. Because it's peer-to-peer, ultimately.

55:09So yeah, if you really want to model the likelihood of Tim Walls saying tampon, you probably have an edge there. I think I'm not sure that's best use of your talents, but that's probably a beatable market, I would think. Like right now, I mean, right now, looking at Polymarket this second and looking at your model, what are some decent bets? And I'm not saying people who are listening to this should go out make those bets because this podcast will come out a few hours or days or whatever after we have this interview. So all the odds are going to change and all the bets are going to change. But like looking at polymarket right now, what inefficiencies do you see compared with what your election models are saying?

55:44Look, in principle, right now they have Trump-Harris 50-50 and we have Harris with a 56 % chance, right? So in principle, you'd bet on Harris. But remember, like prediction markets are more flexible and can account for things that like, by model count of count four, right? So for example, right now you have a port strike where American or not American, I guess they are American, where workers, longshoremen are, are striking on the East coast and the Gulf coast, um, which is expected to lower GDP. If the strike continues for more than a week or two by, you know, 0.5 percentage points, you can put that in the model and figure what's the effect of an 0.5 point, um, 50 basis point decline in GDP on the election odds.

56:26And also there's this, you know, war in the Middle East and everything. So I think it's probably pretty reasonable, but in principle, you would bet the Harris side of the line. I do think that because a lot of the prediction markets are crypto adjacent and Trump has been more crypto friendly, there might be like a slight pro-Trump bias just based on the demographics of the people that are making the trade. But yeah, look, for the most part, we're quite close to Polymarket, our forecast at Silver Bulletin. And so it seems like a good thing to me. What about the popular vote? So Polymarket has, it's like 75 cents to do, by the way, so the way Polymarket works is you bet 75 cents, and if you're right, you win 100 cents, and so on.

57:17So right now, Kamala is at 75 cents to win the popular vote. What's your model saying about her winning the popular vote? Yeah, we have her at 77. So in the grand scheme of things, and also, you know, our model, I mean, it's behind a paywall technically, but like the types of people who were trading on Polymarket are probably, you know, paying for the newsletter. We have the popular vote at 77%. So really, we're very tightly aligned with the model, in part because I'm sure the model like looks at our numbers or the markets look at our model and that affects perception of where the numbers should be.

57:50And you've said in the past, and Pennsylvania is the state. And it reminds me of an, there was an Isaac Asimov story like 40 or 50 years ago. I forgot what it's called, but at first everyone was saying, okay, you can forget about the rest of the country. It all boils down to this one state. Then it all boils down to this one city or then this one neighborhood. And he finally had one person that it all boils down to. And so this whole story was about this person who decides the election. And, but given that it's highly likely Pennsylvania or Michigan, say, will be the state to influence the entire election.

58:28And you give the odds if Trump wins Pennsylvania, he has 92 % chance of winning the election. If Kamala wins Pennsylvania, she has 90 % chance of winning the election. And given that, do you think she should have chosen Josh Shapiro to be a vice presidential candidate? Like, should they be looking at your models more to make their big decisions about who should be the second in command of the country? Nate and I have talked about this so much on the pod and we were both - I know, you guys have talked about this on Risky Business, the podcast, which is, I wanted to mention that. We, I think we both, well, I don't want to put words into Nate's mouth, but I was definitely very much Team Shapiro and I think Nate was as well.

59:08And, right, Nate? Yes. Yeah, look, I mean, Walls was kind of mediocre in that debate, if we're being honest. He was nervous, which I understand. You know, I get nervous sometimes in public appearances. But yeah, no, I think for the Pennsylvania factor alone, you know, and the fact that, like, I think she seemed to take it as a negative that, like, Shapiro was seen as more moderate, but for the most part, undecided voters want reassurance that the candidates are more moderate. So I thought that was a mistake. And if Tim Wall's, I think, totally normal VP pick, but if he was going to prove me wrong, then that debate where he was kind of mediocre would have been the opportunity to do it.

59:50Yeah, I think that Shapiro seemed, I think a lot of things that were seen as negatives for Shapiro are actually positives. Because the other thing that I think was a negative was that he was more ambitious, right? He's someone who has kind of more political aspirations. And Walls was more of a quote unquote team player. It's fucking politics. You want the ambitious guy. Like you want the guy with the killer instinct. You want the guy. I think there was a factor that he was Jewish because at that time, particularly convention time, there was probably more anti-Israel protests then than there are now.

1:00:25And that probably – I'm not saying anybody's anti-Israel or anti-Jewish, but all of these things she had to probably take into account. I mean, I think so. Yeah, I don't know. I try to studiously avoid comment on some of this stuff. But yeah, look, I mean, obviously, there are obviously blurred lines between the way certain people's, because he had a mainstream democratic stance on the war in Gaza, right? Although he had, anyway, yeah, it may have been a factor. I don't know. I think on some level it was a factor. And I think that everyone will vehemently deny that it was a factor. Even he denied it, obviously, because he has to.

1:01:07But I will – I think I've said this before, but, like, yes, I absolutely think it was a factor. You can tell he's probably plotting his revenge, though. Well, that's the thing. Like, Pennsylvanians might have been a little pissed off. Like, that might be a factor in what happens. You know that, like, if – I think he's rooting for Harris to win, right? Of course. but if she loses by like eight electoral votes and loses pennsylvania you know he might he might crack open i don't know if he's a drinker or not might crack open a beer and have a little laugh with himself i would think yeah i really hope that doesn't happen but as i think we've talked about like if harris loses pennsylvania and that's kind of the the tipping point of the election i think they will look back at this and be like oh fuck um you know we we messed up but it's too late at that point.

1:01:58But I also think that the debate would have been very different with Shapiro debating Vance, to be perfectly honest. Yeah, because I think Shapiro was a lawyer. It's like two lawyers. They had a similar kind of training. But this is what I'm wondering is, all of this ultimately is about decision-making. And if people look at good models, whether it's poker, a game theory optimal model for poker, or Nate, your models for elections, if I was the presidential candidate, and I'm not, maybe this is why I'm not, I would have looked at your models and said, okay, I need to really focus on Pennsylvania.

1:02:29So I'm going to pick a Pennsylvania to be my VP candidate, and I'm going to put$100 million in ads in Pennsylvania, and I'm probably going to ignore pretty much every other state. Yeah, look, and to be fair, a vice president usually doesn't matter that much. So you're probably only getting like an extra point or so out of Pennsylvania. but I think they overthought it. Look, after Harris replaced Biden, Democrats were so thrilled. It's what I call the New York Knicks phenomenon, right? Where the Knicks were so bad for so long that, I mean, the Knicks are pretty good now, but when they merely made the playoffs a couple of years ago, people were like, oh my God, this is amazing.

1:03:06So they were kind of riding a little high off the fumes of the vibes. And I think they wanted to keep the good vibes going in the short run, have America's hockey dad or football dad or whatever he is and not thinking about like, we would really like to have an extra 50 ,000 votes in Pennsylvania in November. One of the things that we always talk about, Nate, is that it might be just one point, but in an election this close, every single point matters and every single one of those tiny edges matters because they end up adding up. And when you're neck and neck, you need to be thinking about this differently than if you're an overwhelming favorite.

1:03:49Yeah, I agree. And as you guys have pointed out, I mean, there's so many different factors in these elections that who knows really what will be the critical factor, but you've identified a couple of critical factors that people should pay attention to. But again, in terms of like, let's say I want to just make money, it does seem like there's so many bets out there that are, you know, like you were saying, like Kazakhstani ping pong or whatever, like there's that are inefficient, potentially inefficient. That's going to be now my metaphorical, the Kazakhstani ping pong team, if they exist forever now is in my head as like the inefficient market.

1:04:29Nate, do they exist or did you just plant false information into our heads? No, but I mean, during the pandemic, right, people were desperate to have a sweat on things. And it was like, yeah, Filipino ping pong. I mean, literally people People got into really obscure stuff. DGENs will find a way to DGEN in some way. To DGEN. Do either of you bet on Polymarket or any of these prediction markets? Can't in the US. Polymarket is not open to Americans. Really? Yeah. So you have to use predicted. Is predicted still open to Americans? You need to use CalShe. Yeah. CalShe? I don't think I know. You mentioned that one in a podcast, but I've never been there.

1:05:09There's been news since our podcast. So the CalShit is open again, so that their election betting is open. You can bet at Manifold. The other prediction market I like is Manifest, or Manifold rather, which is free. So it's play money, but has a lot of really smart people trading there and a good community around it. So if you're in the U.S., bet play money on Manifold. If you're anywhere else, then bet Polymarket, I say as a paid advisor to Polymarket. And as a non-paid advisor, I will say if you're in the US, bet real money on Calci. Okay. I'm going to check it out. So, okay. Finally, again, we've talked about how everyone kind of secretly wants to be in the river.

1:05:53Nobody wants to be in the village, i.e. the people making dumb decisions as opposed to well-thought-out, rational, risk-based decisions. How does one – there's really two questions. One is how do you recognize whether you're in the village or the river? because Dunning-Kruger, almost everyone I ever speak to about poker says they're the best poker player in the world. And number two is, you know, a lot of people probably, you know, if they're risk-averse, they don't really know how to be, or not necessarily risk-averse, but I don't know, if they're the sort of person who doesn't make these sort of probabilistic expected value bets, it's hard to change your personality to do that.

1:06:36Well, I'll say a few things as someone who apparently is still in the village, but takes forays into the river, which is that learning poker has definitely helped me embrace risk a lot more in all areas of my life because I'm someone for whom it wasn't natural. And the funny thing is I've taken huge risks in personal decisions, right? throughout my career, quitting jobs without having another job or any money, doing things and making very big changes where I have actually taken pretty substantial risks. But when it came down to it, I started off poker being way too cautious. There are usually two different types of poker players at the beginning.

1:07:28And most tend to err on the side of overaggression, which is good. right? If you have to pick one, like be a knit or be too aggressive, it's better to be too aggressive. And I was definitely a knit at the beginning where it was really difficult for me to kind of take gambles, even when I started learning more and figuring out, oh, I should really be bluffing here. I couldn't pull the trigger. I had, you know, there was mental blocks and I think a lot of it comes from socialization, being female. Like there are lots of social factors that play into this. But once I unlocked that, like my poker playing really changed and my decision making in real life really changed.

1:08:10And so I don't think my personality, like who I am hasn't changed, but I've been able to think about decisions differently, which was part of the reason I got into poker. And it worked, right? It actually did what I thought it was going to do and made me much more tolerant of risk of uncertainty of all of these things. And so that, you know, one of the reasons why I think everyone should learn to play poker, but learn to play poker correctly, is for this because I actually think it's really good to be able to think like this in day-to-day life and to be able to kind of have that gear where you can take the gambles you need to take when you need to take them and when you can kind of understand those types of calculations much more easily.

1:08:53And James, I am definitely not the best poker player in the world. I am probably, I would say I'm not the best poker player on this podcast. I don't know. I don't know. It's probably pretty close. I'm higher. I think I'm a lower floor. I don't know. I'm more gambler than Maria, even though she's trying to be more gambler. I'm more gambler than Maria. Yeah. Look, I even people who have read the book, I hear people who are very much in the village who say, okay, I at least am now starting to think about more things in expected value terms. Understanding that life is uncertain and that sometimes the riskier course of action where there's more uncertainty nevertheless has the higher expected payoff.

1:09:36And we all hopefully live 80, 90 years. We get a lot of decisions that we get to make. The more advanced stuff, the game theory stuff, where, okay, what happens if I'm trying to maximize my EV and so is everyone else? I mean, that's the 201 class. But to think in terms of what's the expected outcome and not being inherently afraid of uncertainty when the rewards that way the risks, I think is something that people definitely can learn. Absolutely. Yeah, and I guess so it's a matter of being comfortable with, like you were saying, with risk. And not only with risk, but with risk size. Like even if you understand expected value, you could still go broke not quite understanding risk size, like how much you should bet.

1:10:17And that's very, you have to be a killer also to understand risk size very well? No, bet sizing is an underrated topic in general, which I think does analogize in some ways. I mean, look, having optionality in life is really important. Making choices where you are able to make more choices down the road. Some people are phobic of optionality and phobic of having too many choices available. But being in a position where you can take advantage of opportunities, I think is kind of like the secret behind a lot of people's success. Well, oh, go ahead, Maria.

1:11:20it's almost like given luck and given the fact that people can go broke, it's almost like you have to make sure you diversify as many ways as possible your opportunity to say you have survivorship bias, right? Because all success for all we know is survivorship bias. Like this, oh, if you're the type of person who pursues your passions, you're going to have success in life. Well, we don't know all the people who pursued their passions who are just homeless right now. But if you kind of diversify all the strategies you use so that you can have survivorship bias in at least one of them, you're doing well.

1:11:58And I guess you choose based on which ones have the highest expected value. Yep. Yeah. I mean, if you want a good critique of my book, it probably suffers from survivorship bias in a lot of ways where you're profiling the successful people. Although there are kind of anti-heroes in the book like Sam McManfred. um although his mistake was that he he felt like uh you shouldn't hedge your risks at all and you should go all in on whatever crazy strategy you have to maximize a v and and even it means like literally if it means ruining your life so there's a lesson there as well i think yeah he he he sized that bet too too large yeah i know i mean basically if you if you bet like i there's a simulation in the book about what if you bet an NFL season and ignoring the Kelly criteria, which is the criteria that most gamblers use to make bets and is actually too aggressive, most people think.

1:12:53And basically, 99 times out of 100, you go completely broke. But one time out of 100, you wind up with more money than Elon Musk. So from an expected value standpoint, it seems good. But most people are sane enough to understand that 99 % happens a lot. and outweighs a 1 % chance of being Elon Musk. And SBF willingly did not care or understand that. And I think it's a combination of that plus hubris and overconfidence because a lot of the things that SBF was making bets on, what if in something like Nate's model that he ran out, there's less uncertainty, right? You know that you're actually doing this simulation.

1:13:38But what if your percentages are off? Like what if any of your calculations are off and you're betting on the fate of the world, right? And you're betting on outcomes like that. Then if you're overconfident and your certainties are wrong, if your confidence intervals are wrong, like if anything is off by a little bit, then you're fucked if you're trying to go for that. Like, well, it's actually plus EV. No, it actually might not be plus EV because you don't know that the math is correct. Well, and also you're making decisions in a world – like Sam Bankman-Fried wasn't just making decisions about investments.

1:14:09he was also gambling his life. And he could have incorporated that into his personal model. Oh, the expected value has to include the fact that there's an infinitely bad outcome in some cases, or an almost infinitely bad outcome. And life decisions have to include the factors of life in your model. I mean, I guess to his credit, I assumed he had a billion dollars sitting in a Swiss bank account somewhere or something, which is probably smart. And he didn't. He really was all in on this scheme to steal, although they were paid back, the scheme to steal funds and reroute them to Alameda and make a bunch of degenerate bets on shit coins.

1:14:52He was truly all in on it, which I think probably he probably found himself honorable to, I mean, he said that people who, if you're not willing to ruin your life, he thinks you're a wimp, is what he told me, basically. and there are a lot of negative adjectives you can use for SPF, but I wouldn't call him a wimp. He authentically enjoys degenerate levels of gambling. I'm a wimp. We're probably all wimps. It's probably healthy to be a wimp in that sense. Well, I can't recommend all of your products enough. I mean, definitely Risky Business, the podcast, or is it, again, I have to look up the name for the second time in this podcast.

1:15:32Correct. Risky Business Podcast, which I listen to despite me having to look up the name every time because I get nervous. Risky Business Podcast is a must listen. I'm going to describe it more, or I did describe it more in the intro. And definitely I would read both of your latest books, On the Edge. Nate Silver just came out and The Biggest Bluff by Maria Konnikova. Maria, when's your next book coming out? Um, probably in, I would say like, oh, I don't want my editor to, uh, to hear this conversation and hold me to it, but I'm working on it. I'm working on it. Um, uh, I would, I would say probably early 2026.

1:16:11Okay. And, and you both have, uh, newsletters, the leap by Maria Konnikova, the silver ball Bolton by Nate Silver, highly encourage. I get these as soon as I get these, I stop everything I'm doing and reading them. No joke. Like they're one of the few newsletters I read. And I really appreciate you guys coming on the podcast. Maria's been on many times. Nate, it's a pleasure to meet you for the first time. Thanks again. It's always such a pleasure. Absolutely, James. Great to chat.

1:16:55Thank you.