Welcome to Pablo Torre Finds Out. I am Pablo Torre, and today we're going to find out what this sound is. This will now be interpreted, and has been interpreted, for reasons that again are predictable and understandable, fundamentally as a broad mandate. It's a narrow mandate.
Are you a caffeine person, Nate?
I'm trying not as bad as I used to be, but like, yeah, I mean, I'm in the post-election haze. It takes a little extra energy these days.
It's not fun, really, because...
But you're on this high, right? You're on this kick for like just really grinding and working hard. You have a lot of adrenaline, and you wake up, you're like, I have a million things to do, and there's these new polls, new polls that came in, right? So you're excited, and now I'm like, oh, I can actually like fall back asleep, but I'm in like a hundreds of hours of sleep deficit, and so...
I mean this as a compliment, and truly I do. I was walking by a Spirit Halloween, and I was like, Nate Silver, I think, can relate to the seasonality of this pop-up Halloween store. Of like, at a certain point in the calendar there are lines, and there is stress and competition to get in to buy this costume. You are American politics for a certain amount of time that I think you're still, you're still in the halo, the outer glow of it.
Sounds like you have to just own the fact that it is seasonal. And look, if you're going to do the election, then you're going to have your tourist season, right? The tourist season where you're selling lobster rolls in Maine, and in August there's a lot of demand for lobster rolls, and then Labor Day weekend you're busiest yet, and you're aware that your life's going to change and become more relaxed after that. But I'm, I'm used to the cyclicality of it.
So this is Nate Silver in something resembling repose. Nate is 46 years old now, which means that he's been publishing and defending his statistical analysis of presidential elections for 16 years at this point. But I happen to be old enough to remember when he was publishing and defending statistical analysis of baseball for places like Baseball Prospectus and Sports Illustrated and ESPN.com.
I saw one of my articles, someone had printed it out and had left it in the bathroom stall, and I'm like, I've really made it now, right? Some article about how good Barry Bonds was.
I think the highest compliment, yeah, absolutely, is to be printed out and into a toilet.
But Nate wasn't just a Moneyball stat nerd sort of writer. Nate saw everything even more specifically in terms of odds, in terms of probability, which is why Nate also became a competitive poker player, which we'll discuss. But his rigorous simulations and nuanced descriptions of the likelihood of events and their expected value, all that stuff is not what made his election analysis over at 538 so broadly compelling and so insanely popular. Even more to the point, and that is despite the fact that his probabilistic way of thinking is now almost categorically embraced, by the way, by the richest people on earth, now as you'll see. In fact, I believe that there is a simpler way to describe the business of Nate Silver. I wonder if you object to this characterization, which is that you are fundamentally in the business of being right.
I, I hope I'm in the business of describing probabilistic things. I think I was blessed and cursed in some ways by having his reputation early on. So 2008, you know, 49 of the 50 states we call correctly, and then it's 50 out of 50 in 2012, and like it's just basically random, right? There was like only a 4% chance that would happen. We had Florida in 2012, back before Florida became a super red DeSantis haven. You know, our final forecast had like Obama with a 50.5% chance in Florida, and he wins it. So therefore you get this reputation, but like, yeah, because I want to like describe the uncertainties in the world, and, and that's a harder message to sell when I give a talk or something, right? Like it's literally in my bio, they call 50 out of 50 states, but like it's kind of against my, against my whole method and brand in some ways.
I've noticed, I've noticed the tension in being the guy who was praised for being in the business of being right, and fundamentally, by the way, you are actively and competitively trying to be right.
I am, yeah, but also, but also fighting the framework, the box of, wait a minute, we need to talk about what probability is. In high school I did debate team, which if you have seen me on Twitter you're probably not very surprised by.
I have notes about how you use Twitter, Nate, and all of this is adding up.
But yeah, you want to, you want to prove yourself, and you want to say, okay, let's actually bet on this if you're going to like talk all this BS or whatever else. And like, and like there's just some intrinsic, some intrinsic part of me just wants to compete.
Yeah, so all of which is to say that you reject the premise of your job being to be right, but your most inner desire is to want to be right.
It's to want to prove people wrong, which is slightly different.
Like if someone else is really wrong about something and I think they have a bad process and they're annoying, then like I probably take more pleasure than that from them being wrong. You know, it's like, it's like if you catch somebody's like bluff in poker, that's an incredibly satisfying feeling, right? Catching a bluff in a big pot is more satisfying than making a full house or something and winning a big pot because you didn't like really have, you know, just the cards went your way, right? And there's some skill in how you get the money in and things like that. But like, but calling out bullshit is like my favorite thing to do basically, which makes people annoyed, which makes Twitter the best and truly the worst possible platform to consume you on.
2016 for you was this, it sounds like, this, this moment at which you realized I need to do a better job, or at the very least devote a lot of time to articulating what this way of thinking is and who actually believes in it, because 2016 in terms of Nate is in the business of being right, yeah, was publicly rough. So where does your forecast have the race right now?
So we show about a three-point lead nationally for Clinton, and she's about a two-to-one favorite.
The supporters there at Clinton headquarters listening intently as the results come in. Some of the people look almost crestfallen. As we were coming into the day, you had your project above 70% for Hillary Clinton. Where is it right now?
I mean, I would look at betting markets, which say that Donald Trump is a narrow favorite to win the Electoral College.
CNN can report that Hillary Clinton has called Donald Trump to concede the race. She has called Donald Trump to say that she will not be president, and I'm not sure the exact words, but probably to congratulate a president-elect Donald Trump.
I come from a world where literally before I ever made an election forecast, I played poker and built baseball models, right? So from my kind of gambling mindset, we were long on Trump relative to the consensus. We had Trump with a 29, round it up to 30 if you want, percent chance of winning, and the betting market said 15%, and the other models said as little as 1%. So from my point of view, it's like if you were heavy on like Leicester City or something the year they were 5,000-to-one underdogs on the Premiership, then like if you were like, oh, they're only 100-to-one against, like that would be a good bet. It would be a very positive EV wager, but you know, it's too mainstream, and it's like kind of not how, it's not how most people think, obviously.
Oh, I mean, even the term EV, a plus EV wager, I do want to spell out expected value because this is the key to what 2016 was for you, how you made sense of it, how in fact it, it was a success according to this framework, while also being the opposite to people who don't speak the language of EV.
Yeah, and expected value, I mean, I'm sure most of the audience knows, but this is the average outcome you would get over the long run given the randomness and uncertainty. I mean, an election might not literally be random in the sense that like a game of craps is, but there are things that are unknowable and are easier to say after the fact. And by the way, some things are nearly random. In 2000, when Al Gore lost by 537 votes or something like that in Florida, and there was a butterfly ballot in Palm Beach County where it was kind of misprinted and people, and hanging chads, and like basically random who won the 2000 election. Some events in this election were kind of random. The fact that like an assassin tried to shoot Trump and grazed his ear, and if he had turned differently, then who knows what would have happened, maybe a very dark turn in American history. But the point is everyone acts like the answers are so obvious after the fact, right, when it's, when it's not always clear. I mean, you know, this year because it was so close, like I said, this was easier to explain than, than most years. Expressing things numerically in probabilities is a sign of humility, and people take it as a sign of hubris instead.
Yes, well, I want to get into that because the key thing you're doing, and I want to translate this to truly the people who don't speak this language fluently at all, is you're talking about, okay, if I'm going to make a bet on Trump winning, given the percentage chance that Nate Silver has outlined for me, right, there is an upside and impact multiplied by the probability, the percentage chance that it would happen. And what you're saying is prevailing betting markets, you're getting five-to-one or six-to-one odds to bet on Trump, and we thought the real chance is three-to-one, let's, or two-to-one, right? So 30% of the time you win 600 bucks, and then 70% of the time you lose 100 bucks, and like that's actually a very positive expected value wager.
Now, play poker, and you're thrilled to get the, the money in when you have the right odds because over the long run you're going to win out, right? And like I know I'm kind of unflappable. I have one home game I play here in New York, and I had a hand where I had pocket aces, the opponent had pocket kings, and we actually ran the board twice, right, right, just a way to like hedge your risk, and like, and I somehow lose both boards, like a one-in-25 chance. This is a fairly big pot, there's a lot of money, and you know, this time I was playing this month, the guy's like, you're so unflappable, you weren't like, you weren't bothered by that. I'm like, that's because I'm winning money in the long run in that situation, and I, and I can't control the way that the dealer like shuffled the cards. I mean, maybe, maybe you don't give as much of a tip.
But the reason why people keep on demanding who you got is because they want, they want the answer, they want your prediction, and what you're doing all the time is trying to tell people that's not what I do. Actually, you are, you're trying to explain to people while also giving them, I guess, some sort of satiation about their curiosity that actually what I'm trying to tell you is that the thing my model made is the answer. Here are the odds we've generated, we've run the simulation a zillion times, and what is so incompatible with maybe this is an American thing, I think it's probably just a human thing, is that that feels like somebody trying to have his cake and eat it too.
I mean, it's not. So literally this election, the last time we ran the model at midnight on Election Day, we ran 80,000 simulations, and Harris won, I think, 40,012 simulations. So like a coin actually lands on heads 50.5% of the time, a coin is a little bit biased toward heads, heads is heavier, and our simulation was like more random than a coin flip. So yes, it's true, on the one hand I'm not really saying anything, although it's a guard against, I think, overconfidence on, on both sides. And it's not always that way. I mean, you know, part of what I think made the site very compelling in like 2012 is there was a lot of talk about how Romney is making his comeback and, and Obama's gonna lose.
Who could forget the era Mitt Romney was making a comeback? Yeah.
And like I'm like, no, actually the Electoral College is pretty good, he's a 90% favorite. So like that's more intuitive for people, but sometimes the answer is we just, we don't know, right? We don't know. Polling is very challenging these days. Harris had like better periods and worse periods in the polls. I mean, look, all things considered, this was actually a relatively decent election for, for the pollsters. But like, I mean, you know, people were, if you were not prepared for Trump winning, then you weren't paying attention to. And by the way, before Harris got in the race, I mean, I think Biden would have lost in a, a true landslide. This is kind of a quasi landslide where Trump wins all the swing states, gets a popular vote, impressive, but like not, he's going to win by one and a half points, not a, not a Reagan-style landslide. I think against Biden you would have had devastation up and down the ballot for Democrats, right?
But the interpretation even of the result, right, this is about a one and a half percent margin, this is about, what is it, maybe 250,000 votes total across Michigan, Wisconsin, Pennsylvania, like that's the margin. This will now be interpreted, and has been interpreted, for reasons that again are predictable and understandable, fundamentally as a broad mandate. And to you it's actually, what, what, what is your characterization of it then?
It's a narrow mandate. I mean, winning the popular vote for the first time as a Republican since 2004, it is impressive. And the way in which Trump won, where you have big gains among Latino voters, among Asian-American voters. By the way, technically speaking, I think Trump will get 49.8, 49.9, so he'll be just a hair short of majority. At the same time, if you are going into a room and 49.9% of people are wearing red hats and 48.4 are wearing blue hats and 1.5% are wearing Jill Stein hats or something, right, you, you wouldn't be immediately obvious kind of whether they're more red hats or blue hats in the room. It's a very close election. But I think it's actually in some ways good for Democrats to lose by a tangible amount where they can't just blame the Electoral College and they can't just blame turnout, right, where they actually, I think, have to think about a new way forward for the party because it was a party that was kind of out of ideas.
You know, I do want to broaden this out though to this way of thinking, because when I think about why probability is hard to sell, I think it is in part because people are demanding a take and you are giving them a process. They're demanding a result, you are giving them a process.
For sure, probability is the art of description more than it is prediction, frankly. You're trying to create kind of a three-dimensional probabilistic landscape that's accurate, right? But you're not saying kind of which square of that blueprint that you wind up on in a given election. So this map that we got with Trump winning 312 electoral votes was the single most likely map in our simulations because the outcomes are correlated, that if you beat your polls in one state by two or three points, then you tend to beat them in all the similar states. And so like this is like the most prominent point in the landscape, but it's a pretty flat landscape overall, right? Wide ranging to the horizons of probability, and, and my job is to like describe this reality that if you simulate the world 100,000 times, you'll wind up in. But lo and behold, you know, I'm going to have what, 20 elections in my lifetime if I live to 80, hopefully a little bit more than that. But yeah, I mean, you're not going to actually ever get to like the long run here, and, and so it's kind of this, you have to be very Zen about it because you're never, you're kind of, I mean, you truly are gambling in some sense with your reputation where like you're not going to hit the long run, and so you don't get this thing where you get to play 10,000 poker hands.
Well, I think there's a very big difference between what your reputation will be and how you actually did at generating the best model, right? Like it's, you are, it's two different things. It is, here it is the thing you've wanted, is the thing I made, check out, and that's not actually what you're being graded on. Actually, it's whether the thing happened that made your percentage chance as described look like, oh wow, he was right and definitive the entire time while he was also trying to say this is not about being definitive.
Sports fans are more rational by comparison, believe it or not, because people recognize like the intrinsic randomness of, you know, in baseball you have a line drive perfectly hit, barrelled, right, that just finds the center fielder, right, or things like that, or a gust of wind knocks the field goal. I mean, it's just people, people get that a little bit more.
Sometimes a bird flies in the path of Randy Johnson's fastball.
Absolutely. I mean, it is different existentially, right? I mean, in theory, if you, if you, if you knew every American, were able to like interview them honestly, then I guess you would have some deterministic aspect to it, but we don't have that luxury. Polling, I mean, it's amazing to me that polling isn't worse, actually. Like almost nobody responds to polls, and yet somehow they kind of get sort of, kind of close, and like that to me is, you know, an underrated feat.
But the other reason I wanted Nate in studio with me today is because I finally read the new book that he wrote this year titled On the Edge, which turns out to be an almost ethnographic study of the worldview to which he himself subscribes, and now that that worldview is eating the world, basically, which is also why it's crazy that we do not understand it better. I want to make the point here that sports and music and entertainment and most everything basically is downstream of this particular type of culture, which might be why Nate calls this group The River.
The River is the world of like deeply skilled, analytical, degenerate gamblers, basically, which happens to be a skill set that is very rewarded in Silicon Valley, in Wall Street, and of course, if you are actually kind of a proper gambler, like a poker player.
It's people that are the cross between being very analytical and, the important part, also very, very competitive. And in Nate's ethnography, The River exists in conflict with another group, a group that embodies the old-school institution full of social norms. And Nate, a former employee of The New York Times, calls this The Village.
The Village is the East Coast progressive, educated, expert class establishment. So it's like The New York Times and Harvard when there's a Democrat in the White House, it's the kind of all the nonprofits and all the different branches of the government and things like that, as opposed to The River, which is very, very individualistic to the point of being contrarian.
The Village is all about the collective good, right? It's collectivist as opposed to the radically individualist River. But as much as the individualistic and contrarian River likes to define itself in opposition to the clustered groupthink of The Village, it is worth noting here to me that the most influential real-life street in Silicon Valley is itself an insulated and almost incestuous cluster, a kind of literal village, actually, because what Wall Street is to the stock market, Sand Hill Road is to the ingenious racket known as venture capital, which is dominated, it turns out, by like five or so VC funds. And the game here is not the stock market, it is to very patiently fund an array of bets instead, an array of bets on privately held high-growth startups. And so even though the odds of any given startup becoming a unicorn are extremely low, that individual improbability arrayed this way makes it so that it doesn't even really matter.
The basic Silicon Valley venture capital equation is that if you invest in enough businesses that have a tangible shot to return 1,000 times your investment, then like that's a really positive set of investments you're making over the long run. And on the one hand, the math is great, where like number one, if you are able to be long-term and make enough bets to invest in things that can give you a 100x or 1,000x return, is like a pretty good idea if you're patient enough to see that through. Right. They also though, like they have these very profound like recruitment advantages, where like the best talent anywhere in the world comes to Silicon Valley, and they get kind of the pick of the litter. It's, it's, it's as though in the NBA if the team that won the championship got the number one pick every year instead of the reverse, right?
But what VCs have determined, and it sounds like in your book you did some modeling of your own to sort of establish how advantageous it is to be a venture capitalist right now, who is doing this at the highest level? Because it sounds like early-stage VC, as you describe it, is not even that quantitative. Like it's not even so much a mathematical secret that Silicon Valley has, it is this larger understanding of we make these bets with this long a timeline that we grant ourselves, and it's going to work out. To what kind of advantage there is?
