Nate Silver on AI, Politics, and Power

Nate Silver—I mean, he really needs no introduction. He writes the Silver Bulletin and is the author of On the Edge: The Art of Risking Everything, now available in paperback with a new foreword.

Before we get going, I just want to say thanks, Nate. Writing on the internet is scary, and you’ve made it less scary. I think the sort of like getting to the point where I can just say things that I know are going to piss off, you know, administration officials and five trillion dollar companies takes a lot. Watching you do that over the years has been a good sort of lodestar for just not caring what powerful and rich people think and just kind of working towards the truth.

I appreciate it, Jordan. I haven’t met you before—we just learned that we’re like about a seven-minute walk away from one another in New York. But I think the equilibrium is just like people in media are way too short-term focused and way too susceptible to peer pressure.

Kind of conversely, once you develop a reputation for:

Doing your reporting,
Doing your reading,
Doing your thinking,
Speaking from a place of knowledge and experience,

but just not trying to sanitize things too much, I think you develop trust with your audience. You carve out more of a distinct niche or brand—if you want to use that overused term.

Yeah, I just think people are too afraid of honesty and differentiation. It’s easy to say if you cover fields like electoral politics or sports that are popular and get a lot of audience and traffic—that much is true. I’m not doing investigative reporting here, but I do think that working hard, being the best version of yourself, and being an honest version of yourself is usually a smart strategy in the long run.

Yeah, I mean it’s something I think is even more difficult with policy writing. I’m closer to a think tank than I am a journalist. The vast majority of people who work in this field have their public comments either off the table because they work in the government or work in government relations for a big company. Even if you are at a think tank, you have to pay the bills somehow, and that basically means getting corporate sponsorship of your work. Which kind of, you know, in some insidious ways, for better or for worse, ends up tracking how you can talk about different stuff.

So yeah, no—look. I mean, I find as a consumer, there are lots of issues including China. I’ve been to Hong Kong three times, never really been to mainland China except the Beijing airport once. But like, you know, China is among many issues where “gosh, I’m not quite sure who to believe or who to trust.”

And I feel like I’d have to invest a lot of time in investigating who I can trust. That requires, at that point, you could almost write about it yourself. And there are a lot of issues like that. Or it’s like, “gosh, I’m not sure there’s any kind of trustworthy authority.”

I mean, kowtowing to corporate power is part of it. That’s part of the beauty of having a Substack model where there’s no advertising. But also, especially in diplomatic international relations, people are always calibrating what they say.

So a lot of the time, I guess I won’t give an example and pick a fight, but the people who I think are smart commentators, you’re always saying,

“Okay, if they’re saying this, then what they really think must be that,”

right? And 20 degrees to the left, 20 degrees to the right—and then it’s just a lot more work than longer formats where you’re able to communicate with more nuance, honesty, and subtlety.

I mean, there’s this recent micro-scandal—Robert O’Brien, former national security advisor, wrote an op-ed saying

“we should sell lots of Nvidia chips to China.”

It comes out three days later that Nvidia is a client of his. That’s the sort of thing where you can get away with it if it’s in an 800-word piece. You can skate over it in that context.

But once you get to the 3,000-word version of it, it’s less “oh, here’s a former national security advisor saying X,” and more like, “Are the arguments convincing?”

And I don’t know, it’s tough. I’m never quite sure whether to say,

“Okay, take everything tabula rasa and ignore who’s making the argument, just assess the argument.”

The fact that people have long-term credibility and reputational issues—one of the things I do is play poker. In poker, the same action from a different player can be massively different things, and a lot of stuff is subtextual. A lot of stuff is deliberately ambiguous. One reason why I… Think large language models like ChatGPT and Claude are interesting because they understand that you can kind of mathematize language. In some ways, language is a game in the sense of game theory, and then it’s strategic. What we say, exactly how we say it, and what’s left unsaid is often powerful. A single word choice can matter a lot.

I mean, you probably think about it in a different context—like official statements by the Chinese government or whatever else. But yeah, no, let’s stay on this though because this was actually one of my mega brain takes. I kind of want your response to it.

We have Nate Silver at 11 years old wanting to be president, and now we’ve had 20 years of him thinking about analyzing presidents and presidential candidates and what they do. You don’t necessarily have to believe in AGI and fast takeoff to think that, say 10, 20, 30 years down the road, a lot of the decisions that presidents and executives take today—an AI will just strictly dominate what the human would do.

That may be the Cerberus moment of what the Swedish prime minister was saying a few days ago:

“Oh yeah, I ask ChatGPT all the time for advice. Just becomes like, no, you should just listen to the AI.”

So I don’t know where you feel free to take that wherever you want, but I’m curious, sort of maybe to start with, or feel free to take that wherever you want, if you want a place to start:

What parts of the things that presidents and presidential candidates do
Do you think are going to be automated the fastest

Presuming that we just let them ingest all of the data that your president or executive would be able to consume themselves.

I mean, look, AI in its current form might be an improvement over a lot of our government elected officials, but that says a lot more about the officials than the AI necessarily.

Yeah, look, I don’t take for granted— and some people do, including people who know a lot about the subject— that we’re going to achieve superhuman general intelligence. There are different valences between these terms that we can parse, if you want.

But you know, some of the reasons that AIs are large language models (I should say) are good now is because:

a) they train on human data, and 
b) they get reinforcement learning with human feedback.

There are cases like pure math problems where you can extrapolate out from the training set in logical ways. For more subjective things like political statements, I don’t know as much.

Some people believe that AIs could become super persuasive. I’m skeptical. First of all, I think humans will be skeptical of AI-generated output, although maybe I’m more skeptical than the average person might be.

Also, it’s a dynamic equilibrium. You can message test, have an AI train an AI, and it can figure out:

“Okay, we can now predict which ads or fundraising emails will get a higher response rate.”

But when people start seeing the same email—Nancy Pelosi says this or that 50 or 100 times—they adjust and react and kind of backlash to it.

In general, in domains like poker, where you are approaching some equilibrium, profits are not easy to have. There aren’t easy tricks; you have to play a robust, smart strategy ultimately. The strength of your AI matters, and how precisely you construct the mix of strategies you take at any given time.

So look, I think there’s a:

25% chance on relatively short timelines AI just blows our socks off
25% chance it does that but in a longer timeline—decades out
50% chance AI is a very important technology, more important than the internet or the automobile, reshapes things, but does not fundamentally reshape human dynamics across a broad range of fields

Things like international relations or politics are, I think, some of the more resistant domains toward AI solutions, quote unquote.

At the same time, I do think another risk is that you’ll have people who view AIs as oracular. We’ve seen cases of people encouraged by ChatGPT to think they’ve developed some new scientific theorem or discovered a new law of physics. They’re very smart at flattering you.

One of the things I do is I build models, and sometimes a bad model is worse than no model, or you’re kind of implicitly… Mental model and so like trusting an all-knowing and all-powerful algorithm, especially in the case where the situation is dynamic, right? The laws of mathematics don’t change but like international relations and politics are always dynamic and maybe changing faster. Whether the AIs can adapt to new situations quickly is also an open question.

Yeah, I mean if we’re trying to bucket the types of things a CEO, a president, or a senator does, I mean we have some sort of personnel management like:

Who am I going to hire?
Who am I going to fire?

We have the outward-facing stuff of like:

What do I say to this interviewer?
How do I talk in the debate?

And then we have these decision points where you have a memo and you could pick A, B, or C, and there are different sets of trade-offs where you could optimize for this thing or for that thing.

I don’t think it’s crazy to think that parts of those different buckets of things you need to do could have AI radically improve—even just to play out the different second and third order effects of whatever you’re negotiating, like the next budget bill or something.

For discrete tasks, yeah, AI can already be wonderful. Right now, I’m doing a little coding on a National Football League (NFL) model. It’s late at night, I’ve been up a long time, had some wine at dinner, and I’m like,

“Okay Claude, how do you do this thing in the language I’m programming in?”

The thing is a discrete task where I have enough experience with these models to expect a good answer, and enough domain knowledge where I can tell if it works or not. I’m not going to have some bad procedure that then chains into other bad procedures in a complicated model.

Yeah, knowing… there’s been mixed evidence on how much more productive AI actually makes people. My stylized impression is that:

It makes the best people even more productive.
It makes the people who are not that smart maybe worse.

And yeah, I worry that it’ll be a substitute for domain knowledge and human experience. But for certain things, it’s already superhuman, and for other things, it’s dumb as shit.

Knowing what is what—I think there is a learning curve for that.

Yeah, I mean, I think there’s this aspect where the human floor can be pretty low, especially if you’re tired or stressed or you’re like the president with a hundred thousand things being thrown in your face. It is literally an impossible job.

So I just think it’s coming. I don’t know, maybe there’s also this weird electoral feedback thing where, presuming AI is really helpful to winning and governance, the folks who trust it more and faster are the ones who perform better in their jobs and get elected to higher and higher office.

No government is perfect; many are backward in a lot of ways. This is very country-by-country too. In some ways, it’s kind of like America has maintained a relatively high degree of international hegemony despite having this constitutional system that’s now hundreds of years old and has flaws.

People who sound a little partisan might say that Donald Trump found a way to exploit some of those flaws. It’s kind of amazing that we still entrust all of this power in one president.

I think probably New York City with eight million people is about as large an entity as you should have maybe one person in charge of, but we haven’t really developed other systems.

If anything, maybe look at the more dynamic places in the world right now. You’d still say:

The U.S.,
China,
And then probably the Middle East.

They kind of cheated. The U.S. is increasingly less democratic, and the other two were not really democratic.

Maybe my assumptions have been wrong.

Taking a step back, in covering the way you’ve covered politics, has it shaped your take or breakdown on a great man versus structuralist view of how things unfold?

I mean the kind of empirical default I think is to be more structuralist. I don’t know. I couldn’t think that Donald Trump is a very important figure. If Donald Trump had… I think Donald Trump is kind of funny. Right, if he had been a stand-up comedian instead, I think the world would look a lot different. I think Elon Musk has been very important to the shape of the world. I’m not saying positively or negatively; I think he is very important.

And, like, no look, I — and things are trending more that way, right? I mean, people, you know. I’m not some like Elizabeth Warren type, but, you know, if the top richest people in the world basically double their wealth every decade, right? And it’s different people cycle in and out of the list, but, like, if you look at the inflation-adjusted wealth of the richest 10 to 100 people, it’s often more about the top 10 now than the 100. They basically double their wealth inflation-adjusted every 10 years. And you compound that over several cycles, and, like, yeah, I mean — that concentration of power, I think, might shift things more toward a great man theory.

I’m sure there’s a substantial degree of randomness too. I have no idea how China formed, for example, but, like, you get in these feedback loops where you have a virtuous loop or a less virtuous loop.

Like, 30 years ago, Americans were worried about being overtaken by Japan, and it happened to Japan a few times. And, like, you know, in some ways, Japan is a very amazing advanced society. So maybe a better example is Rome, Italy, which is like one of my favorite places to travel to. I’ve been there at different points in my life.

There are parts of Rome where if you go today in 2025, they don’t really look that much different than when I was a college kid in 1999–2000. Except for the cell phones, you could really be placed back in 2000 and wouldn’t miss a beat.

Slight non sequitur — my favorite dad book is called Italian Education. It’s a memoir of a cranky British person who married an Italian woman. It’s about raising their kid in sort of early 1990s Italy. The book features really fun, kind of new journalism style writing, but it’s also a fascinating portrayal of Italy at a big transition moment where they’re:

going from super Catholic to more modern
shifting from very localized identities to conceiving of themselves as part of a European project

I don’t have great 2025 Italy takes, but it’s interesting just to think about how much further or not further a country could have gone from that moment to today, especially looking at different parts of the world from 1994 to the present.

I have a friend who’s Irish — like, actually Irish, not Irish American. He moved here in early adulthood and is gay. He said:

“I was growing up, like, Ireland was very Catholic and very anti-gay, and now they almost go out of their way to be pro-LGBTQ rights.”

It is interesting, you know what I mean? Conversely, maybe Russia and that sphere are separating a bit more.

I think the United States is diverging more from Europe too where Europe hasn’t grown a lot economically, and there hasn’t been a lot of innovation. At the same time, their lifespans are increasing and they’re kind of taking this slower-growth dividend into quality of life.

Whereas in the U.S., male life expectancies, even if you ignore COVID, haven’t basically increased in a decade. We are getting wealthier, but I worry a little bit that, as we’re doing things that undermine American leadership and state capacity, we are throwing away some advantages.

We’ve been kind of playing the game on easy mode because:

The dollar is the world’s reserve currency
The U.S. has the world’s biggest military
The U.S. is, relatively speaking, a low-bar, relatively trustworthy player in international relations

If we’re throwing those things away, then it might have a big impact in the next year or so.

When you visit different countries and extrapolate out 3% GDP growth versus 1% GDP growth, and compound that over 20 years, it’s extraordinarily powerful. You see it on the ground — some places are stagnant while others are growing.

I don’t have a good transition talking about prediction markets, but this kind of ties into what the AIs can and can’t do.

You spent a lot of time thinking about poker and had this whole experiment in your book where you spent a year betting on basketball.

The thing about basketball and poker is there’s a lot of data and a lot of track record, which you can base your estimation on. You have to find your edge in very weird corners, but a lot of these markets on… Polymarket and Kalshi are very one-off, like right now is Trump going to put more sanctions on Putin in the next six weeks? Like, is there a regression you can run on that? Not really.

I don’t know, it’s just fascinating because these markets, they’re so much more kind of one-off and open-ended than what you would see in the stock market or in sports betting. Look, I think — and I’m a consultant for Polymarket, so I do have a conflict of interest to disclose — yeah, I think sometimes the one-off events are not as good.

One market where I thought Polymarket, Kalshi, whatever else did not do as well was the election of the new Pope, where pre-bost had like a very Pope Leo now had a very low chance of being selected.

What happens there? On one hand, you have a papal election every 10 or 20 years; that’s not like you have a lot of data. You have everybody leaks, but the papal conclave does not leak apparently, so there’s not really any inside information. Then people apply the heuristics they might apply to other things, like when they have the smoke signals come out quickly, they’re like,

“Oh, it must be kind of like the most obvious name.”

So the favorites went up — that wasn’t true. People had no idea what was going on. So I think there are some limitations there versus things like elections, which are more regular, regularizable.

At the same time, there is a skill in estimation, and it’s one that poker players and sports bettors have. Maybe I’m making a real-time bet on an NFL game, and Patrick Mahomes gets injured, the quarterback of the Kansas City Chiefs, and I have to estimate what effect this has on the probability of the Chiefs winning.

If you just do that a whole bunch, then you get better at it. You have to have domain knowledge; you have to be smart in different ways. And when I have consulted in the business world — not like kind of capital C consulting but people who are actually making bets — you realize that:

A good answer quickly
An okay answer quickly

often makes you money, whereas a perfect answer slowly doesn’t.

I think this is one reason we’re learning that we’ve seen a shift of power away from academia toward for-profit corporations. You can say that’s bad or good; I think we need both, frankly. But you have a profit motive, and you have an incentive to answer a question quickly.

In poker, the same thing applies. If I play a hand and I’m getting two to one from the pot, I need to have the best hand or make my draw one third of the time. Then you go back and run the numbers through a computer solver — they’re called solvers — and:

Actually, here I only had 31 percent equity when I needed 33 percent, so that was a big blunder.  

For most people, 31 versus 33 is the same. But with training, you can estimate these things with uncanny precision. There’s a lot of implicit learning that goes on, and it kind of becomes second nature.

Are you worried about insider trading with all this political betting?

There’s an aspect of: look, these are all crypto markets. You get on these markets with crypto, and there were markets like,

Which way is Suzanne Collins going to vote?
The tail outcomes for a legislative assistant in her office

You can make 10 times your salary in a minute.

What’s your thinking on this?

For sure, I think there are a couple of qualifications:

People on the inside often aren’t as well informed as they think, or
There are downsides to having an inside view and not an outside view; you might drink the Kool-Aid, so to speak, or be in a bubble.

We’ve seen, however, a lot of group chats where people talk about very, very sketchy trades and one-way bets being made in the stock market about what’s going to happen with a trade deal.

You can literally be the person who decides and be betting on the side of it.

If there are incentives to make money in a world of 8 billion people, many of whom are very competitive and most of whom have access to the internet, people are going to find a way to do it.

It’s not just that whatever game theory equilibrium is a prediction of what occurs in the ideal world; it’s what very much does happen. We’ve seen things like in the… In the crypto space, we’ve seen an increasing number of crypto kidnappings. Well, I mean, that’s one of the consequences if people are worth vast amounts of wealth that isn’t very secure. It’s just going to happen until you up security or have better solutions or whatever else.

I don’t think there’s necessarily any more or less insider trading on Polymarket than there might be for sports betting sites. We’ve seen a lot of sports betting scandals or for regular equities. I believe the literature says that members of Congress achieve abnormal returns from their stock portfolios; I’d have to double-check that. I’m sure there’s some debate about it.

People can also sometimes misread insider signals or read a false or fake signal. For example, if they see unusual betting patterns, they might think, “Oh, okay.” There were some tennis betting scandals where tennis is an easy sport to throw because it’s individual—just two people—so you don’t need multiple conspirators, so to speak. It’s something unusual to conclude:

“Oh, therefore there must be an insider trading move or someone’s throwing the match.”

Sometimes it is, other times it isn’t, and it’s very hard a priori to know which is which.

I guess it’s just a new variable, like in politics, where the way you could cash out was:

Becoming a spy for another country (very high risk, lots of downside)
Having a career and then becoming a lobbyist (pays out over years)

But this is something new, and we’re going to have to watch it because I find a lot of value out of seeing these numbers every day and watching how they change. Polymarket, at this point, is like sometimes I go there before looking at the homepages of major news outlets.

However, there’s something that makes me a little queasy about opening up this new realm of betting where perhaps, as citizens, we don’t want the people who we’re paying to do these jobs to have this alternate way to cash in. This applies for journalists too. I know a couple of projects where people are trying to apply a journalistic skill set to make trades—not necessarily in prediction markets, maybe a little bit—but more just equities.

If you have a well-informed view of China, especially in particular industries, then that has big implications for how you might trade a variety of stocks, including American stocks.

My impression is generally that:

People at Wall Street firms trading equities don’t like all this macro risk.
They like to predict what the Fed is going to do.
They like earnings reports, long-term trends they can run regressions on.
They don’t like profound political uncertainty where macro bets are very long-running.
It may be hard to come up with the right proxy for the trade you want to make.

I think you’ll probably see more of a fusion between:

Trading
Research
Journalism

…and kind of every step along that path, potentially.

What do you think your legacy is now, and how would you want to change that in the next 10-20 years of doing your thing?

I mean, look, I think the best work I’ve done is the book I wrote a year ago, The Edge, now in paperback. I think election models are valuable. I think they might be what I’m best known for even though it feels like it’s been like five or ten percent of my lifetime productive work and things like that.

It’s a really hard question to answer; I don’t think I’m quite old enough to answer that yet.

I can answer it maybe I was. Yeah, sure. What do you respond to mine then?

Look, I think there weren’t a lot of numbers in a lot of discussions before you. With elections, people have to put their money with their — I mean, not really put their money with their mouth — but at least respond to facts. You boiled all the polls and demographic and voter registration data into a tangible, grounded set of facts, rather than:

“Oh, you say Marist, I say Quinnipiac. What does that even mean?”

That kind of insight is now spreading into wider arenas, on Polymarket and Kalshi, where lots of different modes of politics now have numbers attached to them in ways they didn’t before. There’s literally a Will China invade Taiwan market like you could not get a more direct example. However much we want to believe anyone has any insight into that, there was no number you could point to to sort of like ground you in that. So, I mean, a great, great man versus trends. There’s been a lot of — there’s a lot more data, a lot more computing power that’s happened over the past 25 years to like enable you to do your thing.

But, I mean, I think it was both doing the modeling as well as presenting it in a voice which is like compelling and engaging that really helped it reshape the way people think about it. I appreciate, you know, I think I’m like, you know, look, I know I have certain talents, right? But I think it’s kind of like if you’re a seven out of ten on the modeling and a seven out of ten or maybe eight out of ten, right, on the presentation, that overlap I think is somewhat rare.

And the overlap is maybe more than the sum of the merely pretty good. I think actually, I think I’m a good modeler, I think I’m pretty good at modeling, but the combination of those is valuable, I think. And I agree that, like, you know, the world is moving directionally more toward prediction markets in particular.

You know, if you look at sports betting, it’s not really growing as much as like the big industry players had hoped, but like, you know, look, prediction markets have had some false starts before. I think now with Polymarket, Calci, Manifold, and the others, right, you have a robust enough and a well-constructed enough trading ecosystem where they are here to stay for a lot of things.

Then, like if I am about to publish a story — let’s say publishing a story on Trump’s latest round of tariffs — and I have to change the tariff if Trump, the article on Trump, has done anything crazy in the past five minutes since I last read the internet, right, I’ll just go to Polymarket.

Look, today did any markets radically change here where maybe there’s some massive event that would make it insensitive to publish a story? If there’s been some earthquake somewhere or something, you know, you just kind of instantly see that news. I mean, Twitter used to be somewhat like that for instant feedback.

Yeah, but it kind of also leads to this gamified ecosystem where, like as a poker player, sports bettor, or for that matter as someone who was a rapid news consumer, you’re always kind of:

Always checking your email
Always checking your phone
Always checking Twitter
Always checking the internet

And you’re kind of always aware of like 15 things at once. It makes it hard to unplug. It makes a distinction between like what is my work life and my real life — where are those boundaries? I mean, I don’t know.

If I’m running on the east side and listening to a podcast, right, like I’m thinking about the next article I’m gonna write, like that’s work. If I’m checking my phone late at night when I’m out at dinner, that’s kind of work. If I’m checking work stuff and not gossip or whatever. And the world’s moving more that way, I think, for better or for worse, and I think it does prioritize quick computation and estimation.

I don’t, the last point tying that to Nate Silver and the legacy — like it’s harder to do big things anymore or what is it?

No, I think it kind of creates more of a barbell-shaped distribution, right? We’re like working on the book, which took me three years — like that was a really important fundamental project. Right now, I’m working on this National Football League model, like I said, that’s a six-week project on a three-year project, but like that’s kind of foundational work that will produce dozens of blog posts and hopefully hundreds or thousands of subscribers for many years for Silver Bulletin.

I don’t know. Having three or four things that you’re really interested in and you’re fully invested in — I think is obsessions, you might call them — I think that’s very valuable.

But also the skill of like quick reaction, your best flash five-second estimate at your best flash in a newsroom setting, right? You know, having the first authoritative take on a Substack.

Right, when Zoran Mandani won the Merrill primary by a larger margin than predicted, I was in Las Vegas at the World Series of Poker. I just busted out of a tournament and I hadn’t really seen any great takes on this yet. It’s a very fruitful subject.

And it was, you know, midnight New York time, nine Vegas time. I just cranked through until 3 in the morning Vegas time, 6 in the morning New York time, and published what I thought was a pretty smart story on it. And so that kind of thing is important.

Whereas the middle ground, the ground occupied by like magazines for example — right, not that places like The Atlantic have become… Digital brands don’t do great work, right? But like, that’s maybe the in-between where the turnaround is a little bit too slow to be the lead story in a very rapidly moving news cycle. Maybe it’s not quite foundational work either.

I mean, I think academia suffers from even more of this problem, where the turnaround time to publish a paper is just too slow. You know what I mean? Versus me, I’ll run a couple of regressions and then give it a good headline and make some pretty charts.

“It’ll be 90% as good as the academic paper in terms of the substantive work and like 150% better written because I’m writing for a popular audience and not journal editors.”

And, you know, I think that’s kind of that exchange of ideas that’s what moves the world. It’s fascinating to see these dynamics.

I mean, I don’t know a lot about Deep Seek, for example, but it was interesting to see the kind of narrative shape in real time about just how China is competitive in the short to medium term on large language models. That was very interesting.

Or to see with Zoran winning the primary, probably the general election too, we’ll see, how much that’s anchored the conversation in different ways and how like stylized, abstracted, modelized versions of the real world become more dominant. We’re all model-building, right?

Even a friend of mine who’s a computational neuroscientist at University of Chicago said,

“Ultimately the brain is a predictive mechanism.”

When you are driving or looking at the road ahead,

“It’s not quite like you’re seeing a literal real-time version of the landscape in front of you.”

It’s more of a stylized version where your brain is making assumptions and filling in blanks because it’s more efficient processing. The message length can be shorter if you focus on the most important things.

You can notice this if you’re in any type of altered state—whether drug-induced, under anesthesia, tired, or under extreme stress or fright—where you process things differently.

On a broader societal level, I think that’s how it operates too. People are like,

Using shortcuts to get things done because otherwise, they never accomplish anything.
But taking shortcuts leads to blind spots.

I don’t think that’s really solvable. AI might scrape off some of the rough edges a little bit, but sometimes the rough edges are created by the market being efficient and dynamic. People key off others’ predictions and form a rapidly shifting consensus. So those dynamics will remain very interesting.

Yeah, it was sort of the Deep Seek experience that was surreal for me because this is a story on China and AI that I’ve been following for five or seven years, and all of a sudden, it is the story.

I think our team hit it pretty well—we doubled our subscribers, great—but watching our little thing try to shape the broader narrative and suddenly all these journalists are like,

“What’s this company?”

And I’m like,

“I’ve been writing about it for two years. Where have you guys been?”

I’ve never had one of my stories really become the main story. It can be an amazing feeling, by the way, where I mean, you talk about the great man theory, this is kind of an analog to it.

Early in a news cycle the way a story is covered is very important. One news outlet or journalist covering a story a different way can shape attitudes about it for weeks to come.

It’s partly why PR people are always like,

“Don’t say so much but be fast.”

You want to preempt things because that founder effect can matter.

But you tell me, Jordan, so what did the mainstream media get wrong about the Deep Seek story if you’re reading The New York Times, The Wall Street Journal, or whatever else?

Yeah, so there was this first narrative that it cost them six million dollars to train their model. That was illustrative. I went on Casey Newton and Hard Fork, and that’s the first question they asked me.

And I’d already written a few things, but I was like,

“No, that is not it. It did not cost six million dollars to make this model.”

You need to hire the people, you need to run the compute… You need to have the compute you need to run a lot of experiments. Like, by the way, actually all-in is probably more like half a billion dollars.

There were enough people for whom that narrative was interesting or financially useful for it to be a thing that spun faster than, like, I don’t know, my loyal crowd of China talk listeners who were listening, hearing us kind of beat the other side of the drum.

I think another thing which has really been shaped by that is the export control debate, where there was this expectation that because DeepSeek exists, that means export controls are worthless. There’s a lot more nuance into that, which we’ve covered in lots of other podcasts we don’t necessarily need to get into here. But the simplified version of drawing that line from A to B, I think, has really shaped the trajectory of American policy towards AI export controls, AI diffusion more broadly.

I feel like myself and the cohort of other folks who understood this sort of money thing weren’t like—there was, like, we have now, quote unquote, lost when it comes to semiconductor export controls. Partially because there was this moment that really ended up reshaping narratives, where the people who agree with my version of the facts were not able to seep into the halls of decision-making.

These narratives that prevail are often, like you said, in someone’s economic or political interest. For example, after the 2024 election, there was a narrative that Democrats lost because of low turnout, especially among younger voters. There is a slight grain of truth in that. However, this was exacerbated because people don’t realize that it takes a month to count the vote fully in the United States. The counts you see on election night, the next day, and the day after are going to shortchange turnout by tens of millions of votes.

But that was a convenient narrative because people wanted to move to the left. It’s not entirely that turnout was low; it’s more that younger voters, particularly young men, shifted against Democrats, and there’s somewhat lower turnout. But it’s still kind of high historically.

It is tricky when you have the more nuanced take versus the easily memeable take. If you’re making sports bets or something, a lot of it’s in the nuance. For example:

We all know this quarterback is good.
Maybe he can both be very good and a little bit overrated by conventional wisdom for X, Y, and Z reasons.
There’s enough of a profit margin for a positive expected value on your bet.

In the news cycle, that’s less forgiving. But I do think Substack and podcasts give a little bit more room for subtlety and exploring things.

I’ve shied away from writing a book all these years. You are very quick twitch but have also done it and made the case that it’s really valuable.

Why don’t you expand on that for me? What do you think having these two book projects under your belt has given you?

For one thing, I like the process of writing a book. Ordinarily, day to day, I guess I consider myself a journalist. But for the most part, the process involves me in a computer. Particularly when it comes to politics, I don’t really want to have—I don’t really want to call a Democrat or Republican source and get their take. People are paying me for my take, and I don’t care to be spun. That increases the turnaround time.

For the book, it’s the opposite. My book, On the Edge, involved interviews with roughly 200 people. A lot of people who are experts or practitioners in lots of fields I find fascinating.

If you weren’t working on a book, if you took two years to interview 200 really smart people about things you’re knowledgeable about just for your own edification, that would make you a lot smarter.

When I was reading the book, this was what I was annoyed by: I wanted the Nate Silver. I wanted to listen to the hour-long conversation you had with Peter Thiel. I think that’d be fascinating if it’s on the record. Anyways, why not sequence it that way?

I think the implicit—I thought people that I spoke with were often very candid, maybe against kind of narrow self-interest.

In writing the book, in some cases, I would never approach somebody and say, “Oh, I want to talk to you about X.” I would say to them, “I’m writing a book about X; it’ll be published in a year or two.”

I would always be very honest about the rationale or the pretense for the conversation. Right, and I feel like it would be, like, I think if there were something Peter Thiel had said that was like super newsworthy—and I’m trying to remember—that was a conversation where they were more explicit. Let’s say it was just like a conversation without ground rules where it’s like, “Yeah, I’ll talk to you, I understand it’s for a book,” and someone says something that’s newsworthy in a non-book context.

I don’t think it’s like over some bright line journalistically, but it’s like a little bit ethically fraught. One thing people say, like with Sam McManus-Freed, that was an explicit understanding. He definitely told me things because he thought the timeline would be enough to insulate him from some risk, or he could shape the narrative somehow. It wound up coming out after he was already in sentence and in prison.

Those people—some, you know, you’ll have a reporter who embargoes some reporting like on a political project. There should be lots of books about Biden’s decline, and the fact is that people will tell you more when they have more protections.

Right, so probably 80 percent of my interviews were more or less fully on the record. Sometimes the background interviews or the author record —and journalists have a lot of distinction for these terms—are in between categories where it’s like you can publish this with approval. I don’t love that, but I did it for one or two important sources where I thought it was the least bad option.

People are more candid if they understand the context of your project and believe it’s coming out in the context of a book that puts everything into a broader universe.

I am reluctant to, like, you know, look: I used to work at The New York Times, and you know, ended in 2013, but now I write for them a few times a year still, so it’s friendly.

If the New York Times calls me for a story, I’m still sometimes reluctant to say anything because you’re going to have one or two quotes put into their narrative that may or may not suit your selfish purposes and may or may not be accurate too.

I mean, The Times is a very popular paper in part because they do write in narratives. Even a boring quote-unquote story, unless it’s like economic data, is now also an interesting case. We’re like, but everything has a little bit of spin on the ball—good writing, a good headline, never anything wrong with those things.

But sometimes there’s more of a narrative that is a little bit reductionist. And they are the best in the business, or among the best.

When you get people who don’t have those journalistic standards then you’ll encounter more problems potentially.

We have these nice historical interludes in the book. Is that a type of, you know, like sort of non-journalism driven writing? Is that something that’s potentially on your horizon as well? Or how do you think about, you know, hanging out in the archives for a year or two?

I mean, history and statistics are blatantly tied together, you know what I mean.

For this NFL project, I’m researching every NFL game played back to the 1920s. It’s kind of remarkable to see how this one sport has survived with significant changes.

But like, you come across something and you’re like, “Why were there no games played that day?” Oh, it’s September 11th attacks or something like that. You encounter changes in real-world behavior and technological changes.

So yeah, any statistical model sometimes can involve kind of extrapolation from first principles, but the most empirical ones are just saying:

We are extrapolating from history and making an assumption

because it is a big assumption—that the trends that existed in the past will correctly extrapolate out to the future.

And often they don’t.

Economic forecasting is notoriously difficult because there are regime changes: the economics of the internet era versus the pre-internet era versus the pre-automation era versus the pre-agricultural era.

Not that there’s great research on these; they are all very different.

I think the field of economic history, sometimes called progress studies, is quite underrated.

The notion of why different societies, great powers or not great powers, rise and fall is vital.

Why is South Korea as prosperous, or more than Japan today per capita, when there was a 10x or 20x gap 40 years ago?

These things seem like really high stakes, really important questions. Because they play out at longer time scales, they often don’t motivate people as much.

But they seem vital and very important— even within… From what I understand, AI companies are not really putting a lot of effort into thinking about what this looks like in five or ten years, even though they tend to have longer time horizons than most. They are not seriously forecasting how the entire world will change if we do achieve superintelligence.

I mean, they talk about it a lot—this is a popular subject on the “In the Door Catch” podcast—but that’s probably substantively more important than what’s in the daily news cycle. There’s this myth of a software engineer who’s annoyed by something in the Spotify app, then joins Spotify for two weeks, fixes it, and quits.

I’m curious if there are any, like, what would a maybe two-week stint look like if you could place yourself anywhere in government? Or maybe what a one or two-year version of that would be, aligned with your skills and interests.

You know, like:

“Let me redesign the constitution”
Maybe we need a ban on partisan gerrymandering
We need to change the Senate and things like that.

To some extent, I’ve done that. I was just satisfied with the way elections were covered. What started as a little two-week project became a life-altering career.

Thinking about the kind of context, I feel like there’s maybe a little more openness now. For example, New York City recently got a new subway map, which is much more legible—that was a nice improvement.

I also feel like I could be a good restaurant consultant, saying things like:

This restaurant’s not going to work because nobody walks on this block.
People walk on Eighth Avenue but not Seventh Avenue; I can’t explain why.
Or figuring out how to get into St. Mark’s.

Things like that, I could help with.

Or the way you’re presenting this really feels like being a kind of editor:

Noticing copy-editing problems in advertisements.
Fixing small issues that make a difference.

So the notion of thinking big enough, come on — what cabinet secretary, what bureau…?

Zohran says:

“I’ll give you any job, Nate. What is it? How to have a good poker scene in New York?”

And I might say:

We need poker rooms but not the rest of gaming.

I guess I sort of agree with the abundance critique that New York takes way too long to build things.

At local government levels, there are often incremental improvements made in different ways. For example, some of the newest infrastructure projects like:

LaGuardia airport
Other airports
West Side development

are all nice, they just took too long. We’re too slow.

But Zohran seems like enough of an outsider and a bright enough guy that he might do smart experimental things that don’t just fall under bureaucracy and inertia.

I don’t know if it was Japan or Korea, but I saw they have little lights embedded in the sidewalk that show you when to cross. That’s pretty cool—simple things like that can make a big difference.

So I think, alright, you’re a department of special projects guy— just making life better on the margin a little bit. Or we’re going to let Nate Silver rewrite the constitution with more barbell theory—think bigger.

Think bigger: everything small, absolutely. Let’s do barbell theory again with money.

Say SPF hits and you’re just his advisor. You have billions of dollars to spend on causes, maybe not dumb stuff—where do you put your marginal

10 billion dollars of philanthropy around politics?

Let’s do it around politics, not just other stuff.

I don’t think politics is a very effective use of money unless it is at the local level.

If you look at projects that were really successful in American history, one of the most successful is the conservative movement’s multi-decade effort to win control of the American court system.

The Supreme Court justices serve five times longer than presidents, on average, so this is ground-level, long-term work that is quite valuable.

But I think the notion of how to make government more… Well, let’s stay on that actually for a second because one of the things that that story did is just invest in ideas and people and kind of like the intellectual superstructure for this movement. There don’t seem to be—I mean maybe we’re starting a little bit now with the sort of abundant stuff and Patrick Collison funding progress things—but there seem to be a lot fewer sort of center or left billionaires who are investing in the interesting intellectual ecosystem to grow the movement.

So college kids or law students are into your thing instead of whatever the default thing is. You probably see more of it on the right. I mean, I think like the Peter Thiel Fellowship program, where he’s paying kids to drop out of college, is an interesting ideological project that has produced some degree of success.

Right? I mean, the effective altruists would say that you want to purchase anti-malaria mosquito nets in the third world and probably that’s very effective, but for big… I don’t know. I mean, as much as I think there’s lots of inefficiency in politics and government, it also reflects a kind of revealed equilibrium from complicated systems and incentives. Maybe change is harder to achieve, and it’s part of the reason why I’m reluctant to give off-the-cuff answers.

I believe there have to be improvements you can make in government efficiency. Why does it cost 10x more to build a mile of subway track in New York than in France or Spain or Japan or whatever? I just think there are a lot of really sticky factors explaining why that’s the case.

In principle, I’d be on board with a project like Doge, for example, but Doge should have been the first thing you did for three months: to study. Not a fake commission but to actually study—okay, where’s the overlap between problems where there really is inefficiency and where it’s tractable and solvable?

That’s not something you can answer off the cuff or just by looking at a spreadsheet.

Would you ever join a campaign? Nate has that. No, it would be about no. I don’t think I’ve ever really been offered that, believe it or not, and it kind of goes against my ethos. I want to study. I want to have the outside view on politics.

I think all this is pretty rough. Campaigns are very tough because you basically have one outcome: you win or lose. In presidential primaries, you have 50 states and they go in sequence, so it’s a little better, but it’s very hard to know what worked or didn’t.

Kamala Harris and Donald Trump had lots of different subtle, nuanced strategies, and the fact is that if inflation peaked at 4% instead of 9% in 2022, she might have won—having nothing to do with strategy per se.

So it’s very hard to get feedback on campaigns. I’m skeptical that you can gain as much through better messaging as you might think. It becomes saturated. The first time the next candidate uses Zoran-style messaging will probably get something out of it. The fifth one who does it might backfire because it seems like a bad facsimile of what he was before—just like there are various Obama imitators or mini-Trumps or whatever else.

People like novelty and some sense of authenticity. What seems authentic is very tricky. Trump, in many ways, is a very fake kind of plasticky person, yet he kind of has this ironic, camp almost level of authenticity that would have been hard to predict in advance.

I was not alone in 2015 thinking, when he’s going down the elevator from Tower, “this is a joke.”

And this highlights the limits of prediction in general. The world is complicated, dynamic, contingent, and circumstantial. Social behavior is contagious. The focal points created by the media and the internet amplify this, where things can change unpredictably and rapidly.

So, I think there’s less political science in running campaigns than in a lot of other fields that might be in my interest set.

Yeah, it’s interesting because the big tactical decision that people are still talking about is:

Why didn’t she go on Rogan?

Even if she did, it wouldn’t have gone well because she does not vibe with that, and it’s a sort of revealed preference of her not doing it. This sort of open-ended media doesn’t fit her, so there is an aspect of like you can only manchurian candidate your candidate to go so far from their essence as a human being. I do think until we’re electing AI models, people can still just get a sense for whether they like or dislike people, and that like is probably two or three percent in a national election just on its face.

When former president, soon to be president Trump, got shot—was it June or July of last year?—like a very sympathetic moment, I’ve never voted for Trump and I wouldn’t, but even then I felt some sympathy. You know that movie—the polls shifted like one or two points, one of the most momentous events of the past 50 years of American campaigns.

When Biden had the worst debate in presidential history, that moved the numbers by maybe two percent. They were two percent that mattered because he was already behind and then fell further behind. There are times when preferences are extremely plastic and times when they’re extremely sticky. Knowing which is which and which interventions are timed to which intervals is probably important.

I’ve started interviewing a lot of these Polymarket investors or traders. It’s remarkable to me that there are no funds or teams. I assume that’s just because these markets aren’t liquid enough. But what would your dream team of skill sets look like if you were going to start the Polymarket hedge fund?

You probably want some AI experts.
You want macro experts.
You want a mix of smart micro traders like poker player estimator types.
It’s more of a barbell approach.
For macro people, you want a China expert.
You want an AI expert.
You want an expert on American politics.

I think on average the takes that Wall Street has about American politics are kind of primitive. You want someone who understands macroeconomics, inflation, and the debt.

In terms of what banks, hedge funds and so forth are doing, different firms probably differ in their thoughts about reputational or enterprise risk to trading. Crypto can be a gray area, prediction markets can be some of a gray area. I suspect there’s probably more of it than you might assume.

Until the last couple of years, there definitely wasn’t enough money in prediction markets overall to be worth it for institutional traders. Now there might be, especially for smaller firms that want to say they are primarily trading non-traditional assets like:

- Prediction markets
- Cryptocurrencies
- Low market cap stuff

There are crypto hedge funds. I’ve worked for a crypto hedge fund, and if they’re getting more into prediction market stuff too, it wouldn’t surprise me in the least. As prediction markets grow, bigger quant hedge funds and eventually Morgan Stanley’s and Goldman’s traders will want to trade them too.

You interviewed a lot of really rich people for this book. Why do they all want to start podcasts, Nate?

They love hearing themselves talk. Remember, it’s not just that they’re rich. These are people mostly in competitive fields, a lot of them in venture capital, where they’ve had success and it goes to their head.

The critique is like what you see in poker: generally, to be in a winning streak in poker is helpful. It improves your attitude, makes people fear you. But you can also go on winner’s tilt, where you’re very hard to stop. When you’re powerful, people start catering to you.

The Emperor’s New Clothes is one of the more accurate fables, alongside The Boy Who Cried Wolf. These are eternal fables that describe human behavior extremely well.

If you’re somebody like Elon Musk, who has made several really good bets (skilled or lucky), consider this: if the fourth SpaceX rocket had blown up, even he told Walter Isaacson, “I was going to be cooked.” It’s very hard if you are one of these people.

One thing I’ve learned from Jordan is that there’s always one more tier of wealth and power behind a closed door than you might expect. There’s kind of one more privilege level even in an event. That’s already privileged. There’s a VIP room, and a VIP room within the VIP room, and then the biggest VIPs of all are not even at that — they’re already at the after party.

And there are smaller worlds too, like the number of people who’s a Keith Rabois or Rabois (I know you say it), who told me like there are really only six people in all of Silicon Valley that matter, which is an exaggeration but maybe not directionally that wrong. Maybe a few dozen, and they all know one another.

I think with the tech types, the VC types, they feel like they’re embattled. Their employees are too woke, and the media is mean to them. People are mean to me on the internet sometimes, but I don’t know. Sorry, so square the like there’s always one level up thing. There’s always one deeper circle of hell to the or privilege or so. You know, okay, but like what is like, okay, but if that’s your theory of the world, then why do you want to have an audience of 10,000 people listening to you talk about the news?

Well, look, part of what I’ve done is go from a giant platform. I used to work for ABC News, which is about as mainstream as it gets. The average watcher is 70 years old at an airport somewhere or something, or maybe a retirement home.

And now with Substack, it’s an audience that turned out not to actually be narrower because the notion of building an email list is a good business model and very sticky. But at first, the stuff that goes behind a paywall is definitely reaching fewer people, but they’re willing to pay, and you’re self-selecting too.

Because the work I do, especially when it comes to election forecasting, is so easily misinterpreted, I don’t mind having a filter for people who come to the problem with more knowledge and can make the writing sharper.

If I am freelancing with the New York Times, I have a good editor over there who often says,

“Slow down, you have to explain this thing better.”

It’s nice to have a conversation where you say, hey, you’re starting with this and that premise, and you’re starting with a memory or you’ve stated this complicated thing before about your political views on an issue that might come into play or you’ve disclosed, and you consult for Polymomentum — it’s a cumulative project, and inherently nobody has time for everybody’s cumulative project.

So to have a smaller audience of even dozens of people, hundreds, or tens of thousands, it’s pretty important. Especially if there is an unfortunate degree of concentration of influence, wealth, and power.

I’m working on a book, and one of the things that surprised me is how many people that I had no connection with were willing to have a conversation with me or at least provide a polite response if they had a good reason not to.

It’s a pretty small world, and people talk to one another. That’s something that’s shifted a little bit again, maybe toward more this great man theory — all of them are great men, all of them are men.

A lot of them are toward the theory where individual agency matters more. My two cents on this is that I think there’s a big cognitive bias of Keith Rabois to want to think that he is the center of the universe.

There are more times than you would think where you have the market providing the discipline or the people or the bigger sentiment when something blows up — like politicians or the media end up reflecting mass opinion more than the opinion of three people who are trying to pull the strings.

One tip I heard in poker recently is that:

“Everybody is the main character of their own poker story.”

If I got caught making a big bluff against the third party earlier in the hand and you’re sitting at the table not involved in the hand, Jordan, you might not even notice that — you might be on your phone.

If you got bluffed by another player earlier and I’m not involved, that affects my surroundings against you more than what I did before because I’m not involved in your narrative, except to the extent I affect you.

I feel like life is often the same way.

By the way, China talk, it’s the only narrative that matters. Keep tuning in for more visions into the future.

This is really fun, thanks, Nate. I guess my last inside baseball question is like Silver Bulletin, it’s a great name. China talk, I don’t know, I gotta get out of it. Somehow, I don’t want it to be the Jordan show, so people didn’t like the China or people didn’t like Silver Bulletin name at first because I came up with it in like three minutes. I thought Twitter was gonna die, and it did. Some placeholder. A lot of names are stupid when you think, you know.

Well, I think about sports names, they are kind of, you know, the Green Bay Packers—it’s kind of a dumb name, you know what I mean? But it just sticks and it kind of seems normal because people repeat it over and over again. Yeah, it’s not like I have names I’d be fine with, it’s more just like the switching cost is not transparent to me—like how much it’ll change listenership or open rates, or whether it puts me on a larger growth trajectory.

Because, like, you know, I wouldn’t change my coverage to do less China. Right now it’s still like 50 percent China, but it would just send a new signal that it’s more than China here.

That’s an interesting point, Kate, because if it’s just a name, it’s a little awkward or because it does include an implicit premise for what the subject is, that might actually—most of them would say, yeah, the switching costs are actually pretty high and to sacrifice brand recognition is costly. Right?

Yeah, maybe you need to start a sub brand or something. I don’t know, that’s trickier than for most people, I think.

Yeah, because the other thing is like advertisers are terrified of China. That’s just… well, that tips over to… yeah. So then I think it’s more like once I have a big contract, once I have Google telling me Jordan will buy five hundred thousand dollars of ads only if it’s called, you know, the Jordan Schneider Show, then you’ll know we’re doing 500K. I think that’d be worth it.

All right, Nate, we end every episode with a song. Do you got like a river song, village song? Got a favorite poker song? I don’t know, a betting song?

I like shoegaze rock, like my buddy Valentine. I’m dating myself a little bit.

I don’t often listen to music when I’m playing poker, but I was trying to, like, you know, when I’m tired, sometimes I’ll do it to avoid the annoying conversations. Listen to ambient stuff like Aphex Twin a lot.

I used to be in the music phase. I played guitar for about six months. Thank you.

大象, 你说你像一头大象, 蜷锁身体把梦隐藏, 我的猫和我都不曾遗忘, 冲装, 低级和重量的冲装, 不要偷走了你的汤,

不定像欲望不断膨胀, 做极其怪的梦, 在空道的这下,

不懂这太阳, 在你跟上你, 冲装的光,

在你跟上你, 冲装的光,

你某种种类, 成为一个之后, 可是大象, 在在在在在在在想求我。

大象, 你说你像一头大象, 蜷锁身体把梦隐藏, 我的梦和我都不愿意。

大象, 你说你像一头大象, 蜷锁身体把梦隐藏, 我的梦和我都不曾遗忘。

大象, 蜷锁身体把梦隐藏, 我的梦和我都不曾遗忘。

大象, 蜷锁身体把梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我都不曾遗忘。

在空草, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我都不曾遗忘。

在空草, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我的梦隐藏, 我都不曾遗忘。

I’ll see you next time.

Josherich's Blog

Nate Silver on AI, Politics, and Power