← New Books Network
New Books Network
Fenwick McKelvey, "SimPolitics: America’s Quest to Solve Politics with Computers" (MIT Press, 2026)
00:51:22
Loading summary
Transcript28 lines
- [00:01]
A
Welcome to the New Books Network.
- [00:05]
B
Hello and welcome to another episode on the New Books Network. I'm one of your hosts, Dr. Miranda Melcher. And I'm very pleased today to be speaking with Dr. Fenwick McKelvey about a book just out from MIT Press in 2026 titled Sim Politics, America's Quest to Solve Politics with Computers, which has been a decades long goal of computers being able to revolutionize both domestic and international polit, making things better. All sorts of people have tried. We're going to be talking about computer programming, but also political scientists, academics, activists. There's all sorts of ways that simulations have been used for elections to predict what might happen in geopolitics for all sorts of different reasons with all sorts of different results too. So I think we're going to have quite a number of things to discuss. Fenwick, thank you so much for joining me on the podcast to tell us about your book.
- [01:01]
A
Miranda, I'm so happy to be here and be able to share about the book today.
- [01:04]
B
Well, I'm pleased you said yes to this. But before we dive too much into the detail of the book, can you introduce yourself a little bit and tell us kind of the origin story? I mean, why did you decide to write this?
- [01:17]
A
Yeah, it's a fun story. I'm a professor in communication studies and so I have a kind of eclectic background in digital policy and digital politics. And I've been studying, strangely enough, or what feels strange at this point, the impact of the Internet and digital technology on elections kind of live for the last 20 years. And one of the parts about that is that each campaign there's a new cycle and there's a new discussion around how Twitter, Facebook, Instagram, and now AI is going to change the campaign. And so year over year I've been able to keep up and track and follow these things. But when I was first starting and first starting and doing my PhD long ago, about 20 years ago, I started looking around to think about where these types of technology are coming from. Where do we start thinking about how computers were part of the campaign? And I remember really vividly wandering around the stacks of my doctoral university, now called Toronto Metropolitan University, and chancing on a book by Robert Sharpbrand called Computers and political campaigning from 1972. It's a fantastic book and you should look at it online because there's this cover of a robot winning on standing on a platform with a computer. And it really dawned on me that there was this much longer history of computers and technologies than this everyday cycle was allowing us and, and I wanted to actually go back and start thinking about where and when did political campaigns become so related to computer technology? And I had a chance in 2012 to actually study this during my postdoc at University of Washington and beginning to look at these organizations and this much longer history. And that led to this almost decade long project of trying to piece together when and where we see this idea that you can take a political problem, something as complex as how a voter might make a decision, and put that and translate that into a computer problem and a computational problem. And I'll get into the kind of history, but I think it's helpful that it comes out as a motivation of trying to make sense of what's happening now and thinking about history as a chance to go back and make sense of what's different, what's changed, that's helpful
- [03:50]
B
to understand the origins of the project. And of course that's a really key question you've highlighted there of kind of wait, when, how, why did this sort of all start? Are there any other key questions that you're asking in the book that we want to lay out as a foundation for our discussion?
- [04:04]
A
Well, when I was looking back and I had more time on this, one of the other images that really stuck with me was from a 1980s, 1981 issue of a magazine called Campaigns and Elections. And so part of the evolution of the book is that I became interested in the history of campaigns, but then had a pause as I started my career and figuring out life as a new professor and became more interested in kind of digital history and wrote another book called Internet Demons, which is a history of the ways the Internet works and how it kind of behaves like an artificial intelligence. So there comes this slosh of ways that these initial questions about political campaigning become interested in this larger milieu of how computers and technology become seen as a solution to, as a way of kind of managing societies. And what's always struck me is that looking back and thinking about the historical work, I chanced upon this issue of campaigns and elections. And there's this advertisement for a company called Aristotle, which sold what we now call a campaign software or what I like to call a political engagement platform. And it basically was an early database that parties could use, but it said, and it described it as the machine to revolutionize politics. And I remember sitting back and wondering how, and this is, you know, 10 years ago, five years ago, how was it that Aristotle was able to make a claim 40, 50 years ago that the computer was going to revolutionize Politics. And how has that claim been able to be made again and again? And so I was really trying then to set up how we see the problem of computers as something, or see the problem of politics as something computers can solve. How is it that computers did evolve and integrate into the campaign? And then I think, how do you see this as something that is a very specific American case? And I think the last thing just to emphasize is that as we see broader histories, media histories, expanding our scope of where we're thinking about these questions being solved and posed. In the thinking of Adina Medina's work working around the Soviet cybernetics, it became clear that we didn't have a similar type of work that was asking about that, about the United States and making that unique. And so it then became this project not only about just how computers were going to be solving political campaigns, but more than how specifically in the United States did it become possible to imagine solving political problems with many types of computers, from databases to simulations to game theoretical modeling, it becomes this whole milieu, and that becomes the thing that kind of drives and animates the book for me.
- [07:04]
B
So thank you for laying out those questions there that you're asking. Some definitely quite intriguing ones. I don't know if we'll answer all of them and as much detail as you do in the book, obviously, but you've laid out a useful foundation obviously for the book and I think also for the rest of our conversation. So moving then from questions to answers, you have specific answers to each of those questions you've raised, but you also have some kind of key ideas that come out in multiple of these answers throughout the book. So can we maybe discuss some of those important ideas like models, prototypes and imaginaries?
- [07:40]
A
Yeah. And I think that as the book developed, there became this opportunity where I saw that how I was thinking through the turn towards thinking about politics with computers was helpful in this wider context, about how we're dealing with the impacts of many technologies in society. And we use and throw around a lot of terms, platforms, imaginaries, discourses. And I felt it necessary to define the kind of method that I was using for the book in hopes that's helpful, and I hope for listeners it's a helpful way of trying to track how we see this link between new technologies coming about and how do we imagine and think about societies more broadly. And the three terms I use is prototypes, models, and imaginaries. And the way I describe it in the book, and the way I learned it was as thinking about like as an iceberg at the top. And the tip of the iceberg is what we have everyday contact with, which is what I call a prototype. And a prototype is a working version, something we interact with, we use as kind of operationalizing a certain theory of a politics. So when I think and I talk about Aristotle's political machine, it's an example of a prototype. It's an attempt to use a computer to understand, make sense of, solve a political problem. And then below that, kind of further down of the waterline of the iceberg, there becomes this question of, well, how is that problem being solved? And how does the prototype encode a certain version or vision of politics? And that's what I talk about as a model. And models are analogs and ways of thinking about how systems work by comparing them to other systems. And the model is, I think, a very kind of important concept in the field. But it's also, I think, you know, helpful to try to then get at what are the politics going on here? Because these models have assumptions. And that's part of what makes a model, is that you have to estimate, you have to make abstractions. And so what are those abstractions? And then below the surface is what I call what we think of as this term imaginary, which I like. And I think it's really become one of the terms I can't get away from, because it's so helpful to say that when we start talking about politics and how we describe it, how we make sense of it, it depends on all these kind of lived assumptions. And this becomes an imaginary, how we think up and dream up American politics. And the imaginary, to me, becomes really central as that through models and through prototypes, this imaginary becomes stabilized and also dependent upon it. And those three concepts, I think, work together throughout the book, both in the context of domestic campaigns in America, as well as the kind of international efforts we see in the book to kind of form and shape what I call a computational imaginary of politics, which is this idea, this kind of discourse that helps legitimate computational solutions to political problems in the United States.
- [10:55]
B
Okay, that's a really helpful explanation of the three concepts and the iceberg way that they relate to each other. So we're going to keep that in mind then, as we go through some of the specific answers you developed to the questions you mentioned earlier. So going back to the kind of biggest umbrella question you've raised, how, when and why do we get ideas and methods developing for using computers to simulate or model all sorts of different political questions and behaviors? What is the kind of origin point and reason.
- [11:27]
A
Well, one of the characters of the book that I think is really helpful is someone named Ithiel de Soila Poole. And for folks in communication studies like myself, that name might be familiar because he's really part of or well known with the MIT Media Lab, this interest in kind of early communication and freedom, kind of a foundational thinker in communication policy in the United States. But he has a much, much longer history. And when he comes about, he really has this backstory about being a defense cold warrior, defense intellectual, that for his early part of his career was really working for contract work for the US Government. And he's this really important figure because in ways he becomes the kind of boundary point for two parts of the book, like the beginning of the book in how computers become proposed as a solution, and kind of the end of the book where he's almost this figure that becomes ambient in this kind of mainstream culture. So kind of is a good way of putting at the end of the book, so Ethio de la Soil Le Poole at the start is this really interesting figure because in the 1950s, late 1940s and early 1950s, there becomes a kind of crisis in how conventional approaches to international domestic politics work. And there's a sense that there needs to be more rigor in the field, and there's a desire to be more specific, more technical, more precise in the methodology. It's often called the new political science in the United States. And this kind of desire to move away from more human humanistic methods to more interpretive approaches to politics is captured and encapsulated by Poole as someone who's then proposing, well, if this is the desire, what we can do is we can build different technologies that are going to solve these and answer these political problems. And he becomes then interesting as someone people believe in and trust. And he's famous for and made famous in Jill Lepore's book if Then, which is the history of the Simulmatics Corporation. And in the early 1960s, he was helping in some ways through the Simlamatics Corporation to create a model of voter behavior that was used to advise the Kennedy campaign. And so the simlamax Corporation was a kind of early entry for me because I saw it as this point where here's this kind of wild idea of using really, really primitive computer technology to model and predict how voters are going to react to the Kennedy campaign and Kennedy's position on certain issues. And it's because there's this desire and moment for something more rigorous for this interest in computer technology to kind of solve it, for this way of trying to move past the limitations of a previous approaches to politics, that we see this potential, this demand that people like pool take advantage of and use to kind of position and launch their careers. Now the problem, and one of the reasons the book went in the direction it is, is the Simulmatics Corporation, in the end, sort of becomes a smoke show. It never really does what it says it wants to do, but it becomes just part of this ambient culture that makes this book so big and so wide, is that it's all these different efforts at the same time that are capturing that same effort, that same desire, and trying to put forward their own prototypes for politics with computers. And so really, to me, part of the book is then attending to these early moments when computers became desirable as opposed to what they were doing previously. And I think a lot of that really depends upon the concerns initially about humans being too inaccurate, too imprecise, and looking for some new ways to inject rigor into the study of politics.
- [15:38]
B
Got it. If those are the goals, though, of injecting rigor, because humans kind of aren't necessarily good enough at it, there's the kind of immediate challenges of if you're trying to solve politics using all this technology, you. You have to deal with the complexity of humans, right. If you're trying to predict voter behavior, I mean, voters are human. So what were some of the different models that were tried? It sounds like some of them did work, some of them less so, to kind of grapple with the complexity of voter behavior. I mean, you mentioned models earlier, kind of running throughout the book. This is a place to me, they seem really important to discuss in terms of what kinds of ideas are literally being encoded about how people react politically.
- [16:18]
A
Well, Miranda, I think that that's also the central tension of the whole book. And I start like, how did politics become so routine in being modeled? Or how did political modeling become so routine? And it's just this strange thing where in tracking this kind of break, these breakdowns, part of what I always want to do is make strange is it's really weird to say that we're going to predict all this complexity of voter behavior or behavior of nations with computers. I mean, it's easier to do, easier to think. That was weird when you're talking about the computers, the 1960s, which are really primitive to today's standards, but even still, it's such a big stretch. So the part for me, and a key takeaway for listeners, is that there wasn't one solution, or there wasn't one answer to how to use computers to solve political problems, but many. And that I think is something I'm really proud of in the book, is it kind of unpacks and open up the politics of the computational imaginary, where you can see how the same type of technology, the same types of prototyping, had very, very different political projects. And in the book, I focus on three, and these come up kind of organically in my studies. And that's part of how the book came about is I'm reading a lot, I'm studying a lot, and I see these kind of trends emerge. And in the book I describe them as cybernetic citizens, rational agents, and data shimmers. And each of these are three different ways that computers are being used to model voter behavior. So when I was talking with Ithiel de Soiler Poole, he's really emblematic of this kind of cybernetic citizen type approach which thinks of humans as kind of primitive computers that are navigating complex environments, trying to achieve something about a balance or what cybernetics would call homeostasis. And that becomes a working theory of how voters are responding or interacting in campaigns, because they'll see certain signals or certain stimuli and they'll react and respond to it. And this creates a really active voter. And something which I think is important is that part of the book is showing and kind of asking us what, what ways or what or how models are proposing certain forms of agency. And so cybernetic citizens are active voters and they're very much aligned with the rise of consumer culture. And this idea that advertisements and messaging and information is going to change behavior. The second type, and I think kind of in contrast and emerges slightly in contrast to this kind of frenetic cybernetic citizen, is this idea of a rational agent where you see scientists and researchers and political scientists trying to find some way of explaining how these cybernetic citizens are working. And they turn towards economics and rational choice theory and game theory to say, well, humans are rational actors. They make decisions based on what's going to benefit them, what's talked about, as expected, utility. And we can create these models that are going to predict voter behavior based on calculations and ways of calculating how voters are making decisions to maximize their benefit. And that's a second school that really is very influential in political science that talks about why people vote and the voter decision theory. Now a third school or a third approach comes up in what I call Data Shimmerez, which is really trying to Identify this trend that we see in parallel where a lot of public data is coming online through the release of a digital versions of the census in the United States and the release of zip codes or certain forms of targeting. And there's a lot of public information available. And savvy campaigners realize that, well, if we integrate and use this data, we can make better guesses or better predictions of the electorate. Now that's a very different version of politics because with census data it happens every 10 years. You're dealing with an assumption that people's behavior is, is going to be the same for a long period of time. And these three different models, and they're just the ones that I found in the book, I definitely think there's more really then kind of demonstrate how through computers there became a way of thinking about how voters would behave actively, more passively, depending on certain types of data or certain ways of using computers. And they become influential in how we think about how voters now behave and how their behavior is really a proxy or closely related or imagined to how they became worked out with computers.
- [21:29]
B
Okay, that's really helpful to discuss the different models, but I want to kind of stay for a moment on what you just mentioned there in terms of became influential, because we all know that it's not a given that an idea of how things work actually gets put into practice or becomes influential. That's not something that happens in every instance. So how did these early models start to get actually used by political campaigners and therefore become influential?
- [21:58]
A
Well, if you remember, I was talking about this book, Computers and political campaigning from 1972. So that book was always a touchstone for me because it was a guidebook for, for political professionals to how to use computers in their campaigns. And by this time, by the 1960s, that's almost 10 years after the kind of famous symbol Max Corporation. It was a really good point of asking, yeah, how were they being used? How were early mainframe computers, these kind of, you know, the big servers we now call them today, that type of computer, how is it being integrated into campaign? And what I found, and it was kind of surprised by, is that certain more kind of wild versions, simulations or cybernetic citizens weren't as influential. And there was a turn to kind of these more mundane versions, kind of data shimmers or rational agents. And so what became, I think part of, you know, writing the book was then actually developing a history of how political campaigns in America developed, evolve and which versions they picked up on. And it isn't necessarily the coolest ones, certainly not the wildest ones. And in the book I talk about a few different examples or cases of this adoption. And the one that I really like is actually Winthrop Rockefeller, who was running for governor of Arkansas in the 1960s and was one of the Rockefellers. He was a Rockefeller. He kind of sconced for various reasons to Arkansas and he wanted to run as a Republican in a profoundly Democratic South. So it's lots of kind of backstory and context. Particularly this is dealing with an era of segregation. And so it's. He's an interesting figure and he basically uses his money to buy one of the first IBM machines to run in the campaign. And so this becomes this kind of interesting test. Here's someone with kind of boundless resources trying to shift the needle in his kind of adopted state and turns towards an IBM computer. And so what does he use computer for? Really what he starts using computer for is collecting data about voters, trying to make sure that they are addressable in this big political database and convince them to come out and vote. And so the way we see it is that the certain different forms of prototypes and certain different techniques of ways of modeling voters become more or less attractive to campaigns. And so one of the big things for me is that these stable versions of voters and these voters that become addressable for computers become really important and really attractive for campaigns. And Winthrop Rockefeller wins the governorship. And a lot of people see it as an important prototype of how computers could be used to win elections in the future. And he's not a well known name now, but I think that it's a really important first case of seeing this kind of transition from this experimental technology into everyday campaigning. And then it just becomes part of standard practice kind of after that. And all the political parties and 10 years later you see this real kind of expectation that there's some form of computer presence or some form of information system for tracking voters now embedded in a political campaign. And that's kind of one of these stories that I look at and try to draw out, trying to gauge these different models and be like, which ones worked and which ones didn't work and which ones became attractive political campaigns and which ones.
- [25:47]
B
I mean, clearly there's enough popularity here that as you said, I mean, it gets embedded pretty quickly. And there's lots of learning, obviously that can happen sort of from cycle to cycle. They're like a regular kind of predictable sort of thing that happens. And in some ways it sort of makes sense. I mean, yes, you mentioned earlier we should think about the strangeness of models being able to predict voter behavior, but at least we're talking about kind of an election cycle. A particular moment, a particular decision that voters have to come to. You know, candidate A or candidate B are at least some kind of limits on predicting the future there. If we then turn to the realm of international politics, it seems like there's so many more variables that it would seem even stranger for computer models to be seen as potentially helpful here. But you discuss in the book, they do start to become potentially relevant to understand international politics. So how, when and why does that transition happen?
- [26:46]
A
Well, I think it's helpful to kind of contextualize some of the history of the book because I think first off, for readers, it's like, oh, there's another whole part about the history of computers in international politics too. And I think that was always, I think, one of the parts that I'm trying to finish this thing, trying to wonder to myself, why did I try to do both at once? But to me, it really goes back to that character I mentioned a bit earlier, Ithir de Soila Poole, is that when I was trying to talk about his narrative, it was way easier for me just to think about it as a story of American political campaigning and just focus on American voters and American voter modeling. But that he's a character that is trying to apply these techniques not only to understanding American politics, but also the Vietnam War. And he's emblematic of a whole industry that's part of this training zone, which is saying, well, we can manage politics broadly, writ large, internationally and domestically better with this more rigorous computational methods. And it became impossible for me not to talk about both these efforts in the United States, but outside of the United States, because they happened at the same time when with the same people using the same techniques in very different, wild, wildly different contexts. And that was then also really important because it helps make the weird part even weirder. And I think it also helps kind of provincialize the problem because it's the United States making these really, really, really detailed assumptions about how American politicians, American voters are going to behave, but then really, really, really broad assumptions about how into, you know, Vietnamese villages are going to behave. And it's this contradiction that makes both the. The politics of this work more important and relevant for me because it's. It's awkward, but it also, I think, helps just talk about the scale and scope of this computational imaginary politics is that it wasn't just something domestic, but it was something that the United States was using to see the world around it and see itself in that world.
- [29:01]
B
Yeah, I think we definitely want to talk more about kind of the ways in which computers were seen as helping make sense of the US in the more global picture. Like, obviously, especially given the complexities of something like Vietnam and the Cold War and kind of trying to predict three moves ahead of what's happening in the Kremlin. Like the US was trying a lot of methods of making sense of what was going on. Can you tell us more about how computers came to be seen as like the way to make sense of things, not just sort of one of many options?
- [29:32]
A
Well, first off, that there's a bit of different context in the international scene than there was in the domestic scene, because the motivations to use computers came from really different sides. In one sense that, well, before computers, there was a lot of attempts of kind of mathematically modeling nuclear war strategy. So talk about game theory, mutual shore destruction. These types of approaches to simulating politics were kind of popular. And if you watch a movie like Dr. Strangelove, one of the characters, Dr. Strangelove, I think works for the Bland Corporation, which is a joke on the RAND Corporation, but which was really seminal. American firm doing a lot of this work. And so they were providing insight into guidance to the American government about its nuclear war strategy as well as kind of international strategy. And a lot of that was mathematical. And so in that sense, you had one side of people inside the RAND Corporation that were frustrated with how narrow the modeling was, too mathematical for dealing with international politics. At the same time, you also had. Which I think was, you know, you know, my thing you might find interesting. But peace activists and peace researchers were also trying to be rigorous and, and try to position themselves with more expertise than some of the hawks that were defining the nuclear strategy at the time. And so this is a bit of a pre history for the book itself. But computers become useful because they're rigorous. They're a way also of being more dynamic, more active than some of the mathematical modeling that you. That you might be seen as circulating as kind of conventional theory. And there becomes this term to using computers as an important middle ground or alternative to these kind of two schools. And the figure that I always think is so central and so forgotten, there's a guy named Harold Gutsko who is a peace activist. He was a conscientious objector and finds himself in these really bizarre, through these really bizarre turns, getting a lot of funding from arpa, which is now called darpa, which is the Defense Research Funding Agency, to develop world models Develop models of the, you know, international relations with and augmented by computers. And in this time there's a lot of different versions of this. And some people are really interested in making games sim, you know, human simulations. A whole other group is trying to create all computer simulations. And so in this moment you see this turn towards computers being kind of interesting. But how much do you want to integrate computers becomes a really important question. And in the second part of the book, kind of paying attention to this, this moment and this become what becomes the breakdown here, where voters, where not voters, but where world international relations scholars become to get sick of using humans and turn to what they call what I call all computer simulations. So they're like, ah, forget this. And so this shift or the way that, that computers come about as the more influential way or very important way of modeling world politics happens because there's a desire to create more rigorous accounts of what's happening. And so you can open up the kind of black box of a computer much better than you can a human to explain what's taking place. Place. There's this also simultaneously a lot of data internationally is now coming online. And so if you use computers, you can use and process all this data to make more and more amazing computer simulations. And then finally it's so much work trying to create simulations with humans. And there's all these great stories of world modelers having to struggle over the summer to get high school students to work on their different human simulations of world politics, to make predictions. And it's just so much effort that it's really appealing to have a digital version of that that you can just run and deal with and not have to stress about, you know, whether the room's too hot or whether the high school students are taking it seriously or not. And that's become this kind of host of factors where computer simulations, all computer simulations become a really desirable way of treating and dealing with world politics. And then I would say, just as a final note that this also gets back to like one of the tricky parts of the book was emphasizing who's paying for this. And often it's the US government that's paying for this. And in this particular context you see how the US government wants a tool, a decision making tool that's useful to them and that I think computers become, in world simulations become at least plausible as a more useful tool for power to have this, you know, to borrow the frights of Donna Haraway, this God's eye view of the world that they can make sense of. And that's better than the alternatives, which would be human exercises or some sort of hybrid of the two. And that all computer simulation is something more useful to the kind of eye of power of the United States.
- [35:23]
B
Very interesting to see the kind of theoretical and practical reasons all sort of combining there. But I noticed you mentioned computer simulations or computer models, plural. Right. There's more than one option at this point and you discuss a number of them in the book. And in fact, it sounds like a bit of a divergence in computer models used to understand global politics. So can we discuss that?
- [35:48]
A
Yeah. In the book? Well, I think that's a great question. In the book I really am trying to kind of map out systematically different ways, different vantage points of this computational imaginary of politics. And by the course of just writing it, I realized that an opportunity to kind of describe differences in the models best when talking about the kind of domestic politics of what I call the first part, sim voters, and that it became easier to talk about the differences in prototyping or ways of putting these in action in the second part about sim worlds. But it is important to say that these three types of voter models that I see in the first part of the book, Data Shimras, cybernetic citizens and rational agents, appear in the second part of the book too. And that was a really big moment for me as I realized that the same types of theories were working out. And that's what I say with such confidence is it really becomes easy to say, oh yeah, that version of that school, so to speak, of modeling voters was doing similar work at the international level. But what I become really kind of interested in is the way that these different schools or different models become operationalized. And in the second part of the book I focus on this kind of distinction between two. One is this more cybernetic approach called Globus. And the second is another approach, more rational agent based approach called factions. And I think these two are really incredible, both as histories, but also to show the differences in this approach to kind of world simulations. Globus is largely forgotten, but for anybody who studied political science, you might know the name of Carl Deutsch, who is a quite influential figure who popularized cybernetic thinking and systems theory and political science. And his last run, his last hurrah, so to speak, was actually working at the, at the Weizenbaum Institute in, in Berlin to develop a world model. And he puts together this kind of incredible team of people to try to build a not entirely benign, but I think certainly a model that's trying to get in the hands of everyday people to help them deal with kind of long term problems. And this becomes this world simulation called Globus. And it runs from the 1970s to the 1980s. And it's this really incredible journey of trying to put together a version of politics really based on kind of cybernetic theories of how states are going to act, interact with each other. That's one of the rare world models that really embeds politics inside of it. And it is working and it works and then it becomes. They just finish it and they're done. But at the end of it, I thought it's really important they develop a way of getting this very complicated model working on a DOS computer and sell it as a software that anybody can use. And I think that that's kind of this really interesting gesture for this one camp, which is trying to see, well, if these world models, how do we increase access and take this tool and put it in the hands of researchers, complicated and clunky as it may be, in a way that's going to foster more discussion and help create this kind of vantage point. The counter version was that with this tool called Factions, which was a way of doing very simple modeling of how different actors are going to interact in a certain situation. It's used for negotiation. And in that sense it becomes a tool that's not about modeling or helping people see the trends of world politics like Globus is. It's a way of winning and giving a strategy to win. When you have multiple actors and who are all working in their expected interest and looking for their own kind of expected utility and trying to figure out a strategy that's going to develop the coalition and ultimately allow one side to win. And that tool becomes really influential, as I can tell in the United States especially, being used for time at the Central Intelligence Agency to be able to help advise some of its strategy. You never know how much. But this is something where the vantage point is not something about helping people understand the world, it's helping people win. And I think the difference between the two is, I think quite striking is that when you're building these world models, what's the intent? What's the type of agency you're trying to give the user? It's a really important consideration. And in one sense, this more clunky version of a world politics in Globus, I think is. Is something that's so compelling because it's trying to make present the complexity of the world versus I think, something which is, I think, a bit more familiar to Us, which is this problem, this new technology is going to be something that's going to help us win or succeed in our interactions with the world as we see in a program like Factions. And that, I think, is an important lesson or is a kind of takeaway today when we're trying to deal with what are these kind of differences in technologies and trying to see these technologies as having very different politics and how we make that more clear.
- [41:33]
B
Yeah, actually, I would love to jump to what this means today, because I think that is an incredibly. Those two models are kind of a clear example of some of the debates I think we're still very much having with technology. So what other sort of influences or legacies of these technologies, but also these debates about the technologies do you think are still around and kind of hope people will become more aware of through the book?
- [41:57]
A
Well, it was always challenging writing the book in that I wanted to talk about the present day. And really, in the end, it became important to bound it as a historical project. And as much as I would have liked to have written an account from the 1950s to the present day, I could only get up to the 1990s, where I feel like the computational imaginary politics really stabilized. And so, in one sense, I'm really proud of the book. And I also really am hopeful that it's helpful to readers because many of the issues that we're dealing with today, you can kind of see the genesis and see there's kind of stabilization in the kind of by the end of the book. And so, like, a figure like Israel Pool has this kind of arc, but then nowadays is just seen as kind of a kind of early figure of a version of the Internet that's really familiar to us. And so for me, the takeaway is that the book really kind of emphasizes two critical questions or two approaches to critical questions that are really kind of relevant is what is the kind of agency that this computer system is proposing for voters, for any type of system? So when I'm talking about models, I'm talking about what are the agencies that the computer is allowing for us to imagine voters to have, whether they're active or passive voters? What are the agencies a system is allowing the user to have, whether they're in control of the world or they're a participant of it? And at the same sense, what are the ways. What's this question of addressability? What's the link between these simulations of this kind of digital world to the real world? And that can widen and narrow. And I think that that's Another important kind of question is like, how do we look at these computer systems and make sure that we can understand this link and what is that link? And I think about that all the time with artificial intelligence, because artificial intelligence is the new technology for the campaign cycle. And AI is something that I was kind of was looking at the book, but many people didn't use it, or the type of AI that was around didn't work as well as other alternatives. And so it never becomes that influential, even though it's a technology that's as old as many of the other in the book itself. But AI today is really making a claim that's kind of reminiscent of cybernetic citizens, where we have the kind of very active, agentic, moving theory that we don't fully understand because there's all of these different factors at work, and that has a risk of basically behaving more actively or more dynamically than voters might behave themselves. And so when I think about AI today, I ask myself, well, what is this kind of agency? What is this kind of theory of political behavior that it's putting in place? And then what are the ways that this is allowing political campaigns or politicians kind of connect or touch with the real world? And in that sense, to me, it's like there's this real question of whether and what we see now in. In both the international and domestic stage, this kind of effort to say, well, we can use large language models as a proxy to voters or world leaders. And so we're also seeing this idea that it's becoming more acceptable to use the abstraction of LLMs as a proxy to or as an equivalent to kind of real world politics. And I think the book itself at least helps expose some of those risks, because those systems are the same thing, and treating them as two really creates risks and problems to how politicians or even worlds are allowed to be understood or rationalized in the kind of games of power that we see play out.
- [45:54]
B
Well, I think that's a lot of things for listeners to take away about history and how we got here and what we might need to think about going forward. So I think a good place to end our discussion on the book. But what, may I ask, are you working on now? What's on your desk? Anything you want to give us a sneak preview of whether or not it's related to what we've been discussing?
- [46:15]
A
Well, there's like, short term and long term. I think I just had the opportunity to give a talk at Goldsmith's about ways that you could take this book and describe it as different versions of political freedom. So in one sense I'm actually kind of excited to, to take and explore, even though I think some of my points have been a bit abstract. Take and push some of the abstract parts of the book and think about them as consequences for theories of democracy. And in that sense I'd like to kind of push and draw out how mainframe computers, personal computers and now AI are kind of proposing different prototypes of what is idealized political freedoms and trying to make sense of whether this moment that, that we're in with AI right now really is a break or distinct break in the computational imaginary than what I laid out in the book. And so I'm trying, I'm trying to make sense of that. And there's parts of me that think that that AI is, is a return to some of the more early visions of computers as being something that, that is smarter and more intelligent than humans that we see at the start of the book and trying to then see how much that plays out. So yeah, trying to make sense of that. The other part is whenever everybody talk about the book now, they start asking me whether it's about prediction markets. And I didn't anticipate that by the end of writing this thing that all the world would be prediction markets, which was an idea kind of as old as many of the ideas in this book that I don't really talk about. And so I am kind of interested a little bit more in how kind of fantasy and games and gaming, which is part of the book, is put into practice through these kind of prediction markets and trying to think about the kind of consequences of something which I don't think is computational, which is this kind of more kind of game like or form of gambling, and how that became kind of a problem, persuasive way of describing and rationalizing how the world works. And so down the road I'd like to kind of try to maybe answer this question around where did prediction markets come from? But I don't know if prediction markets will be around at that time. So still remains an open question.
- [48:38]
B
Fair enough. Well, plenty to keep you busy then. And of course listeners can read the book we've been discussing titled Sim America's Quest to Solve Politics with Computers Politics, published by MIT Press in 2026. Fenwick, thank you so much for joining me on the podcast.
- [48:53]
A
Miranda, thank you so much for the patience and the great questions.
- [48:57]
B
Thank you for listening to this episode of the New Books Network. We are an academic podcast network with the mission of public education. If you liked this episode, please share it with a friend and rate us on your preferred podcast platform. You can browse all of our episodes on our website newbooksnetwork.com Connect with us on Instagram and Blue sky with the handle ewbooksnetwork, and subscribe to our weekly Substack newsletter at newbooksnetwork.substack.com to get episode recommendations straight to your inbox.