The Political Impact of Ideas: Evidence From England | James A. Robinson | 2025 Ryerson Lecture

Thank you. Oh, yes. I have a microphone. I don't need to stand there. Thank you. Thank you very much for those remarks, and thank you very much for inviting me to give the 2025 Ryerson lecture. It's a great honor, and I feel under a bit of pressure after hearing how important it is. So let—I'll try not to disappoint. And, of course, I am going to talk about institutions, but I'm going to talk about some big issues with respect to institutions and the determination of institutions in a particular historical context, which if you know my work, you'll know I'm wont to do. So this is the inception of a project. I think there's just enough to talk about. But I think this is not the moment to rest on my laurels, so I'm going to go for it anyway. So this is the project with Daron, Daron Acemoglu—Daron and I shared the Nobel Prize. We've been working together for 30—we started talking about ideas 33 years ago—Leander Heldring at Northwestern, Steve Pincus, who's a professor in the history department here who unfortunately is giving lectures in Hong Kong this week purely by coincidence, and Santiago Torres, who's a student of mine who's at MIT at the moment.

So this talk is going to be about the determinants of institutions. So as the president was saying, a lot of our work has been trying to study the consequences of institutions for prosperity and flourishing. And part of that is an account of why institutions vary so much in the world and how come some countries end up with institutions that create prosperity and flourishing and other countries don't. So I think if you want to think about what the big picture is here, it's exactly about that variation in institutions. And you could say stepping back and thinking about social science broadly, there's three ways of thinking about why societies have the institutions that they do. One is it's to do with material interests. So institutions are rules in society that create incentives, that create opportunity, that allocate power, that allocate authority. And those institutions may allow the society to flourish, but they also have distinct distributional consequences. So different groups in society, different people may favor different sorts of institutions because of those distributional consequences, consequences of wealth or resources, and a theory of institutions would put emphasis on who likes what and how much power do they have to get what they like. So I'm going to call that the material interest perspective on institutions, which I've done a lot of work—Daron and I've done a lot of work on that.

Another view increasingly popular in economics emphasizes the cultural determinants of institutions. So I don't want to get into a big discussion of culture in 40 minutes. When I start talking about the context and the evidence, I'll be very concrete about what I mean. But in economics today, culture—we think of culture as a kind of evolutionary dynamic of people's preferences and values that characterize a society. And the focus today, which is the focus of a lot of our work at the moment, is on what I'm calling—what I'm going to call ideas. And by ideas, I'm going to mean normative justifications about the way to do things. So you could say those are in some sense cultural, but they don't move slowly like we think of culture as moving. And they're subject to innovation and entrepreneurship. So I think what I'm going to try to convince you today is that thinking about institutions, at least in this particular context, you have to think of not just material interests or maybe slow-moving culture but also the ideas people have about the institutions in society. What sort of institutions should we have? And there's contesting ideas about institutions, and that's what we're going to try to make some headway on. And I want to say here on this slide that social science, including our own work, has been dominated by this approach of material interests. So people think of institutions as being driven by the underlying material interests, which are relevant for particular sets of institutions. But by the end of the day today, I hope that I'll have shown you a little bit of evidence that in this one historical context we have to take people's ideas about institutions seriously. I'll be very concrete about that in a minute.

So I'm going to talk about 17th century England, and I'll try to justify why that is. But I just want to start moving into that mode by saying if we look at 17th century England, it's an exciting era and country to study, of course, because there's enormous intellectual ferment. There's huge institutional change. There's massive innovation in ideas. There's a lot of economic change. So everything is happening. You could even say there's cultural change. And if we look at famous political events, famous changes in political institutions such as the Civil War or the Glorious Revolution when James II, the King of England, was overthrown, parliamentary sovereignty established, that led to the emergence of many modern state institutions in Britain, material interests were clearly at stake. Here's George Jeffreys—Hanging George—he was called the hanging judge because he liked hanging people. And he also liked promoting the interests of the East India Company. So the East India Company was this monopoly. It was very much favored by King James II. They liked promoting the East India Company at the expense of other traders. So you could say you could look at this, the Glorious Revolution—here's this is George Jeffreys being arrested at Wapping shortly after James II left the country. You could say, of course, they want to get Jeffreys out of the way. They wanted to abolish the East India Company monopoly and create more opportunities for merchants. So that would be a kind of argument about you want to get rid of the king and you want to make this transition because it's going to distribute wealth and economic opportunities differently.

You could also think about the cultural commitments at play. This is just my favorite illustration of this. This is Charles I to go back to the Civil War. I'm going to talk about the whole century. So going back from the 1680s to the 1640s, here's Charles I. Now Charles I was a deeply religious man who really believed that God had made him king and not just king. He believed that God allowed him to heal people. So here he is laying on hands healing somebody. Now you might think, well, how seriously did he take that? Very seriously! So according to one historian who wrote a book about this, he did it 92,000 times. So that's—so he was a deeply religious man, so we want to think about religion and institutions as scholars have done. And finally, of course, the focus of the discussion is going to be about ideas. So when Charles I got into trouble being bankrupt and Scots were invading, parliament presented—John Pym, parliament presented this famous remonstrance, which was a list of all the grievances. 204, if I remember correctly, and involving institutional change—change in political institutions, constitutional change, new ideas for how to do things. So you see all these things at play—interests, culture, ideas.

I alluded to why this is so interesting, and the contribution of our research is really going to be to provide a methodology which will allow us to think about all these different forces—ideas, interests, and culture. I think interests and culture, we understand how to think about that empirically. The main contribution and the kind of conceptual challenges with respect to measuring ideas and trying to investigate the impact of ideas on people's actions and people's beliefs and people's attitudes. So I said why the 17th century. There's a lot of interesting stuff going on. There's political change. There's the Glorious Revolution. We have a republic. We had a republic for a little bit in England. We had this republic, and then they took it away again in 1660 and it never came back. We had the Glorious Revolution. Well, I talked about that. We have all this incredible ferment, intellectual ferment, philosophy. We have Hobbes. We have Locke. We have the diggers. We have the levelers. We have all sorts of ideas. As the great British historian Christopher Hill used this expression—well, he got it from a contemporary tract—the world was turned upside down. So that makes it a particularly interesting moment to look at, and, of course, so many of these ideas have had profound implications since. John Locke's ideas played an incredibly important role in the American Revolution and American constitutional process, for example, and there's huge economic changes going on at the same time—the colonization of the Americas, the Virginia Company, the colonization of the Caribbean, the founding of the Bank of England. So there's all sorts of stuff going on. So this is why look at 17th century England. Why not look at United States today? Well, you can apply the methodology to that.

So how to think about these events, if you look at the literature and to which we've contributed, you can see all these types of arguments. Karl Marx, Marx and Engels in The Communist Manifesto when they talk about this. That's a material—it's the rise of the bourgeoisie. There's economic change. Economic change redistributes power, resources. The bourgeois revolution they talked about 1688, the Glorious Revolution as a bourgeois revolution. We did empirical work with Simon Johnson on exactly that topic. Steve Pincus's great book on 1688 talks about the interests involved in that. Cultural explanations, also one of the most prominent explanations for both the Civil War and the Glorious Revolution due to the British historian John Morrill is it's really about religion actually. He says this is the last of the European wars of religion. It's really about the fact that Charles I and James II were devout Catholics, and the rest is—most of the rest of the people didn't want a Catholic on the throne of England. So it's a cultural struggle. What I'm going to emphasize—what we're emphasizing is actually the ideational conflict because at the same time there were interests—there was culture—there was also a lot of conflict over what's the right way to organize the state. What's the legitimate way to organize the state? In particular—and I'm going to zoom in today on a dichotomy between—there's lots of different ideas we could talk about the nature of the state and society and—but today I'm just going to focus on the most salient dichotomy between the divine right of kings. Charles I, James II argued that they ruled because they were chosen by God to rule. They had the right to rule because of the divine right of kings, and there was an alternative view which said, no, that's not right at all. There's no divine right of kings. God didn't make anyone king. The people choose the king. There's popular sovereignty, and the people have the right—have a duty almost to resist the abuse of the sovereign. So that's so-called resistance theory. So there's a clash of ideas. And what we're interested in in some sense is sorting out what role do these things play. What role do interests play? What role does culture play? What role do ideas play? My examples are meant to suggest probably all of these things are in there. I'm not dismissing any of them, but I think the contribution is to have a framework where you can compare the relative effect but also just to use this idea—bring ideas into the discussion because they're not in the discussion in social science.

So we're going to focus on that dichotomy as I've already said, and we're going to use for now two books as a placeholder. I can talk much more about that for these two different points of view. So John Locke's—I already mentioned John Locke's Second Treatise of Government. We're going to use that as a placeholder for popular sovereignty or the fact that people are able to resist the king if they want if the king is behaving badly or ruling badly, and we want to contrast that book, the arguments in that book, the ideas in that book to the ideas in the polar opposite book, which is by Sir Robert Filmer called Patriarcha. So Patriarcha was the go-to book for the people who wanted to justify the divine right of kings. So both of those books make lots of different arguments. So for the divine right of kings against popular sovereignty, against the divine right of kings for popular sovereignty, and what we're going to use—what we're going to do is we're going to extract those arguments from the book using techniques in AI. And then we're going to look at people's arguments. We're going to look at people's speeches, and we're going to look at what language they use and what arguments do they use. And we're going to try to look at the similarity between the arguments that parliamentarians use, the arguments that politicians use. And we ask and ask is that argument that Locke made, or is that argument that Filmer made? So the crux of the technical contribution is just trying to get at that similarity. That's a hard problem, and I'll talk a little bit about that. And then we want to look at what's the consequences for making those kind of arguments. If I make a Locke or a Lockean type of argument, does that make it more likely I'm going to vote against the king or oppose the king or I'm going to make a speech, the stance of which—I'll talk a little bit about how we code that—the stance of which is anti-monarchy or pro-monarchy? So those are the dependent variables, but the big contribution is to try to quantify this idea of similarity as a way of bringing ideas into the discussion, and then I can put that into an empirical framework where I can compare it to interests and culture and other things.

So as I said—I'll show you in a second some of the sources we're using—the technology that we've developed over the last few years can be used for anything. And we're—I'm just—for now, I'm going to use as I said Locke's second treatises. It's both treatises actually. It's not just a Second Treatise, but Locke's Two Treatises of Government and Filmer's Patriarcha as a—that's the kind of source of information. But we're expanding that as I said, and once you get the technology working, you can basically—there's lots of computational complications in terms of change in the use of language and is it written in Latin and et cetera. But I'm not going to get into that. So we want to compare similarities, people's—real people's similarities to the argument in Locke and Filmer. And then we have information on votes and support. I'll say a bit more about that in a second. And then we also can look at speeches and code what's the stance of this speech. Is this a pro-king speech or an anti-king speech? And then we have a lot of information from the history of parliament about people's economic interests, about religion, about all sorts of things. So what's the main finding so far? What I'll try to show you by the end is actually this method of quantifying ideas provides evidence, in fact, that ideas did play a role, an important role, in influencing the way people behaved. And so if I can convince you that there might be something there, then that—I'm happy today.

So the sources—let me not spend too much time on this—there's an enormous amount of stuff out there that we basically tried to digitize using OCR. Won't bore you with all the details, debates in the House of Commons. Of course, the period that we're looking at is before they have Hansard or the modern, very systematic way of recording what went on in parliament. So we have to digitize that. There's a lot of stuff already digitized that we can systematize also. So these tend to be much more informal parliamentary diaries, not necessarily official records of what went on in the House of Commons, but it's very useful for the purposes at hand. So I think we have the entire—we have the entire population of sources that we could have on this that we've digitized and we're using as the database. And then here's Patriarcha and Locke's Two Treatises we're going to use to distill these arguments. And I'll talk more—to try to talk more about that as I said. So, of course, these arguments about the divine right of kings and popular sovereignty, they don't start with any of these books, and, in fact, they go back into the 17th century—sorry, they go back into the previous century, into the 16th century, and probably much earlier. But one of the power we think of the technique we have is that we can actually identify. I don't have any results to show you today. We're still grappling with it, but we can actually identify innovations in ideas. So one of the things that we're very, very excited about is using the methodology to look at innovations like what's new in Filmer. What's new in Locke that wasn't in some previous book? So that's—we can—I'm not going to show you any results on that, but I think that's one of the ways where we're going to look at intellectual innovation. That seems to be one of the most exciting things. And then we can ask what's most important, the new ideas or the old ideas? Let me not talk about that. I talked about the history of parliament. This is just aspirational. Here's all the books we've digitized, but I'm not going to show you any results with that, but you'll get the idea. Once you see the methodology, you can see it's going to apply to anything.

So a lot of information comes from the history of parliament. This was a project started in the 1940s in Britain by Lewis Namier. I think it's still ongoing. Has fabulous information from the period we're looking at about the background of members of parliament, education, economic activities, occupation, religion, et cetera, et cetera. And here's the information we have on the—part of the information we have on the outcomes. So the column on the left are in black. They're real votes we have. So for example, 1689 after James II left England in the Glorious Revolution, parliament voted on whether or not he'd abdicated. So the idea was if he'd abdicated, then we can make William or we can put new king and queen on the throne and everything will be great. So his supporters said, no, no, he didn't abdicate. He hasn't—he's still the king. So they can define a clear for the king against the king. The bill of exclusion was a bill to keep James II out of the succession because he was a Catholic. So there were several attempts to stop him succeeding Charles II because he was a Catholic. The Civil War, we have this very famous division at the time of the start of the Civil War about who was in favor of the king, who was against the King. So we have some votes. Unfortunately, votes are often not recorded. So what we do have is from these parliamentary diaries and other sources lists of this person supported the king or this person opposed the king. So for today it doesn't really matter so far, but I'll just put all of those together. So think of vote support as being the same thing. But as I said, we're also going to analyze the speeches and look at the stance of the speech. Is the speech in favor of the king or against the king?

So here's the technical bit. What do we do—what do we do with this? Well, we set it up—we use ChatGPT basically. And we take—I'll show you an example in a minute. We take, let's say, Locke's Treatises of Government, and we feed Locke's Treatises of Government chapter by chapter or section by section, if the chapters get too long. And we asked ChatGPT, what question is being asked in this chapter? So tell us what question is being asked in this chapter and what arguments are being used to answer the question. So that sounds very straightforward, but actually it needs a lot of guidance. So there's a lot of testing, test sets that have to go on here, meaning you hand code a lot of stuff yourself and just make sure that it's getting something sensible. And then you go back to the drawing board and you keep working like that. But think about taking a chapter of Locke. Here it is. It's chapter 19 of Two Treatises of Government. Here's the text. I'm not going to go through it in detail, but we feed that into ChatGPT. And we asked it to classify what questions are being asked and what arguments are being used. So we do the same thing for speeches. We can feed the speeches, we can feed Filmer's book in. We can feed any book in there, you can see. So once we've got that, we ask to retrieve the questions. So here's a prompt. This is a ChatGPT prompt. Just for concreteness, it's only a fraction. The above is a chapter relating to the scope, validity, needs, or origins of the power of kings. What are the questions addressed in this chapter? There's more to it than that, but it generates a load of questions. There's a lot of questions. John Locke had lots of questions. Any chapter has lots of different questions. Here's one of the questions that comes out. "How does the concept of Popular Sovereignty challenge the Divine Right of Kings?" Great. So there's a question. What's the answer? So here's the answer. So it's a big thing. "The concept of Popular Sovereignty fundamentally challenged the Divine Right of Kings by asserting that—". I'm not going to get into it. So there's a question. There's an answer. It's a big, complicated answer. So what do we do with that? Well, and here's the trick is we want to break that into premises and conclusions.

So comparing arguments turns out to be—I was going to give you an example of if you just use ChatGPT to do this, it's really bad at comparing arguments. Now of course, we have some political philosophers in the audience, and I'm sure the political philosophers do a very good job at comparing arguments and seeing is this argument similar to Locke or is it not similar to Locke? I guess you could say in our defense, what we can do with this technology, we can do this at massive scale. So we can do this with thousands of books and thousands of arguments and thousands of speeches, and we can bring it all together. So of course, humans can do this. And humans have done this. And there's a lot of knowledge on this. And in some sense, that we can use in these test sets and in calibrating the technology. But I think we're able to do this at scale. But it's not trivial to do it at scale. And the kind of technology, the kind of methodological breakthrough here is in some sense, the problem with many of these large language models is that when you ask it to compare arguments, it takes account of all sorts of stuff that you don't care about, like syntax or tone or word choice. A lot of stuff from our perspective is kind of irrelevant. What you care about is the logic. So what we've done is to take an argument like that and break it into premises and conclusions, and then break up the conclusions using some ideas, some kind of logic into what's called conjunctive normal form, which is just representing the conclusions as like ands and ors, et cetera. I'm going to show you a concrete example in a minute. And so once we break up—let me show you the concrete example. So here's the argument that I showed you. On the left are the premises. I'm not going to start going through the premises. On the right is the conclusion. "The principle of Popular Sovereignty—". So we took that argument. We broke it up into premises and the conclusion. The conclusion is, "The principle of Popular Sovereignty asserts that the legitimacy of government—", blah, blah, blah. And then we can break it up into this CNF, this conjunctive normal form. "Popular Sovereignty establishes that government legitimacy derives from the consent of the governed, and popular sovereignty enables resistance to tyranny, and popular sovereignty enables the reclamation of rights." So the point about doing all of this is that this makes it much easier for these AI algorithms to compare what's similar and what's dissimilar, that once you've broken it into these premises and conclusions, but you represent the conclusions in this particular logical form, then it gets much more straightforward to define what similarity, and then by comparing premises and conclusions, we can create an index basically of similarity. Or we can create a kind of similarity score. So the trick is really to break the arguments down into the kind of logical components. And that makes it much more accurate the whole thing. And then we can construct this matrix of the similarity of different types of arguments.

All right. So I'm not sure how much time I have left. Somebody is going to wave 10 minutes at me or five minutes. They haven't waved it at me yet. Where is the lady? She's there.

All right. Great. So with all of that, I wasn't—I'm either going to go into too much detail or insufficient detail presenting something like that. So what do we want to do? Well, so now we have—let me just collect my thoughts and say, what we're doing with all of this mechanism, what we're doing with all of this here is to say, let's take an argument from John Locke, OK? And we're going to boil it down into these component parts. And now I'm going to take some speech by somebody, the Earl of Clarendon. And the Earl of Clarendon makes an argument about John Locke or not John Locke, but a Lockean type of argument. About what? Well, the government comes from an original contract between the people. The government is established by an original contract, by the sovereignty of the people. The government does not come from God or something. So you can look at—that's not actually an argument that the Earl of Clarendon would have made. But anyway, hypothetically, and Lewis Namier, by the way, would really have disapproved of the whole exercise. So just to remind ourselves where we are here before I start running regressions, doing the statistical analysis, is we're looking at a particular person. That particular person makes arguments. We do the same thing we did with Locke and Filmer. We break down the arguments into different parts. And now we can compare the arguments to Locke and Filmer. How similar is that specific argument that the Earl of Clarendon made to these arguments that Locke made? Or how similar is that argument he made to an argument that Filmer made, filtered into this particular set of arguments. Of course, Locke made all sorts of arguments. Locke made arguments about the origin of property rights and the origin of wealth. And so that's interesting. But that's not the focus of what we're doing today, so filtered to look at these issues of constitutional, what's the right constitutional government? Is it about divine right of kings? Is it about popular sovereignty? So now we can take an individual and we can look at what they say. We can look at their speeches. And we can say, how similar is that to something Locke said? Or how similar is that to something Filmer said? And we can create a data set where we know that individual's behavior, how they voted on a particular measure or a particular division, who did they support, or a speech they made, was that against the King or in favor of the King? And we can look at the influence of different factors on that person's behavior. We can look at their economic interests. We can look in a very crude way at culture—a lot more to say about that. And we can look at oh, they voted against the King. Did they make arguments in their speech, which were similar to Locke's or similar to Filmer, or neither? So with that, we can go to the statistics.

So we're interested in estimating a kind of simple model like this. On the left-hand side is do I vote in favor of the King? Do I vote against the King? So that's just coded 1 minus 1. You could do that in different ways. And it's not just vote, it's also support. Remember I said, sometimes we have these votes. Other times we don't really have the votes. We have lists of people who supported the King or opposed the King. So think of vote as being really vote or support. And then we're going to try to understand, we're going to try to model the variation in that by looking at the similarity between the arguments that a particular member of parliament made and the arguments of Locke or Filmer. But again, think of that as a placeholder. We can bring a lot more material into this in the future and we will. So we can control for people's characteristics. Were they a merchant? What were their economic interests? We can control for whether they are Catholic. I sort of alluded to the fact that one of the big cleavages—as John Morrill said, this was the last war of European religion was this conflict between the Stuarts, who were Catholics, and the Parliamentarians, some of whom were Catholic, but many of whom were Protestant. So an argument about the Glorious Revolution, for example, the timing of the Glorious Revolution would be, well, James II had a son. And suddenly there was a son, and there was a Catholic heir, and all hell broke loose, if you'll pardon my French. So just this is the basic sort of exercise.

And you might say there's a lot of issues here to think through that perhaps I shouldn't go into details, but someone who gives a speech makes many arguments. And so there could be several arguments in John Locke that were similar to the arguments that he made. So how do you deal with that? Well, you could aggregate that in different ways, which you could average across the similarities or something like that. So maybe that's too in the weeds. So maybe this is even more in the weeds. So I don't know if this is—I always say, I apologize with students, and I always say, I don't know if this is a regression audience. I think I did that last week. I don't know if this is a regression audience or not, but let me just talk you through this. And so we're trying to explain the propensity of people to vote against the King or oppose the King. And what this suggests is that there is a positive correlation between the extent to which members of the parliament made arguments similar to Locke. So the more similar arguments they made to Locke, the greater their propensity to vote against the King or oppose the King. The negative coefficients on similarity to Filmer means the more they made arguments that were similar to Filmer, the more the propensity to support the King. So that's what it means. And then the columns are just let's look at what happens when we start controlling for economic interests. I'm just calling that merchant. We can do all sorts of things. Merchant, why merchant? Well, I mentioned the East India Company earlier on. If you read Steve's great book, 1688, chapter 3, which is called "Going Dutch," is all about the economic factors leading up to the Glorious Revolution. So a lot of the contest was over global mercantile expansion, these restrictions on trade. So a big hypothesis is that there's a huge cleavage between people who are merchants, who have particular mercantile economic interests, and they were threatened by the East India Company. What I want you to take away from this—we're still working on this. We also have whether or not they were Catholic. What's interesting about this so far is we don't find this is done very crudely. So I don't want to get too excited, but we don't find this matters very much at all. In fact, the estimates of the kind of ideational factors, meaning, do you make these arguments that are similar or similar to Filmer, don't seem to be correlated at all with economic interests and your religious commitments.

So that's the first table. Now, just in case this wasn't a regression audience, we had some pictures. And so this is just the kind of partial relationship between—on the horizontal axis is just this similarity score, other things equal. And the vertical axis is you could think of that as your propensity to vote against the King. So you can see the more similar you are on the left panel, the more similar you are to Locke, the greater your propensity to oppose the King. And on the right panel, the line slopes in the other way. That's the partial relationship between your similarity to Filmer and your propensity to support the King. And just to say what I was saying with the table is what we find is that this relationship between these similarity scores and your behavior doesn't seem to be influenced by your economic interests or your religious commitments. But again, there's a lot more work to do on that topic. And you see that visually here, which is if I start controlling for that, statistically this is whether you're a merchant, you have a background as a merchant, you see nothing really happens. And if you're a Catholic, again, nothing really happens. So that's just to show that the partial relationships don't seem to be influenced by your economic interests or your kind of religious commitments measured in this way. Of course, there's a lot more to culture. Why am I talking about religion? Well, because so much of the literature on 17th century English politics has talked about it in terms of a kind of religious conflict, which it was in many ways as well. So again, these results are very stark in the sense that they emphasize these ideational factors. But I started with those examples to say, well, probably all of these things are relevant. So that's looking at just votes and support.

We can also look at this at the level of the speeches. So we have a lot more speeches than we do votes and divisions in the House of Commons. So there's a lot more variation in the speeches. So that's kind of exciting to play with from a statistical point of view. And also, if we look at, say, votes, we don't observe the same person at many instances over time. But speeches, English people like to talk a lot. And English politicians like to really talk a lot. So we have a lot more speeches made by the same person. So that's very exciting, because we can look at how they change their opinions over time. So this regression allows us to look at that, meaning as they started making arguments that were more similar to Locke over time, they became more likely to oppose the King. So looking at that, what I would call the within variation is sort of appealing for lots of reasons that I won't go into. So hopefully you get the idea. And we find exactly that. So when you're able, looking at the speech level here, so what's the outcome? The outcome, remember, is we have to use a large language model to decide, is this speech pro-King or anti-King? So I guess if you're the president of a major university, you get very good at identifying, is this a pro-president speech or an anti-president speech?

There you go. So it's easy to code. Maybe we didn't need any highfalutin technology to code that. But sorry, couldn't resist that. Just to remind you what the dependent variable here is. So you see very similar things. So as people articulate arguments that are more similar to Locke, they get more likely to be making speech, which is anti-King. And the opposite is true with similarity to Filmer.

So I'm going to conclude now. So now I'm feeling I should have spent more time on the nerdy bits. But anyway, OK. So what are we trying to achieve here? I think at some general level, everyone sort of agrees that ideas, ideology—I thought about the word "ideology" somehow has a pejorative connotation in some contexts. So we decided to talk about ideas. I think most people seem to agree that ideas matter for human motivation and action. And in fact, if you read an undergraduate comparative politics textbook, it says, comparative politics is all about the three I's—ideas, interests, and institutions, and how those things interact. But after page one or page two, ideas disappear from the book, and it's all about interests and institutions. So that's pretty out of sync, in fact, with the way people actually discuss these things in real life. I'd find it very difficult to discuss what was going on in the United States today without thinking about ideas, frankly. Yes, we're thinking about institutions, and we're thinking about what institutions are kind of equilibria and how fragile are the institutions and what's going to happen with the institutions. But we're also thinking about interests, and we're thinking about ideas as well. So I think ideational arguments have really not been integrated into modern social science. I don't know why that is. I think one, you could think of different hypotheses. One hypothesis would be Marxism. Traditional Marxism was extremely hostile to ideational types of arguments. Marx had his idea of the substructure and the superstructure. And ideas were kind of froth, which they came in to justify what was happening, but they didn't really play an important, autonomous role in driving things. I think in this context, and that's one of the reasons why we like this context, it seems pretty plausible. And many historians and political philosophers and scholars have kind of made specific arguments that the ideas are important. And that's another reason why we've focused on this. But obviously you can think of other examples like the French Revolution, or whatever. And maybe there's different reasons. And I think a clear reason is just the difficulty of measuring ideas, measuring them, and conceptualizing their causal impact on action. And I think what's so powerful about all these modern AI methodologies is they give you a fantastic set of tools for actually measuring these things in a new way. And so I think we just took a few steps in that direction today. But I think it's an incredibly exciting area for young people and even for old people to be working in.

Now we have questions. Somebody else is in charge. I don't like being in charge.

Yeah. I mean, I think the traditional view would be yes, it mattered. The religious convictions of people mattered. And a lot of you could say a lot of these cleavages would be over this act of exclusion, for example, trying to stop James II becoming king, you could say because some people didn't want that because they thought he was a closet Catholic or he was an overt Catholic. But there's variation of opinion. Some people genuinely thought, OK, he's a Catholic. We don't really want a Catholic king, but he's the legitimate monarch.

No, I think you have a mixture. You have a mixture. And there's variation. As I'm saying, I think these ideational issues cut across religion also, meaning it's not just there were Catholics here and there were the Protestants there. If that were true, you'd find something very strong in those regressions, which we don't find. I think that suggests these other things are cutting across that. And in particular, there were Protestants who believed that he really was the king, that God really had chosen him to be king. And so I might not like him and I might not like his policies, but he's the legitimate king, and we just have to put up with it.

Yes. So absolutely, yes, it could. And all of these things are interrelated at some level. Obviously, the economic conditions could stimulate ideas. If you picked up a book about Confucian thought, they'd say, why did Confucius have all those ideas when he did? Well, it was the Warring States period, and central authority had fallen apart in China, and people were looking for different models of governance. And so I think these things are all interrelated. That's why social science is so difficult. I don't want to get into that today. But what I do want to get into is that can be true, but it can also be true that ideas have the autonomous impact that I'm suggesting here.

Can I pass on that one? I don't know. That's a good question. I think it's a good question. If I look at the world—Daron and I have a sort of reading list of philosophical works. That might be on the reading list. There's a very interesting book by a political philosopher, I think he's called Rosen, Michael Rosen, I believe, which is about conservative philosophy, basically, starting with Oswald Spengler and going through for the last 100 years. And I would say, if you read that book, if you send me an email, I can give you the citation. It's remarkable how many of anti-liberal philosophy, if you like—it's remarkable how many ideas in that book are flying about at the moment.