Subscribe
Share
Share
Embed
Is war in long-term decline? Steven Pinker's The Better Angels of Our Nature brought this previously obscure academic question to the centre of public debate, and pointed to rates of death in war to argue energetically that war is on the way out.
But that idea divides war scholars and statisticians, and so Better Angels has prompted a spirited debate, with datasets and statistical analyses exchanged back and forth year after year. The lack of consensus has left a somewhat bewildered public (including host Rob Wiblin) unsure quite what to believe.
Today's guest, professor in political science Bear Braumoeller, is one of the scholars who believes we lack convincing evidence that warlikeness is in long-term decline. He collected the analysis that led him to that conclusion in his 2019 book, Only the Dead: The Persistence of War in the Modern Age.
Links to learn more, summary and full transcript.
The question is of great practical importance. The US and PRC are entering a period of renewed great power competition, with Taiwan as a potential trigger for war, and Russia is once more invading and attempting to annex the territory of its neighbours.
If war has been going out of fashion since the start of the Enlightenment, we might console ourselves that however nerve-wracking these present circumstances may feel, modern culture will throw up powerful barriers to another world war. But if we're as war-prone as we ever have been, one need only inspect the record of the 20th century to recoil in horror at what might await us in the 21st.
Bear argues that the second reaction is the appropriate one. The world has gone up in flames many times through history, with roughly 0.5% of the population dying in the Napoleonic Wars, 1% in World War I, 3% in World War II, and perhaps 10% during the Mongol conquests. And with no reason to think similar catastrophes are any less likely today, complacency could lead us to sleepwalk into disaster.
He gets to this conclusion primarily by analysing the datasets of the decades-old Correlates of War project, which aspires to track all interstate conflicts and battlefield deaths since 1815. In Only the Dead, he chops up and inspects this data dozens of different ways, to test if there are any shifts over time which seem larger than what could be explained by chance variation alone.
In a nutshell, Bear simply finds no general trend in either direction from 1815 through today. It seems like, as philosopher George Santayana lamented in 1922, "only the dead have seen the end of war".
In today's conversation, Bear and Rob discuss all of the above in more detail than even a usual 80,000 Hours podcast episode, as well as:
• Why haven't modern ideas about the immorality of violence led to the decline of war, when it's such a natural thing to expect?
• What would Bear's critics say in response to all this?
• What do the optimists get right?
• How does one do proper statistical tests for events that are clumped together, like war deaths?
• Why are deaths in war so concentrated in a handful of the most extreme events?
• Did the ideas of the Enlightenment promote nonviolence, on balance?
• Were early states more or less violent than groups of hunter-gatherers?
• If Bear is right, what can be done?
• How did the 'Concert of Europe' or 'Bismarckian system' maintain peace in the 19th century?
• Which wars are remarkable but largely unknown?
Get this episode by subscribing to our podcast on the world’s most pressing problems and how to solve them: type 80,000 Hours into your podcasting app.
Producer: Keiran Harris
Audio mastering: Ryan Kessler
Transcriptions: Katy Moore
Unusually in-depth conversations about the world's most pressing problems and what you can do to solve them.
Subscribe by searching for '80000 Hours' wherever you get podcasts.
Hosted by Rob Wiblin and Luisa Rodriguez.
Rob Wiblin: Hi listeners, this is The 80,000 Hours Podcast, where we have unusually in-depth conversations about the world’s most pressing problems, what you can do to solve them, and whether you’re more likely to have a heart attack if you live in a county or a state. I’m Rob Wiblin, Head of Research at 80,000 Hours.
Is war on the decline?
Getting the answer to this right really matters.
The US and the PRC are for better or worse entering a period of renewed great power competition, with Taiwan as an obvious possible trigger for war between them, and Russia is once again invading and attempting to annex territory from its neighbours, something that has a very 19th century feel to it.
If war has indeed been gradually going out of fashion for the last few hundred years, we might reassure ourselves that however troubling the current situation feels, fingers crossed modern culture should continue to throw up power barriers to another global war like those of the past.
But if we’re as war prone as ever, one need only inspect the record of the 20th century or the 19th century, or the 18th century, to feel a suitable level of fear about what might be in store for us in the 21st century.
In 2019 political science professor Bear Braumoeller wrote a book analysing the question of whether war is in long-term decline, titled Only the Dead: The Persistence of War in the Modern Age.
It was in part a response to Steven Pinker’s book The Better Angels of Our Nature, and other optimists who think that modern ways of reasoning and learning about the world, and modern moral attitudes, are gradually reducing the proclivity of states to launch wars.
As you’ll hear, unfortunately Bear does not agree with this.
Given this is a live controversy with smart people on either side, I approached this episode wanting to get into the weeds in order to help me figure out what I actually think about this question personally.
That means that when we turn to Bear’s data analysis, we really get into the nuts and bolts — things like how exactly the number of wars is coded when there are more than two countries involved, how he corrects for the fact that his graphs test many different hypotheses at once, and the results Bear got for at least four different plausible measures of warlikeness.
This is the best discussion of these specifics that I’m aware of, so I’m really proud of what we’ve managed to put together here.
But that is on the challenging end, so the show’s producer Keiran did a little reorganisation of the conversation to put the most engaging and general interest topics first.
So first up, we’ll give a short summary of the conclusions of Bear’s data analysis in Only the Dead.
But we’ll then press pause on that, and discuss the ideas and values of the enlightenment, and whether we should naturally have expected them to reduce rates of war since 1700.
We then push on to what Bear thinks is the real primary driver of rates of conflict, which political scientists call ‘international orders.’ To illustrate that, we go through two international orders from the 19th century which I knew very little about before reading his book — the Concert of Europe and the Bismarckian system. It’s great stuff if like me you enjoy history.
We then talk about what can be done to reduce the risk of war today if Bear is right that the risk remains troublingly high, and it’s the nature of international orders that do the most to determine their prevalence in any given period.
We then loop back and dive deep into Only the Dead, discussing the datasets he used, the specific results he got for different measures of warlikeness, the power law for deaths in war, the biggest weakness of the analysis, and quite a few others things.
I hope that section can provide a lot of clarity for people who, like me, have found it very confusing how much disagreement there seems to be on this issue.
All right, without further ado, I bring you Bear Braumoeller.
Today I’m speaking with Bear Braumoeller. Bear is a computational social scientist and a professor in the Department of Political Science at Ohio State University, where in 2020 he founded the social science research lab known as MESO, which stands for Modeling Emergent Social Order. He’s been studying issues of war and peace for over 30 years. In 2013, he published The Great Powers and the International System, and in 2019, he published Only the Dead: The Persistence of War in the Modern Age, which is the topic of today’s conversation.
Thanks for coming on the podcast, Bear.
Bear Braumoeller: Thank you so much. I’m very glad to be here.
Rob Wiblin: I hope we’re going to talk about whether war is in decline and what factors most influence its prevalence. But first, what are you working on at the moment, and why do you think it’s important?
Bear Braumoeller: What am I working on at the moment? That’s a few things.
First of all, with my research lab, which you just mentioned, I’m working on studying the relationship between international order and international conflict. It’s important in part because preliminary results show that there’s a big relationship between order and war, but there aren’t very many studies, and there isn’t very much theory. And so as a result, it’s difficult to translate that relationship into anything approximating actionable foreign policy advice, for example.
Second, I’m here now in Oslo, Norway at the Centre for Advanced Study with a group of people who are spending the years studying the causes of escalation and warfare. That’s important because that was something else that came out of the book: I realised that some wars get really, really big and other wars don’t, and the extent to which wars can escalate is really shocking. You read the book, so you know that. At the end of chapter 5, I was trying to find a silver lining and say something optimistic, and I just sort of threw in the towel. I’m like, “No, this is really bad. It sucks. There’s nothing good to say about escalation.”
But one thing that a few of us have realised is it’s just incredibly understudied. We study war initiation; we don’t study escalation. So there’s a group of statisticians and social scientists here who are getting together for an entire year to hash out our differences, in some cases, and try to get a better handle on why it is that some wars get really big and others don’t.
Rob Wiblin: Yeah, fantastic. It seems like your research project is taking all of the things that you weren’t sure about at the end of Only the Dead and trying to answer them. And some of them are very difficult questions, so you need a whole team in a research lab just to make some progress on it.
Bear Braumoeller: Right, exactly. It’s accumulative research. It’s funny, I used to tell people that I study the academic equivalent of bright shiny objects, because I would just study whatever I was drawn to. But yeah, this is definitely an extension of the work that I did in Only the Dead.
Rob Wiblin: Yeah. Well, let’s talk about Only the Dead. What prompted you to write the book?
Bear Braumoeller: Well, it’s sort of funny. I wouldn’t call it statistical analysis, but I had played around with the data about trends in warfare before, but only in the context of teaching. When I teach classes on war, one of the things that I tell my students is, “You may not know this, but there are data on wars, and people actually analyse data on wars, and you can get some interesting findings out of them.” And so during the first lecture, I would just talk a little bit about some of the things that we can try to figure out using data on wars. And one of them was: are we seeing a rise or decline in conflict? I mostly showed them why that was a hard problem to answer, rather than trying to come up with a comprehensive answer.
But the funny thing is that it’s not the way that political scientists study war. We try to look for primarily things that cause war or ways to predict war. We never really look at questions like “Is war on the decline?” And so I never really thought of that as a very interesting exercise. But then Steven Pinker’s book came out, and I realised people are really interested in this question. And so I thought, “I have to get this book. I have to take a look and see what it says.” And I got it, and I started reading it, and I thought, “Wow, I am really not satisfied with these answers.” Not because of what the answers were, but because of the way in which he arrived at them.
So this started out as a paper. I wrote a conference paper. This is what academics do when we get annoyed about something: we sit down at the computer and start typing away. So I started writing a paper. I invited Steven Pinker to an Author Meets Critics panel, which ended up being cancelled. So I just kept working on the paper, and it kept getting longer and longer.
And at one point I was fortunate enough to have a conversation with Dave McBride, who is the editor at Oxford University Press. And I said, “I’ve got this paper that I’m wrangling, and I think I might be able to turn it into a book — or I might have to, if I want to say everything that I want to say.” And he had just read Better Angels of Our Nature. He had some of the same reactions I did. He was over-the-moon excited about having me write this book.
And this is something that not a lot of people know: academic books do not get marketed the same way that trade books do. They’re just completely different streams.
Rob Wiblin: Do they get marketed at all?
Bear Braumoeller: Mostly to university libraries, and directly to other academics. So my first book, I naively expected to be able to walk into the local Barnes & Noble and find it on the shelf. And I was sorely disappointed. And so I talked with Dave, and he said, “If you think this debate deserves to be in the public sphere, we can market this as a trade book. The challenge is you have to write it in plain English. You can’t write it like an academic would.”
Now, this was actually a fun opportunity for me, because before I became a political scientist, I wanted to be a writer. I actually very much enjoy writing, and I’m painfully aware that social science has done terrible things to my writing. So this was a great opportunity to try to undo some of those things and write for a general audience. So I thought that this sounded perfect.
Rob Wiblin: Yeah. Did the book get as much public attention as you were hoping? I mean, it’s a super hot topic. As I’ve learned preparing for this, it’s a controversy in academia, but it’s a controversy in academia that has definitely spilled over into the public sphere.
Bear Braumoeller: Right. It did, and it didn’t. So COVID definitely refocused people’s attention. This was something that I think publishers noticed across the board: that book sales were down in general.
Rob Wiblin: I see.
Bear Braumoeller: On the other hand, Ukraine has sparked more interest in the topic, I think. But definitely, it’s night and day compared to a regular academic publication, and it certainly seems to have gotten a lot of people talking, which is great.
Rob Wiblin: Yeah, I think you’ve done an excellent job explaining really quite difficult social science analysis. In a way that, if you had no knowledge or no experience with these issues, it would be heavy going. But if you do have some idea about statistical analysis, or you have some preexisting interest in social science, I think you should definitely be able to follow it, and you’ll just learn an enormous amount reading the book. So if anyone does enjoy this interview, I can definitely recommend going away and reading it. Before we dive into the substance, where does the title come from? Only the Dead.
Bear Braumoeller: This is a quote from George Santayana, who’s a philosopher who is mainly known for his quips, I think. And one of the things that he said was, “Only the dead have seen the end of war.”
Rob Wiblin: I see. OK.
Bear Braumoeller: And that’s funny, I don’t know why, but in the military, that’s attributed to Plato a lot. And I didn’t want to screw it up when I put the attribution in the book. So I did some double checking and there’s no evidence that Plato ever said that.
But the reason for the title: I thought Better Angels was a fantastic title. I really like that part of the book. And I wanted something with an equivalent kind of literary pedigree that would convey the opposite message. At the same time, as you saw going through the book, I don’t want to be confrontational. I just want to try to answer this question. So there’s a book that came out I think last year called The Darker Angels of Our Nature by a group of historians. And that was a title that crossed my mind, and I just let it cross. I thought this is much more about the question of whether war is in decline than it is about Steven Pinker per se.
Rob Wiblin: Yeah, absolutely. It might seem like we’re picking on Steven Pinker a little bit as this goes on, or focusing on him a little bit — but I suppose that’s the consequence of being enormously successful, and getting your book out and your message out so much: you become the standard bearer, I suppose, for a set of views that are actually pretty mainstream already.
Rob Wiblin: So, the many findings in the book are presented, as I was saying, with substantial complexity and subtlety. And it’s going to take a while to explain many of your concerns with past research, then lay out your methodology, and then work through the various different tests and specifications that you provide us with. But so the audience can’t miss it, is it possible to summarise in a few minutes the key bottom lines that you really wanted people who read the book to remember and come away with?
Bear Braumoeller: Yeah, I wish there was just one bottom line, but there are a few. The first one is that there are a lot of reasons to be sceptical about the data and arguments that have been presented in favour of the decline of war argument. And this was a big part of my frustration in reading Better Angels in particular. I was looking forward to these nuggets of data, and the statistical tests, and so on and so forth, and I just didn’t see that.
Some other scholars have accused Pinker of cherry-picking the data. I didn’t do that in the book, but I’ve read that criticism, and I don’t think it’s completely unfair. Just for one example, he provided a graph of war in decline from a conference paper that never got published, and I dug up the conference paper and it was one of six graphs, and it was the only one that showed war in decline.
Rob Wiblin: It doesn’t seem fully evenhanded.
Bear Braumoeller: That’s one issue. Another is that the analysis of the data is really lacking. And this is a point that Nassim Taleb has made often and at length in particular: that there are no actual tests of the argument, in the sense of a formal statistical test that would help you distinguish signal from noise. That’s what statistical tests do. So it looks a lot more impressive, I think, to people who don’t do data analytics than it actually is to the people who do. That’s one bottom line.
Rob Wiblin: Right. OK, what are some other key messages?
Bear Braumoeller: The second bottom line is that when you do test the argument that war is in decline, it’s really hard to find support for it. And I was clear about this in the book: I wanted Pinker to be right. My hope was that I was going to find trends that were going to support the argument. But I looked at three different meanings of the phrase “decline of war,” and I only found a decline in one of them, and that was only at the end of the Cold War. Mostly, there’s just been no change over the past couple hundred years. And you could even argue that there’d been some evidence of increase in warfare prior to the end of the Cold War. So that’s number two.
Rob Wiblin: I guess the key scary finding is just that there’s not really any evidence that would make us feel confident that the threat that we face from war today is materially less than the threat that people have faced at many other times through history — times in which enormous, horrific wars broke out, and war was one of the most regularly destructive forces out there.
Bear Braumoeller: Correct.
Rob Wiblin: So we just shouldn’t be sanguine about the issue, I suppose is the bottom line. And what’s another key conclusion?
Bear Braumoeller: I think the third bottom line is that there is some hope. I don’t mean it to be an entirely negative conclusion. As you mentioned, patterns of international order I think are much better predictors of variation in international conflict than human progress over time. You don’t see a tonne of variation in conflict across the entire system over time. You do see variation across groups of states, and specifically across and within international orders — like the liberal international order, the Soviet communist order, the Concert of Europe, and so on.
So I think that in some ways I kind of do agree with people who argue that we’ve made progress. I think we’ve created something really impressive. Michael Howard wrote a book about international order where he referred to it as “the invention of peace.” So I think there is some hope if we can get a better handle on international order. The thing is we don’t really know how to use it. We don’t have a very good sense of how to optimise configurations of international order to create as much peace as possible.
Rob Wiblin: Fantastic. OK so let’s park the data analysis in Only The Dead for now. Listeners can rest assured that we’re going to come back and discuss that in plenty of detail in an hour or so, as I have a lot of questions about its various strengths and weaknesses.
But for now let’s push on to a different issue that you cover in the book, which is the age of Enlightenment, and reason, pacifist values and humanism and so on, and what impact we might expect that to have had.
I’m really interested in this issue, because I think I probably had a strong preconception that the Enlightenment — and all of the intellectual advances that stemmed from that over the last 250 years — that that probably was pushing us towards a more reasonable world, in which conflict was less likely. That certainly would have been my assumption, coming in. I’ve changed my mind on that over the last few years, or at least I view it now as a lot less clear. And I feel like I had all the knowledge necessary to realise that that was a much less clear claim than what I thought it was.
So there are ways in which intellectual changes in the 17th and 18th centuries might push us towards peace, but there are also ways in which they might not — and in fact, did not. Can you lay out some of those?
Bear Braumoeller: Sure. The word “enlightenment” just has such positive connotations. It’s hard to think…
Rob Wiblin: “I became enlightened and then went to war.”
Bear Braumoeller: You tend to attribute all good things that came after the Enlightenment to some degree of enlightenment. But you’re exactly right that the Enlightenment gave rise to a lot of ideas, and some of those ideas made war more acceptable for new reasons that hadn’t existed before. Herder and nationalism, for example. Hegel and Marx on the value of a strong state. Justifications for uprising against monarchies, the invention of wars for peace in the Greek Civil War. Interventions in order to stop war. All of these are grounded, very firmly, in Enlightenment ideas. Everyone loves to talk about Emmanuel Kant, but Kant wasn’t the whole Enlightenment. So that’s one general point.
The other one is Enlightenment ideas, even when they have been sort of progressive and positive, have given rise to very illiberal reactions on the part of people who were dissatisfied with modernity. The recent waves of populist movements, for example, is a good illustration of that. These are fundamentally nationalistic, illiberal movements. So there are a couple of ways in which you can’t really draw a straight arrow from the Enlightenment to peace.
Rob Wiblin: Right. So yeah, this is a big, big shift in my thinking here. I was thinking about it: What is the Enlightenment? What is it fair to say is a consequence of the Enlightenment? Is it kind of unclear? It’s obviously a vague concept. It’s a difficult, undefined question.
Bear Braumoeller: Now, again, the caveat is I’m not an Enlightenment specialist, so take it with a grain of salt. But you know, you see much more of a fundamental reliance on reason as a vector to truth than revelation or faith. And it’s a fundamental reorientation of society. It had a tremendous, just incredibly pervasive impact, the more it spread. There were actually a variety of different enlightenments in different places at different times, but the overall trend was more toward a focus on better ways to truth. And you see that in the rise of universities, the rise of science, all sorts of different ways. But reason can lead you to some pretty dark places too.
Rob Wiblin: Yes. And even if it’s leading you in a good direction in the long term, it could be an extremely bloody path to get there, arguably.
Bear Braumoeller: Yes.
Rob Wiblin: So if you construe the Enlightenment that broadly — that it’s a turn towards reasoned argument as a way of assessing ideas and trying to arrive at truth — it’s almost easier to ask what in the present world isn’t influenced or isn’t caused by the Enlightenment than to say what is.
So obviously if anything is caused by the Enlightenment, surely the French Revolution qualifies. The French Revolution was not exactly super peaceful. Then you’ve got the anticolonial wars and various wars of independence fought by various different groups — the United States then Latin America, Haiti, then many other countries — that are not exactly super peaceful.
Then you’ve got communism. Definitely the revolutions were deadly. The wars fought by communist states to foster communism elsewhere were very deadly. Communism, good or bad, it was not necessarily a peaceful process. You’ve got all the antimonarchist wars, which was a massive ongoing turmoil in the 19th century, deciding how countries should be governed. Again, maybe antimonarchism was good, but it caused a lot of wars.
Arguably, the Enlightenment justified a whole lot of colonialism that otherwise people might not have been interested in. Of course, countries were engaged in colonialism for other reasons before, but the Enlightenment provided new possible intellectual ammunition that people could use to dominate other people claiming that they’re doing the right thing and supposedly using reason to justify that.
Then plausibly you could say all of the new weapons that we invented, maybe that’s caused by the Enlightenment in a sense — because science has, just as it’s enabled medicine, also enabled nuclear weapons. Then you’ve got nationalism. Now we already had nationalism, but there’s kind of new flavours, new styles of nationalism that are justified on Enlightenment grounds.
I think maybe the one thing you could say is that it’s a stretch to say, “I think that fascism or Nazism is a kind of enlightened” — of all of the modern ideas, it’s one of the least Enlightenment-influenced, but at least it’s a reaction against Enlightenment ideas, you could say. So it’s still, in a sense, a broader consequence.
So, wow. The Enlightenment has just caused so much, that to say that it has in general fostered pacifism and peace… I haven’t even mentioned, of course, now people have all kinds of ideas about humanitarian interventions and going to wars in order to protect people, in order to promote liberalism, in order to promote ideas that they think are good. If you put that to someone in 1400 — that they should go to war overseas in order to promote liberalism, or in order to protect people from being oppressed by their own government — that would’ve been regarded as ludicrous.
Bear Braumoeller: Right.
Rob Wiblin: OK, so that’s a very long rant, but suffice it to say it’s a complicated picture. I think maybe you could make a stronger case that democratic liberalism has tended towards pacifism and antiviolence. But again, it’s a complicated picture even there, although I might rate my chances more at making that case.
Bear Braumoeller: Yeah, and it also gets more complicated when you look into the relationship between parts and wholes. So certainly there have been arguments that liberal democracy is conducive to peace. One of the things that I find in a later chapter is that the Soviet communist order was also relatively peaceful under threat of force. So it was not exactly peace in the fullest sense, but it was the absence of conflict, right? And certainly the way that Lenin and Marx initially envisioned a communist state, peace would have been a big part of the goal. Lenin in particular argued that the Enlightenment led to liberalism, which led to capitalism, which led to imperialism, which was a major driver of war — and that communism was the solution to that. And you could make those arguments. But it turns out, when you put capitalism and communism in the same international system, that is not a recipe for peace.
So it’s complicated in many ways. In the book I discuss the Enlightenment a little bit. I don’t say much more than that. It’s a lot more complicated than this. You gave a fantastic recitation of all the different kinds of things that have followed from the Enlightenment, and I thought about how do you aggregate that into a big overall plus or minus? I don’t know. I have no idea. But it’s way more complicated than just, “We need more Enlightenment. We’ve got to double down on the Enlightenment.” On everything that came out of the Enlightenment? I don’t think you want to do that.
Rob Wiblin: Yeah, I want to make that joke: “Has the Enlightenment led to peace? It’s too soon to tell.”
Bear Braumoeller: Right, right.
Rob Wiblin: I followed really closely this pro-Enlightenment or uncertain-about-the-Enlightenment discourse. I imagine that a lot of what’s going on is people talking past one another. The Enlightenment is a very broad term. People mean different things. One group, the pro-Enlightenment group, is probably talking about a particular single strain of thought, or a class of thought, that has emerged from the Enlightenment among many others, and arguing that that style of reasoning leads to peace. And probably there they have a strong case.
Whereas the people who are against it are saying, “What about all these other things? What about colonialism and the reaction against that? What about communism and all of the wars that resulted from that?” If you consider it very broadly, then the effect is so broad, the question starts to seem almost silly in a way.
Bear Braumoeller: But there is sort of a cottage industry in big think history books, and people arguing that modernity is good, and modernity is bad, and we really were happier when we were hunter-gatherers, and so on and so forth. So it’s one of those things that people love to weigh in on. I envy them the ability to think broadly enough to put all those pieces together and come to an aggregate judgement.
Rob Wiblin: Yeah. Do you have a take on democratic peace theory? This idea that democracies tend to be very hesitant to get into conflict with another clear democracy?
Bear Braumoeller: Yes, a fairly complicated one. Well, first of all, there was no way that I could, as a scholar of international relations, live through the 1990s and not have an opinion about the democratic peace. We killed more trees with studies about the democratic peace in the 1990s than you can imagine. There’s strong evidence for it. There’s a fair amount of evidence that it’s strongest in the post–World War II period. And there just weren’t that many democracies before World War I, and they were pretty far apart, so you can see why it would be a little more difficult to say.
Rob Wiblin: Hard to tell, yeah.
Bear Braumoeller: But the post–World War II period is also the period when there was an ideological split, with democracies on one side and communist states on the other.
Rob Wiblin: So they were all within the same order.
Bear Braumoeller: Yes.
Rob Wiblin: So maybe that explains it, rather than democracy, per se.
Bear Braumoeller: Exactly. And Kant talked about what Bruce Russett has called the Kantian Triad — which is trade, shared institutions (for example, the international order), and democracy. And there’s a neat piece by one of my colleagues at OSU, Skyler Cranmer, along with two coauthors, Elizabeth Menninga and Peter Mucha, in PNAS in 2015. These folks are network folks, and they do a good job of dealing with the characteristics of the data. And they argue that the Kantian Triad as a whole does result in a reduction of conflict.
Rob Wiblin: Sorry, what’s the Kantian Triad?
Bear Braumoeller: Sorry. Trade, institutions, and democracy. Those three things. And they argue that those three things do produce peace, but that, of the three, democracy matters least, which is kind of an interesting finding. So the upshot is that, historically, democracy is embedded in larger historical structures — like networks of trade and international order. And it’s hard to divide up the impact of different variables in that sort of complex of things.
Rob Wiblin: Because they all tend to come together.
Bear Braumoeller: Yeah, exactly. And it’s not even clear that you should try to divide them up.
Rob Wiblin: Because they cause one another.
Bear Braumoeller: Yeah, exactly. They’re very interrelated, and as a complex, as a structure, we can say that they have this impact. And you can argue about what parts of it are most important, but there does seem to be a “there there.” That said, something that people often miss is that the Concert of Europe was a fundamentally conservative institution. This was a bunch of monarchs who had seen the French Revolution — they had seen liberty, equality, and brotherhood — and they wanted to stamp it out. “Forget it, we don’t want any more of that.” So that was a fundamentally antidemocratic international order, and it maintained peace for the better part of 50 years.
Rob Wiblin: Yeah. I didn’t know what the Concert of Europe was until I read this book. So we’re going to get to that in a minute, and explain what it is in a bit more detail.
I can’t remember if you discussed this in the book, but I think a lot of people have the perception that religion is a key driver of war through history, and maybe still today. And perhaps that’s one thing that has influenced a lot of people to think that a more secular modern world ought to be more peaceful, and that’s one channel by which the Enlightenment has reduced war. Do you have any reaction to that idea?
Bear Braumoeller: Not a big one. Again, I can’t claim to have studied this systematically, and I don’t want to run roughshod over the findings of people who actually have. But I can say, at a broad level, that religion at other times has been a driver of peace. I just mentioned the Concert of Europe, which was a collection of the major powers in Europe. They got together, five major powers, and decided that they were going to cooperate to prevent revolution. Tsar Alexander of Russia very much saw this as a Christian organisation. He referred to it as the Holy Alliance. And there are all sorts of stories about the other people at the conferences sort of snickering behind his back and saying, “Sure, Alex, that’s fine.”
Rob Wiblin: “Whatever you say.”
Bear Braumoeller: “Whatever floats your boat. Just sign the paper.” But to him, Christian unity was sort of the motivation behind the Holy Alliance, which merged into the Concert of Europe. It became the Concert of Europe. And that was in part because if you were a believer in Christianity, you were a believer in the divine right of kings. And the divine right of kings is what the French Revolution was about. So the Concert of Europe was a repressive institution. It was designed to sort of roll back the clock, and then maintain the status quo. But it was, in some part, religious reasoning, and it maintained peace for quite some time. So it’s a complicated thing to answer.
Rob Wiblin: It’s a complicated picture. Yeah, absolutely. You can imagine both religion inspiring war between countries that have different religious views, and also promoting peace between countries that are neighbours, that have common religious values.
Bear Braumoeller: Right.
Rob Wiblin: So my substantially less informed opinion is just, I kind of scan through a list of all of the wars that I’m aware of in my mind. And I think, in how many of these was religion the key issue? Now, obviously, there were periods in history where religion was a key issue in most wars, the wars resulting from the Reformation, and so on. It’s very, very clear that religious issues are very central there. But in the modern era, there’s a lot of wars fought over issues that don’t really seem to be about religion. Or where religion is mentioned, but it seems like that’s a cover for other issues, at least arguably.
Bear Braumoeller: Yeah. There’s a wonderful book by Kalevi Holsti, written years ago. And I reproduced a graph from his findings, where it shows trends over time: religious wars are big in this period, and smaller in this period; and nationalist wars are small in this period, and rise up in the next one. That sort of ebb and flow is really fascinating.
Rob Wiblin: Yeah. Now, we don’t fight over religion. We fight over the appropriate way to govern a country, the appropriate system of international order.
Bear Braumoeller: I don’t know if you’ve read a book called Nathaniel’s Nutmeg?
Rob Wiblin: No. I haven’t heard of it actually.
Bear Braumoeller: Really neat book about the discovery of the East Indies, and the race to obtain as much nutmeg as possible. And there were actually wars fought over the control of nutmeg, which they believed, at the time, cured the plague.
Rob Wiblin: Right. Wow.
Bear Braumoeller: So yeah, people fight over nutmeg. There’s a lot of things people fight over.
Rob Wiblin: Whoever controls the spice.
Bear Braumoeller: Yes, exactly.
Rob Wiblin: OK, pushing on. The last third of the book is about international orders. You think changes in international orders seem to be one of the key drivers, the most plausible key driver, of changes in international rates of war initiation and escalation.
Bear Braumoeller: Right.
Rob Wiblin: Now, international order is a kind of complex concept that different people define a bit differently, but you put it as, “A multilateral security regime that involves one or more major powers, and that is legitimised by a set of principles that are potentially universal in scope.” We could stay on the abstract level, but I think it might be more useful just to talk about some different examples of international orders that people might be familiar with, and why they qualify as international orders.
Bear Braumoeller: So the Concert of Europe, which I’ve now been talking about for three answers in a row, I think, is a great example — because it involved multiple great powers, and the principles involved were taken to be universal, and they thought they had a fundamental understanding of the best way to govern the international system. So that’s a clear example. The Bismarckian system in Europe, in the 1870s and 1880s is a slightly less clear example, because Bismarck clearly believed that he had a good idea for how to manage Europe, and how to manage Germany’s security. It’s a little more arguable as to whether you could see that as potentially universal in scope, but err on the side of bringing it in, just to see what the results look like.
Rob Wiblin: What about some ones that people might have more context on? So what was the international order during the Cold War?
Bear Braumoeller: I think there were a few orders, and they were sort of nested within each other. So the Western liberal order, or the liberal international order, was what we think of as “the West” during the Cold War. The Soviet communist order is what we think of as “the East.” But I also think that there were a couple of other orders. The UN system is sort of an overarching international order that’s less binding, in a sense, and less — I don’t want to say less formalised; it’s extremely formalised — but less deep, in a security sense.
Rob Wiblin: Less strong, I suppose.
Bear Braumoeller: Yeah. And at the beginning, the great powers gave themselves a free pass by giving themselves a veto in the Security Council. So in that sense, it’s a limited international order. But also, I think especially after the Cuban Missile Crisis, there was an order between the superpowers — where they tried very hard to improve their ability to predict one another’s behaviour. I don’t talk about that order in the book — it’s more of a “this is my hunch” kind of thing than that I’ve got concrete data or historians’ accounts to back it up. But it does strike me that, if you take a look at the conflict management that the great powers engaged in, you could say that there was kind of a nascent major power order there. That’s what I mean by international order. So you’re right, it’s a broad concept.
Rob Wiblin: You talk about how there’s these different eras of international order. I suppose you could say, after the Cold War, there was a new one. And arguably, I suppose we’re entering into a new international order, with China being more assertive, and refusing to be constrained by the order that predominated in the year 2000, for example. Are there different “classes”? If you had to list “Here are the four different kinds of international order,” what would they be? Or is that the wrong way to think about it?
Bear Braumoeller: No, it’s fine. I mean, very broadly, I think the categorisation that makes most sense to me is: there are orders that are negotiated — the liberal international order would be a good example of a negotiated order among countries. There are orders that are imposed — like empires, where the subsidiary states don’t really have a choice, or the choices that they’re given are so unpalatable that they effectively don’t have a choice. And then there’s a third category that libertarians are particularly fond of: the spontaneous order — where interactions just spontaneously give rise to mutual expectations and rules of the road. So in a very broad sense, I think you can break orders down into those types.
Rob Wiblin: And is it fair to say that basically, the story with international orders is that once a bloc of countries has established an international order, internally, they tend to have substantially lower rates of violent conflict. However, if you have more than one international order on the scene at any one point in time, then their tendency to fight with one another is pretty substantial. And so, whether they’re good or bad, it’s kind of a balancing between these two things: that internally, they’re relatively peaceful, and externally, they can be quite aggressive.
Bear Braumoeller: Right. Yeah, exactly.
Rob Wiblin: That doesn’t comment on the anarchy situation, or the situation where there is no order. How does that perform?
Bear Braumoeller: Well, anarchy is, in a way, kind of the baseline condition. And it tends to be certainly more conflictual than life under international orders, and generally more conflictual than relationships across orders as well.
Rob Wiblin: Really? OK, so it’s really kind of the worst case.
Bear Braumoeller: It’s a really interesting question. So I wrote this book, and what I observed was higher rates of conflict initiation across international orders. I just sort of assumed what happens is you get two different groups forming, and once you have an in-group effect, you get a cross-group conflict. Well, it turns out that there’s a whole psych literature on this, and that doesn’t generally happen. In-group effect doesn’t automatically lead to out-group conflict.
So my lab and I came up with a model of how it is that hierarchy, and hierarchical order, and international war are related to each other. And we also found, in that model, that just the formation of multiple hierarchies, in and of itself, doesn’t cause an increase in conflict between them. So, it’s a really interesting puzzle of why it is you see increased rates of conflict initiation. And I’ve got a couple of answers that are potentially worth considering.
One is that multiple international orders don’t cause increased levels of conflict; increased levels of conflict cause multiple international orders. So the world becomes more dangerous, and as a response to that danger, you see international orders form. Which actually maps very well to the trajectory of the Western liberal order and the Soviet communist order after World War II. They fought on the same side, and only gradually did each realise that the other was a threat. And as that perception of threat increased more and more, there was an increasing tendency to try to form orders, to kind of counteract it. So that may be one reason that you see more conflict across multiple orders: that conflict creates orders.
Rob Wiblin: It’s reverse causation.
Bear Braumoeller: Yeah, exactly.
Rob Wiblin: Yeah. I guess the technical term is that all of these things are endogenous. So it’s extremely hard to do causal attribution, because everything is causing everything else. Well, it plausibly is.
Bear Braumoeller: Well, that’s why theory is really important, right? Because unless you have a concrete theory of how these things happen, it’s very easy to convince yourself that shirtsleeve calculations or broad reasoning will lead you to the right answer. But we actually constructed a formal computational model to simulate international systems. And we fed it a set of premises, and we watched how the model behaved. And we came away saying, “OK, so why is there this conflict?” And we do see it happening, but we see it happening for exactly the opposite reason that we had thought.
Another possibility is that orders tend to increase ideological polarisation. When you say, “You can become a member of NATO, but you have to be a liberal democracy,” what that effectively does is it means that the process of forming an international order increases tensions between the member states and states outside the order, right? Because it increases the sort of ideological differences between the two, that can lead to war. So one possibility is that this process of making states more democratic, or more communist, draws them more into conflict with one another than they otherwise would’ve been.
So there are a couple possibilities for what this mechanism is, but we know surprisingly little, actually. And that’s one of the reasons that we’re doing this work, is to try to straighten this out.
Rob Wiblin: Yeah, fantastic. I learned so much from this last bit of the book. I mean, I know a bit about a decent number of different times in history, but it turned out that 19th century history was a much greater gap in my mind than what I had appreciated. I feel like it’s very natural that — like most people, I suppose — when I’ve looked into history, I’ve looked into World War II, then maybe the Cold War, then maybe World War I, then maybe you think about the modern era.
But at what point do you start studying the Bismarckian system? That seems like it’s perhaps more advanced than an amateur like me might typically get into. But I think it really would be helpful for people who are commenting on international relations, even in an amateur capacity like me, to have a broader reference class, a broader set of knowledge of different eras, and how countries got together in those times. Because having a tiny dataset of only four eras — especially four eras where the issues were often quite similar, and carried over between them — just means that you know very little. And you can plausibly just say things that will be very silly to someone who does understand the 18th century and 19th century, because there’ll be great counterexamples to what you’re saying.
Bear Braumoeller: Right. And actually, this is a general problem in studying international orders. There just haven’t been very many of them, right? So this is one of the things that I love about my research lab. The other members are graduate students, and you’ll be seeing their dissertations as books before too long, I hope, and I think they’re going to be really interesting.
They’re doing just incredible things, comparing the Syrian Empire to the Ottoman Empire. Or in another case, comparing the Concert of Europe with the formation of the United States, and saying, “Why is it that you get a unified state in the United States, but you don’t get a United States of Europe in 1815?” And they’re incredibly creative about how to come up with these comparisons, and incredibly fearless when it comes to diving into the literature on these places.
So it is a problem. Even if you go into the 19th century, you still have a pretty limited number of examples. I’m very lucky to be working with these smart, intellectually adventurous people, who are teaching me a lot about the Ming Empire. It’s really terrific.
Rob Wiblin: If there’s one thing you can say about history, it’s that there’s a lot of it. I think we need to stop producing it, until we can, preferably, deal with the history we already have.
Bear Braumoeller: Just have a pause. Yeah.
Rob Wiblin: The era that I was most keen to share a little bit about with the audience is this Concert of Europe period, where there was a very clear order that you’ve already discussed a little bit. To do a bit of setup here, in the late 18th century, you’ve got the French Revolution, a lot of new innovative and controversial ideas going around. Then you have the Napoleonic Wars, from around, was it 1800 to 1815? Incredibly bloody period. Napoleon never saw a country he didn’t want to invade. So there’s just been constant bloodshed. Finally, in 1815 or so, they managed to get rid of Napoleon, and they’re like, “All right, no more of this.” So just talk about the agreement that the European powers have put together at that point?
Bear Braumoeller: One of the reasons that Napoleon was so dangerous is that he kept running around Europe and saying, “You, Poles. You’re Poles. Be Poles. Be independent.” “You, Italians, you’re Italians.” They say, “Yes, we’re Italians,” and they’d rise up. So he became an enormous pain for these multiethnic empires that existed at the time. And if there was one thing that they could agree on at the end of the Napoleonic Wars, it’s that they were done with that; they’d had enough of this whole nationalist uprising thing.
And so they came together at the Congress of Vienna. And it’s interesting. There was a lot of negotiation, there was a lot of discussion, there were agreements made about the idea that they would work together to govern the European system, at least, and ensure that it would remain conservative and monarchical.
Even the UK was part of this, although Castlereagh was kind of out on a limb. Foreign ministers were not monitored with the same frequency that they are today, and he did a bit of freelancing. So there was this agreement, and partly because people knew that it wasn’t going to pass muster in Britain, it wasn’t formalised. And as a matter of fact, Castlereagh had to write a sort of strained paper in 1820, saying, “As you know, we will not be a part of any attempt to suppress democracy on the continent. Wink, wink.” So keeping the UK in was a bit of a challenge.
But interestingly, it was very thinly institutionalised. They met at a few congresses to discuss problems as they arose, but there was no central body. There was no UN building that they would go to, and meet on a regular basis. It was just a very ad hoc sort of thing. So I’m not surprised that people haven’t come across it, because it was very —
Rob Wiblin: Not so conspicuous.
Bear Braumoeller: Yeah. Now, it started after the Napoleonic Wars, as you mentioned. Interestingly enough, because it was so thinly institutionalised, it’s sort of hard to say exactly when it ended. It just kind of slowly faded away. But if you could point to one end point, I think it would be the Crimean War in the 1850s. But in that period, it was remarkably successful.
Rob Wiblin: In the book, actually, you say, “The four decades following the Napoleonic Wars were, by a significant margin, the most peaceful period on record in Europe.” Is that true, even considering the 1990 to 2020 period?
Bear Braumoeller: Sure. I mean, there was the war in Yugoslavia. Now, in part it was peaceful because anytime there was a threat to the peace, the great powers intervened. Sometimes bloodlessly, sometimes not. But I think, after a few interventions, people tended to get the message.
Rob Wiblin: Yeah. Right. So it was like active peacekeeping, but it was effective. Why did it break down?
Bear Braumoeller: Largely because Russia, the UK, and France, having managed successfully to govern the international system for decades, stumbled into a war that none of them actually really wanted, in the Crimea.
Rob Wiblin: It’s one of those wars where people are like, “What were they doing? What were they thinking?”
Bear Braumoeller: Yeah, right, exactly. And I think it often wasn’t entirely clear to them. But it just made it very clear that you could no longer take consensus among the great powers for granted. And once that happens, you have Prussia saying, “Well, I do have some territorial ambitions that I’d like to realise, actually. And if nobody’s watching, if Mom and Dad are no longer punishing us for territorial aggrandisement, I have a few territories that I’d like to pick up.”
Rob Wiblin: Yeah, right. I guess it might seem a little bit odd to talk about this as a global international order. And obviously, it wasn’t.
Bear Braumoeller: No, no.
Rob Wiblin: It slightly is a confined set, because those major powers in Europe didn’t face external threats, really. So they could maintain peace between one another there. From their point of view, there were no wars being fought against them, or no wars that they weren’t choosing.
Bear Braumoeller: Right. Most orders are regional. Even the liberal international order is a regional international order.
Rob Wiblin: So what can we learn from that period that could possibly be relevant today?
Bear Braumoeller: It’s an idiosyncratic period. And most of these countries were ruled by individual people, or small numbers of people. So it’s kind of difficult to apply the lessons. But we can say, I think, that it showed us that major powers can prioritise international peace over domestic ideology. There were substantial differences, and growing differences, in domestic ideology over this period. But for a time, at least, they managed to subsume those differences, and actually pursue joint interests in peace.
Rob Wiblin: I see.
Bear Braumoeller: The other thing is, Michael Howard — Sir Michael Howard, you’ve got to give him his due — refers to the Concert of Europe as “the invention of peace.” And his argument — and I’m not going to argue with a historian of his calibre — is that this was the first international order in which the maintenance of peace was an explicit objective. And so, in that way, it’s a watershed event in the international system.
Rob Wiblin: Was it unusual to have a treaty between so many different major powers, where they all agreed to try to avoid fighting one another? This wasn’t typical?
Bear Braumoeller: Yes, that’s right. As a matter of fact, there was a discussion after the end of the Cold War about the possibility of a major power concert of some sort. That idea was floated a couple of times. And the problem was, I think a lot of people looked at the Western liberal order — or the liberal international order, as they were calling it by that point — and said, “If it ain’t broke, don’t fix it.”
Rob Wiblin: I see. Yeah. Possibly we could live to regret that decision.
Bear Braumoeller: Yes. Yeah, we could.
Rob Wiblin: Do you think that the main reason for people taking this novel approach was just that they were so scarred by the Napoleonic Wars — and I suppose also internal revolutions within lots of countries — that they were willing to set aside their territorial ambitions, and their interest in invading one another, and say, “We’ll leave that for another day, or maybe never, and let’s just focus on not getting overthrown?”
Bear Braumoeller: Right. Yes. And I think, in part, that’s wrapped up in the theories of war that they had in mind. So the major power conflicts that had occurred previously had often been limited in scale, certainly relative to the Napoleonic Wars. They’d involved professional armies. The Napoleonic Wars were really a revolution in warfare. And I think there was an uncommon unity. Just about all the powers were affected in one way or another, so there was an uncommon degree of leaders being on the same page about what the problem was — which was revolution — and how big the problem was, which was: revolution anywhere threatens the peace everywhere.
If you’ve just had a big war, you may overcompensate and think that wars are going to pop up all over the place. International order, I think, benefits from that perception. It’s one of the reasons you see international orders form after large wars. People say, “I don’t want to do that again.” But the fact that this order was able to form, despite everybody not quite being on the same page about principles of international governance, is kind of inspiring. And I think the historians who wrote about “Do we need a new Concert?” saw that as inspiring as well.
Rob Wiblin: Later on in the 19th century, possibly going into the early 20th century, Europe followed a different system, that was also reasonably peaceful, called the Bismarckian system. What were the characteristics of that one?
Bear Braumoeller: Bismarck was a really special case. He designed a system of alliances — actually three systems of alliances — in order to deter conflict. He did this because conflict would almost inevitably involve Germany. Germany had just won a series of wars and declared itself a unified German state. And in the process, it had made basically a permanent enemy out of France. Anybody who attacked Germany from the east could count on France attacking from the west.
So Bismarck had very good reasons for becoming a man of peace in 1871. He basically used alliances to tie up the other great powers and prevent coalitions from forming, in particular coalitions involving France. He did this, like I said, in three different clusters of alliances, over the course of about 20 years. Most statesmen, I think, and certainly not his successors, couldn’t manage that balancing act.
Rob Wiblin: I see. The idea was that he would ensure that there was never a coalition or an alliance large enough that it would think that it was a good idea to launch a war because they would be able to win and gain territory. Things would always be sufficiently finally balanced that everyone would be too nervous to actually start a war.
Bear Braumoeller: Correct.
Rob Wiblin: OK. Do you think he got lucky?
Bear Braumoeller: It’s a great question. It’s very common to attribute the Bismarckian system to sheer diplomatic genius. I don’t think I’ve seen substantial movement away from that interpretation, although I’m not exactly steeped in the literature on Bismarck. But he may have been fortunate in that the countries that were closest to him, for the most part, had just lost to Germany.
Rob Wiblin: They also saw the virtues of peace.
Bear Braumoeller: Exactly. But the big caveat to that is France really wanted to retake the territory that it had lost.
Rob Wiblin: Is there any potential relevance to us today from that system? It seems like it’s a little bit more of an anarchic one than perhaps what we have right now. But nonetheless, you can balance powers off against one another and keep them quiet that way.
Bear Braumoeller: Yeah. It’s funny. The people who refer to themselves as political realists, people like Henry Kissinger, have been aspiring to be Bismarck for generations. There are definitely people around in the foreign policy establishment who have ideas about how it is that you can align politics in such a way as to reach peace.
Now, often it’s impractical. I’ll give you an example. Henry Kissinger argued that the best way to avoid war with Ukraine and Russia was for Ukraine to declare neutrality. He’s probably right. That probably would’ve worked, but the Ukrainians weren’t going to do that.
Rob Wiblin: They didn’t want to.
Bear Braumoeller: It’s not going to happen. You know?
Rob Wiblin: Yeah. I’ll just throw in here that one of the most important points that you make — throughout the book, actually — is that we could choose peace. We could have peace, if only we didn’t care about the issues at stake enough to be willing to fight. If we’re willing to surrender and give up those things, then we don’t have to fight. But however, people have always cared about some things, or at least sometimes have cared about some things, more than peace. That remains true today. That’s why we’re willing to fight. That’s why we might be willing to fight a massive war, because there might be something incredibly important to us at stake.
Bear Braumoeller: That’s right. I quote John Lennon on this, who, during the Vietnam War, had a wonderful song called “War Is Over.” People would ask him, “How do we end the war in Vietnam?” He said, “Just stop fighting. Put your guns down and walk away.” Now, Lennon was not exactly the cheerful Beatle. I’m pretty sure that what he had in mind when he wrote that was, “You could stop war, but I know you’re not going to. And that’s because there are things that you care about more than peace.”
Rob Wiblin: Yeah, it’s true. It’s a quote that could, from one point of view, seem incredibly naive, and from another point of view, could seem incredibly smart. “You’re choosing to do this because you care about other things as well as peace.”
Bear Braumoeller: Yes, exactly.
Rob Wiblin: It’s always an active choice. OK, let’s talk now about what recommendations might possibly come out of your research. Of course, all of these issues are pretty tentative. It sounds like the international order research project is at an early stage and potentially has a long way to run. Fingers crossed there’s a lot more to learn.
First up, there’s a reason why we’ve done a lot of interviews this year on war. It’s very much in people’s minds these days. Firstly, the Russian invasion of Ukraine, and more recently, a lot of sabre-rattling over Taiwan. Looking around, what characteristics of the current international order stand out to you as important and salient?
Bear Braumoeller: I think the way to describe it is that we have a fairly well-defined order in the liberal international order. There are a couple of smaller regional orders. Russia, I think, is the easiest one to point to. The Russians for a long time have believed that the countries in the “near abroad” — that is, the former Soviet states — are only really nominally independent: they’re independent by the good graces of Moscow. With the exception of the Baltics, I think, which were obtained later, and I think that Russians don’t see themselves as being that —
Rob Wiblin: Invested in it.
Bear Braumoeller: Yeah, exactly. But the other former states of the former Soviet Union, I think that Russia has been spending part of the post–Cold War period essentially reestablishing its political dominance in those areas. A big part of what’s happened is that we’ve run out of wiggle room. We’ve run out of empty space. The liberal international order has expanded to the point where we’re now talking about incorporating former Soviet states. That’s a point at which international orders bump up against each other.
Rob Wiblin: Push comes to shove.
Bear Braumoeller: Yeah, exactly.
Rob Wiblin: How about China?
Bear Braumoeller: It’s funny. There was a statement early on in the Ukraine War from, I think, the Chinese foreign ministry. It started off by saying, “People are drawing these parallels between Ukraine and Taiwan, and they’re completely unwarranted. The situations are completely different.” I thought, “Great. That’s terrific. I’d love to believe that.” Then the statement went on to say, “Because Ukraine is an independent country, whereas Taiwan is actually part of China.”
Rob Wiblin: I’m reassured.
Bear Braumoeller: Yeah, exactly. I’m just like, “Oh, that’s so much better. Yeah, you’re right. That’s awesome.”
Rob Wiblin: The one that shouldn’t be at war is at war, and the one that should be isn’t yet, I guess.
Bear Braumoeller: Yeah. Part of what’s going on here is that the stakes are abstract and not really easily divisible. If Russia and Ukraine were fighting over Crimea, and only Crimea, that’s something that could be resolved fairly easily. What they’re fighting over is what international environment they want to live in. Russia wants to live in one in which the former Soviet states are subservient, and Ukraine is not interested in living in that world. And you can’t really split the pie.
That’s a problem with Taiwan as well, and possibly even a more serious problem. Because the war in Ukraine has gone on for a while. Sometimes you’ll see people saying, “My guess is the way that this is going to end is that Russia de facto swallows part of eastern Ukraine and they declare peace and call it a day.” I could conceivably see that happening just because of the prevalence of Russian speakers in the eastern part of the country. I don’t see any of the same dividing lines in Taiwan. Russia invading and doing really poorly militarily did not increase Ukraine’s willingness to give up part of its country. I’m not saying this is an obvious or easy deal to be made, but I don’t even see that in Taiwan.
Rob Wiblin: Are you saying it’s quite hard to reach a peaceful negotiated settlement between Ukraine and Russia? Like one of the reasons why you ended up with a war was that the issues at stake are so expansive that these countries are going to have with one another — about the entire international culture — that you couldn’t simply sit down and make an easy agreement with various different points, because everything is on the table?
Bear Braumoeller: Right.
Rob Wiblin: Yeah. Interesting. Are there any other important aspects of the international order that we haven’t mentioned that are worth discussing? We haven’t really discussed Latin America or Africa or Southeast Asia very much, but I suppose potentially, from a great power war and nuclear war point of view, they’re not the most important actors.
Bear Braumoeller: I think the thing that I would say is, when people talk about things that can be done to increase the prospects for peace in the current international structure, there are a handful of ways to think about it.
Let’s start with an understanding of war that comes from Clausewitz — but actually was really well articulated by a guy named Jim Fearon in 1995 — called the “bargaining theory of war.” He pointed out that, yes, war is an extension of politics — but in the sense that you have an issue, you negotiate over it, and war is what happens when you can’t reach a negotiated settlement. So we need to understand why you can’t reach a settlement.
From that perspective, what you’re seeing in these cases is that issues arise that people are willing to fight and die over, and they can’t arrive at a settlement short of war. How do you fix that? You can make them less committed to the values that they currently have. Ukraine could say, “I don’t really want that territory.” That’s not likely going to happen. One in which the likely costs of war are very high here —
Rob Wiblin: Is nuclear deterrence?
Bear Braumoeller: I was going to say one theoretical way to do this would be to give Ukraine nuclear weapons. Do you want to do that? No. I don’t want to do that. I think it’s a terrible idea. It might increase the odds that the war will end, but if it doesn’t, it’ll probably increase the costs dramatically.
But actually, I think that the unbelievably broad, crippling sanctions that were implemented in Ukraine are a fantastic signal for China, and a warning for China that they can expect the same in Taiwan. I think actions like that actually can make a difference. It’s hard to make a difference in the current conflict, but it can make a difference for the future.
Rob Wiblin: By setting an example that you’re willing to pay a high price in order to punish territorial conquest or annexation.
Bear Braumoeller: Right. Exactly.
Rob Wiblin: OK. I’m not quite sure what the practical implication is. I suppose things are going on in the world all the time, and I want to have some opinion. Ideally, I want to say, “It would improve humanity’s prospects if country X does Y.” Then I’m trying to bring this international order framework to bear and then think, “What does this imply about policy for the UK or the US? In principle, what would I want Russia to do?” It seems like one bottom line with the international order framing is that the most peaceful state of the world would be one hegemon who is powerful enough to either dominate or ideologically sway everyone in the world, so now there’s peace within this one order and then there’s no one else to fight — I guess, unless aliens come along.
Bear Braumoeller: Or, less uncharitably, provide security for everybody.
Rob Wiblin: I see.
Bear Braumoeller: Because one way to think about it —
Rob Wiblin: If you fight amongst yourselves, we will punish the aggressor.
Bear Braumoeller: Yes, exactly. I think that’s a common understanding of how hierarchical international orders form: that it’s a trade of security for autonomy.
Rob Wiblin: I see.
Bear Braumoeller: The smaller states say, “We’ll give you some policy thing that you want. We’ll vote with you at the UN. We’ll give you basing rights. We’ll pay cash, whatever — and you will provide security for us.” If there’s a hegemon that’s powerful enough to make that trade with all the countries in the world, then that would definitely be a situation in which you’d find more peace. We are pretty powerful, but we’ve decided to limit the scope of our security deals to democracies. That inherently limits the scope of the order.
Rob Wiblin: Is that for ideological reasons or just because it’s so impractical to try to make those deals with everyone? The US military is powerful, but it’s not omni-powerful.
Bear Braumoeller: I don’t know. I think they’d rise to the challenge. We already outspend the next 14 countries — or whatever it is, 11 or 12.
Rob Wiblin: Yeah, but you’ve got to ship everything over. It’s a whole effort.
Bear Braumoeller: Right. No, I think, actually, it’s because that would mean agreeing to outcomes that are normatively unacceptable. We would have to be cool with the way the Chinese treat the Uyghurs, or at least agree to look the other way.
Rob Wiblin: Sorry, in what way?
Bear Braumoeller: If we were to include China in a broader sort of concert system.
Rob Wiblin: I see. Right. And we feel uncomfortable because we’re making allies out of people we have ideological and moral disagreements with.
Bear Braumoeller: Right. Exactly. Again, this gets back to the John Lennon point. The simple answer to how to stop war is stop caring about things, which…
Rob Wiblin: It’s a great example of ways in which the Enlightenment doesn’t universally encourage peace. Without the Enlightenment, we might not give two shits about a group of people we know nothing about, on the other side of the world. We might not care about the conquest of Taiwan and the rights of Taiwanese people. That would be bad in one way, but it would be good in the way that you’d be less likely to have a great power conflict over those issues.
Bear Braumoeller: Right. Exactly. Steven Pinker talks about the rise of empathy as a cause of peace. And it probably is, but it also leads us to go fight wars in faraway places for people we empathise with.
Rob Wiblin: Yeah. I want to ask a question about what kind of policy suggestion you might have about how the liberal international order ought to shift itself over time, or how it ought to interface with China. What agreement might it ideally make with China or Russia?
Bear Braumoeller: Yeah, sure. How do we get from our current understanding of international order to a more peaceful system? One way, as I just mentioned, is to care less about things. But that’s just not going to happen.
Rob Wiblin: It’s not going to happen. Yeah.
Bear Braumoeller: Another is to increase the likely costs of war, so the states are more reluctant to go to war.
Rob Wiblin: But that’s a pretty double-edged sword.
Bear Braumoeller: Yeah. As I say, the way that I’m trying to do it is to say, “Hey, escalation is really bad. Escalation is probably more costly than you realise.” I’m trying to get that message out there. If you’re trying to predict how old a person is going to be, the older they are, the fewer years you give them. With a war, it’s exactly the opposite. The bigger it gets, the bigger you predict it’s going to be. People don’t get that intuition. I’m hoping that helping to get that to sink in will at least help people appreciate the costs of war that they already face.
Rob Wiblin: I see. Something that’s unambiguously bad is if people are taking actions that incur a high risk of a really catastrophic war and they don’t realise it. Because then you bear all the costs of the war, and you don’t have any of the benefit of deterrence.
Bear Braumoeller: Right.
Rob Wiblin: I see. I think one problem we face is that we haven’t had a massive war in a long time. So people do not appreciate it. They don’t read the news about Ukraine and imagine New York being a smouldering ruin, and fully contemplate that actions that they take regarding Ukraine could result in the complete destruction of their societies — in a way that for people in 1946, that was probably very front-of-mind.
Bear Braumoeller: Right. They also don’t appreciate things like the Iran–Iraq War, which is not a war on a global scale, but it was a massively costly war for those two societies. Unbelievably deadly. Even if they don’t see the potential for the spread of war across the international system, they should appreciate the possibility that a war that’s reasonably well contained is going to kill a lot of people.
Rob Wiblin: Right. I see. Just making it more salient to people, that even in the present day, wars escalate to a phenomenal degree and kill a shocking fraction of the population of the combatant countries. We should remember that, because it could be us next. Are there any incremental changes to the current international order that would lower the risk of war, that are actually plausible?
Bear Braumoeller: Right. Like I said, I think the most hope lies in information — information about the dangers of escalation, information about other states’ likely costs of war. This is why I think sanctions were such a good thing, because they conveyed information to China that the likely costs of invading Taiwan are probably higher than they think. I think there’s room to work at the margin there. There’s also room to work by increasing the costs of invasion, which we do by bolstering the troops in the area and so on and so forth. I think there is some wiggle room there, but in terms of something large and transformational that’s going to produce a change like the end of the Cold War, it’s hard to imagine that without a fundamental restructuring of international order.
Rob Wiblin: Do you think there is a case for a more transactional — arguably, “accepting of evil” — arrangement between the US and China, that would say, “The war is just so costly that we have to sacrifice other values”? To some extent, you already are. The US did not invade China because of the Uyghurs. It complained a whole bunch, but it knows that that is not anywhere near within its power. We are transactional in some ways. We just don’t call it that.
Bear Braumoeller: One of my students, Andy Goodhart, is actually looking at this question. Not the China question in particular, but he’s looking at international orders that have based their legitimacy primarily on performance. Like, “We will keep you from being invaded,” rather than legitimacy in the sense of, “We have a normatively correct way of ordering society, and we think you agree with us. So let’s form an international society based on those principles.”
I think China, to a large extent, is already doing this. If you ask people in China, “Why do you support the Chinese state?”, they’re not like, “I’m a true believer in communism.” I think a lot of the answers you would get are, “My parents grew up in a hut and now I drive a BMW.” I tend to think, at least domestically, China already sort of operates on those principles. I’m not entirely sure.
It’s a huge, big, open, interesting question as to whether we can reach an accommodation along those lines. But the problem is you have to convince people to restructure what they would say is the most successful alliance in history.
Rob Wiblin: Also going half of the way there could be the worst of all worlds. Because let’s say you start giving more transactional signals to China that you’re not really going to have a values-based foreign policy. And then they invade Taiwan. Then your population’s not actually with you, and maybe your heart wasn’t really in that in the first place. So now you do respond. Now, you’re at war, because you’ve given indications that you’re not going to retaliate. But, actually, you always were, in your heart of hearts. It’s just a nightmare.
Bear Braumoeller: Yeah. These kinds of international relations are full of perverse outcomes like that.
Rob Wiblin: Yeah. So lots of people in the audience would love to help reduce the risk of both smaller wars and larger wars, if they can. What could listeners take away about ways that they could potentially contribute, or that their countries could contribute?
Bear Braumoeller: I’d suggest graduate school in political science for one thing.
Rob Wiblin: Honestly, it does feel like that’s the answer sometimes, that maybe one has to be extremely knowledgeable in order to give good advice on these things.
Bear Braumoeller: Yeah. The way that we’ve structured our pursuit of understanding war is not one that maps really well to understanding the relationship between order and war. Academia, I think, needs a fundamental reorientation in terms of how it thinks about conflict and aggregation across levels and dynamics and things like that. This is where the book gets pretty dark. I think the best thing that people can do is think about that card analogy — 96 cards, do you really want to draw one — and try to get a better fundamental understanding of how escalation works, and then vote accordingly. Because I think if I had understood just how dangerous war is, prior to writing this book, I would’ve been a lot less evenhanded in policy arguments in the past 20 years or so. I would’ve seen the maintenance of peace as a much higher priority. I think that that can help.
Rob Wiblin: It seems like making a blockbuster movie about a war between the US and China, that occurs in a realistic way where both sides get absolutely screwed, and at the end of the movie these countries are in shambles. Hopefully, it’s watched on both sides, and it really makes salient to both sides the risk that they’re running, and maybe causes both to be a bit less aggressive and try to find more ways to avoid it. It’s a little bit harder to see how that backfires.
Bear Braumoeller: That’s a good point. I don’t know if you remember a movie called The Day After?
Rob Wiblin: Yes. Horrific.
Bear Braumoeller: It really brought home the dangers of nuclear war. You could tell, for a while at least, it had that impact on society. It was all anybody talked about. If I remember correctly, it was even before social media, and it was clear that everybody was talking about it everywhere.
Rob Wiblin: OK that’s all my questions on international orders. Let’s now come back to the core topic of the book about which we left everyone hanging earlier on, which is the empirical question of whether, by the numbers, war is in long-term decline or not.
So as you mentioned earlier, the book is in part a response to The Better Angels of Our Nature by Steven Pinker, where among other things he argues that indeed war is in long-term decline. On balance, obviously you think that when it comes to interstate conflict, Pinker is wrong to think that that kind of violence has a long-term trend downwards, and we should be less worried about it today than maybe we should have been 50, 100, 150, 200 years ago.
But before we dive into that, what do you think Pinker gets right? Because he makes many arguments in The Better Angels of Our Nature and you’re far from disagreeing with him on everything.
Bear Braumoeller: No, I don’t [disagree with him on everything]. And I think it’s sort of interesting. I think he gets the broader story about human cooperation right, despite not getting the particulars.
There’s actually been a lot of literature recently on human prosociality that points out, for one thing, that humans are — by the standards of other species — insanely willing to cooperate outside of their immediate kin group. We always talk about ants being cooperative, but ants have never been to the Moon. It’s just incredible what humanity has done by cooperating. And that strand of the literature, I think, is what he was picking up on when he was thinking about this book. At least that’s my sense. And that literature has a really fundamentally good point to make: that cooperation is the key to humanity’s rise. He just misses the fact that we’re really warlike too. And those two things aren’t necessarily contradictory.
As far as the other forms of violence that he mentions — like crime and so on — I’m totally agnostic about those. I don’t want to pretend to be an expert in something I’m not. And I read those chapters, and I thought, “Maybe, I don’t know. That sounds plausible.” And I think all the people who write about crime read the chapters on war and went, “That sounds about right.” It’s just when it gets into your own field that you start saying, “I don’t think so.”
Rob Wiblin: Everything gets more complicated the more you know about it. Or most things, anyway.
Bear Braumoeller: Right. But I also think that he’s right about the fact that we’ve made progress in reducing conflict. I just think that progress is very limited, and we don’t really understand it fully. And I don’t really find his causal mechanism compelling.
Rob Wiblin: Yeah, that’s a super fascinating issue that we’ll come back to later. We should maybe be clear about what exactly Pinker was claiming that you disagree with, because I feel like it’s possible to get a bit confused between two different claims.
One would be that, say, the present era — since 1990 to the present day — is more peaceful than maybe the long-run average. Or perhaps that since the end of World War II to today, that’s been more peaceful than the long-run average. That’s one claim, and I think maybe that is actually more plausibly on the table.
The other bolder claim perhaps, is that there is a long-term trend towards a reduction in interstate conflict since, say, 1700 or 1600 or so on — which I think you are maybe more sceptical still about that one. Is that right?
Bear Braumoeller: It’s funny, I’m sceptical about some of the mechanisms that he talks about. The arguments about the role of the Enlightenment, for example, I think don’t really pass superficial muster. But I’m not an Enlightenment scholar; my area is understanding numbers in war. And so the main part that I disagree with is when we get into the actual data analysis and the claims that he makes. He actually lays out four of them right there in his book. The main two are a long-term reduction in conflict. And the second one, actually he argues for an increase in the lethality of warfare that declines after World War II — which seems like it would be a slam dunk, since World War II sort of rewrote the book on lethality in warfare. But I think there are pretty compelling reasons not to buy those stories.
Rob Wiblin: So if I recall, in the book, Pinker focuses on this quantity — I think it’s the percentage of the global population dying in wars per year — and he argues that has trended down on average towards the present day over some period. I think to many people that might seem to settle the issue that if a smaller fraction of the population is dying per year in wars, then that is a pretty strong indication that war must be in decline. Why do you think that wouldn’t settle the issue?
Bear Braumoeller: So first of all, I should note that in my book I take a look at three different measures of war: the raw lethality of war, lethality per capita (that is, the battle deaths divided by the populations of the combatants), and then lethality as a percentage of world population.
And I should note that people who study war basically never use the last one. That is not a measure that exists in the literature. I think the reason is that they think of war as behaviour, and they’re trying to explain that behaviour. How you measure something really depends on what you think that thing is. If you think of war as a cause of deaths, like heart attacks, then you probably do want to take a look at war divided by world population one year after another. And you’d find that it’s really small in recent years. You’d also find that it’s really not small in 1916, 1917, 1918, 1940, 1941. Nobody talks about those years, but you would find that it’s small in recent years.
But what does that tell you? There is this Oka et al. piece from PNAS that I cite in the book. They argue that as societies get bigger, the size of the war group — the army and military — gets smaller as a percentage of the population. So maybe that’s what’s going on: maybe the state system is just growing, and states are, relative to previous decades, just smaller and weaker and farther apart. There are a lot of reasons for changes in the annual rate of war deaths per capita worldwide that aren’t people getting less warlike.
Now, if you think of war as behaviour, you want to know how often wars happen. You want to know how big they get. Interestingly enough, if you forget about the year-to-year count, if you look at the lethality of wars measured as deaths divided by world population, the lethality of wars has not declined even by that measure. Which really should scare you, because world population has grown a lot. And if the lethality of wars tracks with that, that’s pretty bad.
So the reason that I don’t think people should be satisfied with that measure is that wars are sort of like pandemics. Pandemics are really not a huge cause of human death at all most of the time — unless you happen to be in a year with a pandemic.
Rob Wiblin: It’s incredibly clustered.
Bear Braumoeller: Along the same lines, theoretically, everybody’s at risk from a heart attack, but the population of China is not at risk of a war that’s happening half a world away that they have nothing to do with. So I think that’s the main dividing line between the mindsets of people who find those two different measures appealing.
Rob Wiblin: In the book, you refer to this approach — of thinking about it as a percentage of the global population dying in war per year — as thinking about war as a public health problem. So you’re trying to evaluate how large that is, as a fraction of all deaths, say.
Whereas you are more interested in studying cultural and political and international relations changes — where you’re instead trying to study human behaviour and culture and propensities. And for that, we kind of want a different thing. We want to see how often, when there’s a plausible opportunity for a war to start, does a war start? So instead, you start looking at these war initiation rates and things like that, which we’re going to talk about later.
You were saying that even if warlikeness remained constant over time, because the population of countries is getting larger over the last few hundred years, we might expect fewer people to be dying in battle — because as countries get larger, they have a lower fraction of their population in their army.
Bear Braumoeller: Correct.
Rob Wiblin: I suppose because they feel like they’re hitting declining returns, and they just don’t need to have so many people in their army in order to plausibly be able to defend themselves. So we might have expected, just on that fact alone, for war to be in decline by this health or fraction-of-all-deaths measure. And yet you think it’s actually not declining even by that measure, or at least not reliably declining.
When I was putting together this question, I went to the Our World in Data website and looked at their graph for “Global conflict deaths: a comparison of different data sources, number of deaths, 1400 to 2017.” And to me, it just looked like a complete mess. Sometimes it’s kind of low, and then it goes up a bit, and then it goes back down again, and then it goes up again, and then it goes down again. And it just seems to do that basically since 1500. I suppose World War I and World War II stand out a bit, but there’s been enormous periods of serious conflict in the past as well. And zooming out that much, the current period just doesn’t look special at all. At least not to me, just eyeballing it.
Bear Braumoeller: No, I think that’s exactly right.
Rob Wiblin: So then, what is the fundamental crux of the disagreement between you and Pinker? If you could talk for a long time, why does he think it’s declining from this health point of view when you don’t, and eyeballing it seems like it’s not declining?
Bear Braumoeller: I would have to know more about him to know why he’s reading the data in the way that he is. And I don’t, and I don’t want to guess.
Rob Wiblin: Thinking about the broader group of people who endorsed this decline of war theory, do you have any sense of what the crux of the disagreement is between them and people who are more likely to agree with you?
Bear Braumoeller: Well, abstracting away from the particulars for a minute, my sense of why Pinker and I are coming from different places — and this may relate to broader groups of people and society in a nutshell — is that he’s a psychologist and I’m a political scientist. He locates the drivers of behaviour inside individual humans. And I think a lot of people reflexively do that.
I study the world. And political scientists, and sociologists, and other social scientists in these areas study the way that individual incentives produce group behaviour. And fundamental to that perspective is something called the ecological inference fallacy — that is, the idea that you can infer group behaviour from individual characteristics or vice versa.
Which clearly isn’t the case. One example: in the US, wealthier states vote democratic. But you can’t, from that, infer that wealthier people vote democratic. In fact, wealthier people vote Republican. So the characteristics of individuals don’t necessarily have anything to do with the characteristics of aggregates, because aggregates don’t just transmit people’s preferences directly.
So while he does acknowledge the role of the state at the start of his story, most of his story is about what’s happening inside human minds over time. And I think that’s a story that is easy for people to wrap their heads around, and intuitively connects to the way that people think about the world. But one of my biggest problems when I was reading through the book is that human aggregates have their own dynamics and characteristics. And where those come from is actually a fascinating set of topics, but you can’t just take it for granted.
Rob Wiblin: Yeah, that’s fascinating. So people whose fields naturally study individuals might notice that on an individual level, people’s attitudes towards violence are changing radically. Perhaps someone in 1800 would have been far more likely to endorse violence as a legitimate response to something than someone in a survey today. And then it’s very natural to go from that and say, “Well, doesn’t that strongly suggest that war must be on the decline?”
And you’re saying, not necessarily — because war is mediated by many things, not just individual dispositions. It’s about behaviour in groups. Maybe the incentives for states have changed and the technology has changed, and all these other things have changed that could cause war to be even more common and more destructive, despite people disliking it even more than they did, oddly enough.
Bear Braumoeller: Well, it’s not even necessarily that those factors would be outweighed by other factors. My colleague at Ohio State, John Mueller, is another one of the folks who argues in favour of decline of war. And I think he makes a really good point, which is that antiwar sentiment — the idea that we have to abolish war — really reached its peak between World War I and World War II. I mean, the 1930s was the heyday of the “Let’s make war illegal” movement — the Kellogg–Briand Pact and so on and so forth.
Rob Wiblin: Did it work?
Bear Braumoeller: It didn’t work. That’s the thing. And John’s reaction to that point is, “Well, that has to do with Adolf Hitler. It was just one guy didn’t get the memo.” Well, arguably…
Rob Wiblin: But it seems like a peace that fragile…
Bear Braumoeller: Yes. If one guy not getting the memo can cause World War II, that’s a big problem. Then we shouldn’t be studying what people in general believe at all. But also, you could make a good argument that the antiwar sentiment in the 1930s made Hitler’s rise easier: that people were complacent; they thought that they’d managed to find a lasting peace. And as a result, were less willing to stand up to it.
Rob Wiblin: So this is a great example of the difference between individual attitudes and then group outcomes. Where you can have a situation where a large group of people become very opposed to war, and so they disarm themselves. And this creates an incentive for a wolf to enter the scene, basically, and think, “Wow, this is great. I can hunt all of these sheep and get away with murder.” And so maybe that happened, maybe it didn’t. But it demonstrates the point that you can’t necessarily go from individual attitudes to understanding global trends.
OK, let’s get into the weeds a little bit and acquaint ourselves with the dataset you are mostly using in the book, which is the data from the Correlates of War project — which I learned has been running since 1963, and I think in the early 2000s had to move to Penn State?
Bear Braumoeller: I think it’s now at Appalachian State or UT Austin, it just moved just recently. It’s under new governance.
Rob Wiblin: OK, so it’s the child of many parents over many years. A lot of time has gone into it. What time period does the dataset cover?
Bear Braumoeller: There are actually multiple datasets. There’s a dataset on Inter-State Wars, which just covers wars. There’s a dataset on Militarized Interstate Disputes, which covers the threats, displays, and uses of force. And then there’s some less used datasets. There’s the Extra-State Wars dataset, which is a dataset of wars between countries and non-country entities — like colonies, for example. There’s a dataset of Non-State Wars, which are wars among non-state entities. And then there’s a dataset of Civil Wars.
Rob Wiblin: So a lot of different categories.
Bear Braumoeller: To answer your question, they all cover the period 1815 to something like the present — like 2010, 2015.
Rob Wiblin: OK, so there’s all of these different categories of conflicts. As I understand it in the book, you just focus on state-versus-state conflict, basically. One reason for this is that people who study these issues think that it’s possible to have quite different trends for state-versus-state international conflicts as opposed to civil wars. You could have civil wars increasing and state-versus-state conflict going down, or vice versa. They tend to follow different dynamics. And you’re excluding state-versus-non-state actors and vice versa. Is that also right?
Bear Braumoeller: Well I do bring those datasets in at one point, just as sort of a robustness check, but for the most part I didn’t use them. And the reason is, we’re not interested in the frequency of conflict initiation over time — just the raw count — because the international system gets bigger. So just naturally the number of conflicts is going to increase as the number of opportunities increase.
What we want to look at is the rate of conflict initiation, which is the number of initiations divided by the number of opportunities. Now, to know the number of opportunities, you have to know how many actors there are. And there’s the clear answer to that in the data on the state system. There is no clear answer to that in the data on extra-state wars: they don’t have any data on the possible wars or wars that didn’t happen, and there’s no comprehensive list of non-state actors during that time.
Rob Wiblin: Or potential non-state actors.
Bear Braumoeller: Right. Exactly, Exactly. I mention at one point in the book that prior to unification, India was made up of something like 525 princely states. Do we want to say that the British could have gone to war with all of them, each of them? So that gets to be a conceptual model. So one reason is that the kind of analysis that I was doing just wasn’t possible outside of the interstate data. Another is that specialists in the field, for better or for worse, tend to treat interstate conflict as a distinct thing — that the causes of civil wars or the causes of extra-state conflicts tend to differ, and the dynamics of those conflicts differ.
Rob Wiblin: For my purposes, I’m particularly concerned about great power conflict and the risk of nuclear war — or a war so large that it could really take humanity off track as a whole. And I think for that purpose, state-versus-state conflict seems by far the most likely possibility. So it’s actually great from my point of view that that’s the focus area. How big a problem is it that all of these datasets start right at the point that the Napoleonic Wars end, and an unusually peaceful period of European history begins? It seems like that might skew the trends a bit.
Bear Braumoeller: It’s funny. I’ve seen this criticism, and to be honest, I’m not sure I get it. I think the idea is that this whole 100-year period of relative peace, the 19th century, is like a statistical fluke. But the point of using statistical tests is not to be misled by statistical flukes. So I’m trying to analyse the data in a way that would account for the possibility that this is just a glitch. Now, for Pinker’s thesis, I think the existence of that peaceful period in European history is a real problem. And I think he glosses over it. He mentions it only very briefly, I think on one page.
Rob Wiblin: I guess the concern might be that I assume that they’ve actually chosen 1815 because that is when the Napoleonic Wars ended. It’s not a randomly chosen start date. And so the fact that it’s not a randomly chosen start date potentially negatively biases the beginning of the dataset. Now, I suppose if you’re doing the statistical tests properly, then hopefully that should take that into account, and it shouldn’t really matter whether you begin it at the start of the Napoleonic Wars or the end of them. Hopefully, the findings should be robust regardless.
Bear Braumoeller: I like that way of summarising the issue. From that perspective, what I’d say is that most other datasets start at 1945. And the reason that they do, first of all, is data availability. But also there’s this sense that the years after a major war are coherent in a way that they might be dissimilar to the years prior to that war in fundamental ways. So they want to start at the beginning of a new international system or international regime. So from that perspective, just the fact that they started at the end of the Napoleonic Wars doesn’t necessarily mitigate one way or the other in terms of conflict. As you saw, after 1945, rates of conflict worldwide were actually quite high at the initiation of a new international system. And there’s no reason that that couldn’t have been the case in Europe as well.
Rob Wiblin: Yeah. OK, so we talked about how there’s a challenge counting the number of actors in this system. You talked about the interstate system. So, going back to how many states were there in 1916 — not a trivial question to establish — but even having established how many state actors there are at any given point in time, how do you count wars?
So I’m wondering, for example, when in the Second Gulf War you had the US invading Iraq, but you also had the UK invading Iraq. So is that one war or is that two wars? And then if you start thinking about World War II — the sheer number of different states that were in conflict with one another at different points and switching who they were in conflict with sometimes — it seems like counting the number of conflicts occurring is itself quite challenging.
Bear Braumoeller: The two datasets that I primarily use do these in different ways. In the interstate war data, one war is one observation. So World War II was just one conflict. The militarised interstate dispute data counts every bilateral use of force as a case. So they do it both ways. I take the point that, looking at World War II, it’s a ridiculous number of conflict initiations.
I’m speculating a little bit, but I have some basis in the sense that the Correlates of War project was born at the University of Michigan and that’s where I did my graduate training, so I knew the Correlates of War folks pretty well. I think they would argue that states are agents, and that they have autonomy: that they can choose whether or not to become involved in a war. And what they want to do is study that choice. Sometimes states will make a contingent promise to go to war, like, “I’ll form an alliance with you, and then you get into a war and I’ll get dragged into the war with you.” But that’s no less my choice in a sense, right?
Rob Wiblin: I see. So I guess if you have a war where there’s 10 allied parties who all declare war on a group of another 10 allied countries, then there’d be 100 wars going on, by this measure.
Bear Braumoeller: There’d be one war, but 100 uses of force.
Rob Wiblin: 100 uses of force. I see.
Bear Braumoeller: Because they’re not fighting each other, they’re just fighting across. So it’s 10 plus 10 plus… you had it right. One hundred, yeah.
Rob Wiblin: OK. Another thing that I’m not quite sure how it gets treated is: during the 19th century, as you were saying, there was a point at which there were, in a sense, 500 small states in India — because it wasn’t a unified actor. And then for some period of time, Britain ended up in control of almost all of it. By one measure you might think that’s going to result in something like 500 wars. But in this dataset, I think it doesn’t show up that way. How would that appear?
Bear Braumoeller: Well, first of all, it wouldn’t show up in either of the interstate datasets at all. It would show up as a colonial war, which would show up in the Extra-State Wars dataset. And in that dataset, one war is one war. It doesn’t look at initiations of use of force. So even if the British did go to war with all of the princely states in India at the same time, it would only be counted as one conflict.
Rob Wiblin: As I understand it, each of those princely states don’t meet the threshold for being a state within this dataset. What deaths are included? It doesn’t include civilian deaths — we’re talking about battlefield deaths, because that’s far better measured. Is that right?
Bear Braumoeller: That’s right, yeah. Even in recent years, there have been tremendous debates about how many civilians were killed in a war. So they just decided at some point that they had far more reliable records on battle deaths than they did on civilian deaths. That was sort of a pragmatic decision.
Rob Wiblin: So in terms of trying to understand trends in warlikeness, it’s fine just to use battlefield deaths — so long as the ratio of battlefield deaths to civilian deaths is roughly constant. I guess it depends again on what exact question you’re asking, but in terms of how worried we should be about war, if far more civilians are getting killed now than in the past, then that increase in source of concern might get masked in this data. Because those civilians aren’t being counted, because we’re only trying to count battlefield deaths because that’s more practical. How much of a concern is that?
Bear Braumoeller: Without concrete data it’s hard to say. But there are a handful of these issues. Tanisha Fazal actually wrote a terrific article about battlefield medicine in war and pointed out that her criticism of Pinker’s book was that he failed to take into account the fact that between 1815 and now your odds of surviving being shot in battle have dramatically increased. You look at the statistics on the US Civil War for example, and you lost a ridiculous number of people — not to the primary injury, but to the disease that followed.
Rob Wiblin: Right. Bacterial infection.
Bear Braumoeller: Exactly. So we’ve taken great strides in improving battlefield medicine. And her argument is that part of the reason he sees a decline is that he’s ignoring the fact that battlefield medicine has gotten better. The disturbing thing, from my point of view, is that I don’t find a decline, and I do believe that battlefield medicine has gotten better. And if you put those two things together, it sounds like we’re trying to kill even more people, but we’re only succeeding at the same level on balance.
Rob Wiblin: And just failing. Pulling all of this together, how much should we trust this Correlates of War dataset? I suppose you’ve used it because it’s the best thing going. It was the best thing that you had at the time in order to try to answer these questions. But how much further do we have to go towards trying to produce the perfect dataset that we could really trust?
Bear Braumoeller: Oh, I don’t know that we could get there. I think political scientists are going to argue no matter what. But it’s a great question. As you said, the Correlates of War project started in the 1960s, and some of the ways that it was set up, people have argued that they’re too restrictive. The coding rules for who belongs in the interstate system are, especially in very early years, extremely West-centric and Eurocentric.
I think those complaints are totally fair. And that’s something that I think has to change before too long. I mean, there have been enough criticisms now that COW really needs to revisit their state membership criteria and expand the dataset. The challenge is that since the 1960s, other datasets have been built on the membership list for the international system. So there’s a degree of institutional lock-in, but they could at least make a start.
On the other hand, it is a well-funded effort. It’s a lot of scholars from a shocking range of sources. It’s benefited from an amazing array of expertise. Lots of scholars have had eyes on it over the last six decades and have been able to complain to the Correlates of War people about codings. So I think the data that are there, you might not like what the Correlates of War project does, but to me it does what it does pretty convincingly. At least we can be fairly confident that major errors that would’ve been flagged by the conflict studies community have been corrected.
Rob Wiblin: OK, pushing on. As you talked about earlier, a graph of war deaths is always dominated by a handful of especially bloody wars. They just stand out as peaks and then sometimes other wars just kind of disappear on the y-axis. In the book, you describe how this has potentially led people to see patterns in the data where none exist. Can you explain that phenomenon?
Bear Braumoeller: So wars tend to be very big, very rare events. And depending on when the biggest ones happen, it’s really easy to see trends when there are none. So just imagine that over the past two centuries, the timing of wars has happened essentially at random. And the lethality of wars can go from 1,000 battle deaths — which is the minimum criterion — all the way up to World War II. But when that happens is essentially random.
So depending on when they happen, it’s really easy to see trends where there are none. If World War I and World War II happened at the beginning of the period, you’re going to say there’s a downward trend. If they happen at the end of the period, you’re going to say there’s an upward trend.
We usually misjudge trends for one or two reasons. One is that wars are really rare, and we haven’t seen many of them recently. This is like people cancelling their earthquake insurance because there haven’t been any earthquakes recently. There will be, just wait. But they go a long time without an earthquake and it seems like it’s a safe idea. And the other one is that recent deadly wars tend to make us overcompensate in the other direction. That is, we tend to be far more worried about big wars happening than probably we should be.
Rob Wiblin: So the fact is that wars follow this thing called a power law — which we’ll come back to — which is basically just saying a small number of wars cause most of the deaths. But everyone who studies this is aware of this. I think Steven Pinker is surely aware of this; he’s a very smart guy. Doesn’t seem like this issue is rocket science, at least for someone with statistical training. How is it that people are potentially over-reading trends in the data when this is just such a clear problem?
Bear Braumoeller: Well, now, it’s funny. Power laws are really hard. I was going to say even if you have statistical training, but sometimes even especially if you have statistical training. They fool the naked eye in the way that I described. They can also fool standard statistical tests. Anyone who’s taken a statistics class in your audience is probably familiar with the central limit theorem, which is the theorem that ensures that if you have a sample, the average of that sample will be pretty close to the average of the population. So you can poll 100 people, as long as they’re a representative sample, you can ask them “How much do you like the president?” and get a nationally representative answer (within certain percentage points). But with the central limit theorem, people often don’t read the fine print.
Rob Wiblin: It makes a lot of assumptions.
Bear Braumoeller: Only two. One is that you have at least, I think, 30 observations or something like that as a rule of thumb. And the other one is that the distribution of the data is not too far from normal. And power law distributions are very, very, very, very far from normal. And this was something I ran into that I hadn’t anticipated when I started the project, but all of the tools that you learn in a statistics class end up not being useful for measuring and testing trends in warfare.
Rob Wiblin: You just need to go and find completely new tests. Maybe just invent new tests.
Bear Braumoeller: Yes, both of which I did in the course of this project. But, let’s put it this way: I thought I understood power laws before I started this project. And what you saw with me being really horrified at the end of chapter 5 — saying, “This is very, very bad. We need to pay attention to escalation. It’s way, way worse than I thought it was” — was kind of the education that I got in the course of writing the book and getting a better handle on the mathematics of power laws.
Rob Wiblin: So to get around this issue of accidentally inferring a change where none really exists and it’s just a chance difference, you use something called “change-point analysis” to test statistically whether there are different periods in history with greater and lower rates of interstate violence than could be explained by chance alone. I suppose you’ve chosen the specification that works around or tolerates the fact that deaths or occurrence of wars can be clustered and distributed in a power law. Can you explain how those tests work?
Bear Braumoeller: I’m not sure you want me to, but I’m happy to. Let’s say you measure some characteristic across two groups — like the heights of men and women. You’ve got a group of men, you’ve got a group of women, and you measure their heights. And you’ll find that on average there’s a difference. Men are taller. But you want to know: is that difference bigger than what you’d expect to see by chance?
One way to answer that question is to go to your statistical textbook. There’s a statistical formula for it; you can run a test. But there’s also a more intuitive way to answer it — a lot more work, before computers, but it exists. You just reshuffle the people into new groups of the same size thousands of times. Then you measure the difference in heights across groups, thousands of times. And that tells you how much height varies by chance. You literally randomise the members of the group and then see how much their height varies. You do that lots and lots of times.
Rob Wiblin: And then you see how much the actual sample stands out. Is it really far on the tail different from what you’d get.
Bear Braumoeller: Yes. Exactly.
Rob Wiblin: I was taught this under the name “bootstrapping.”
Bear Braumoeller: That is close. “Permutation test” is what it’s called. And it’s the same family of resampling inference. Bootstrapping and permutation tests, I love both of those. They’re both in my top five favourite tests. They’re intuitive, they’re easy to do, they’re hard to fool. They’re really, really handy.
Anyway, so the technique that I used, in a nutshell, picks different points in time and then uses a permutation test to compare rates of conflict initiation before and after those points. It just throws CPU power at the problem, and it comes up with the points at which there was a statistically significant break or change in the underlying rate of conflict initiation.
And some of those points I was struck by when I looked at the results of the analysis. Just looking at the graph was what made me think about international orders. Because one of the change points happened in the early 1850s, which is right around the time that the Concert of Europe disintegrated. Another one happened right at the end of the Cold War. And so those were big flags: “I think I know what happened in those years.”
Rob Wiblin: “1989, you say?”
Bear Braumoeller: Exactly. Exactly.
Rob Wiblin: I suppose it’s very reassuring when the statistical tests, you haven’t told it anything about World War II or the Cold War or anything, and yet it’s picking out almost exactly these breakpoints that you might expect based on your gestalt knowledge of international relations.
Bear Braumoeller: Right, exactly. That was very reassuring.
Rob Wiblin: Yeah. These change-point analyses, and these simulations that you’re running to see how far away from a random distribution you are, do they face a multiple testing issue? Just to explain this: as I mentioned, you’ve got the period from 1800 to 2000, and you’re basically trying to split it up into different periods. And you could choose any of these different years as breakpoints to establish periods. Say you’ve got the period 1800 to 1850, and then 1850 to 1920. Say those are two different periods, and then you’ve got a third one perhaps after that. It seems like in theory you have so many different options, that you could test so many different possible breakpoints, that it’s a bit unclear what the threshold should be given that you’ve asked so many different questions. “Is there a breakpoint here? Is there a breakpoint there?”
Bear Braumoeller: This happens a lot in this literature. You get people saying, “I want to run a test of interstate wars and see if there’s a change in this period, or a change at this point, or a change at this point. Now let’s add civil wars and see if there’s a change at this point, a change at this point.” And then run lots and lots of tests, and then one of them comes up positive and you sort of declare victory.
Rob Wiblin: What a surprise!
Bear Braumoeller: Right. There was one, I’d have to look at the exact numbers, but I came across one that ran like 21 different configurations, I think, and found three significant differences, and said, “Hey, we found the answer.” So, that’s definitely an issue in this literature, within this test, that’s taken into account, just by virtue of the way the test is constructed.
Rob Wiblin: I see. Because when you do the simulation to see whether there are changes, you are effectively doing exactly the same process, asking the same number of questions in that simulation as in the real test. And so it picks it up. OK, that’s really good.
Bear Braumoeller: What I’m doing is downloading a statistical package in R that was written by people who are smarter than I am and who knew how to create this test in the first place.
Rob Wiblin: This is a slightly separate issue, but while we’re on these methodological questions, a lot of people believe that rates of violent death before the rise of modern states and empires — like the Persian Empire, the Roman Empire, and so on — that rates of violent death, including homicide, was far higher than it is today. So now we’re kind of talking about hunter-gatherers compared to the Roman Empire. How confident can we be about that claim, that over that very long time periods, rates of interpersonal violence have declined?
Bear Braumoeller: So this one is a bit outside my area, but just based on looking at the evidence, it strikes me as being really debatable. The archeologists who have responded to this question have argued about things like “How do you know when someone’s killed in battle versus just buried with a spear next to them?” — that kind of thing. And you get heated debates over that question.
From my perspective, the key thing is that small groups naturally produce higher variation in outcomes than large groups. A good example from recent years is you look at maps of COVID prevalence by county. Probably everyone here has seen one of these maps, where you see there’s a lot in New York City, there’s less in Philadelphia, that sort of thing. But when you look at the western part of the US, all of a sudden you get this incredible array of checker-boarded rates: some are really high, some are really low. And that’s because some of those counties have 10 people in them. And if 10 people get COVID in that county, then…
Rob Wiblin: 100%.
Bear Braumoeller: So what you see in those counties is more variation. At least in my data visualisation class, my students are always trying to read something meaningful into that pattern in the map data, and say, “Oh, what is it about this county?” I’m like, “It’s small, that’s what it is.” That’s what’s going on.
Rob Wiblin: Let’s just pause and emphasise this, because this is an incredibly important thing for people in the audience to have in mind when people make a very wide range of claims.
A really interesting case where this shows up is if you look at rankings of schools in terms of test performance or learning or any measure of quality. You’ll very often find that the ones at the top are very small schools. Now a very natural interpretation is that small schools are better. Maybe education is better provided at a boutique scale rather than an industrial scale. However, you absolutely have to check that the worst-performing schools are not also small. Because of course when you have a tiny cohort, it’s possible for them to knock it out of the park one year and then be absolutely garbage the next.
Small schools and small anything are going to have far more higher variance, far more dispersed outcomes. And you need to look at the full range, rather than selecting on the dependent variable — or, in this case, selecting based on high test results or low test results — because that is going to absolutely wreck your conclusion. And the application here is that, of course, pre-state societies — hunter-gatherer groups — were far smaller. So it would be possible for them to have 10% of their population die in a conflict in a given year, in a way that’s extremely difficult to happen in the United States, given its size today.
But doesn’t that also mean that some of the most peaceful-seeming societies, on average, it should cancel out? And both the most violent and the most peaceful ones will all be hunter-gatherer groups, basically?
Bear Braumoeller: Yes. And this is actually where Pinker gets accused of cherry-picking the data, because archeologists have come up with a whole bunch of, “Why didn’t you talk about the Hopi?” for example. “Why didn’t you talk about these other groups that had incredibly low rates of death by warfare?” Again, this is not my area, but that’s part of the criticism.
Rob Wiblin: So the concern is that there is going to be very, very different levels of violence between different hunter-gatherer groups, perhaps just by chance or maybe in different places, different cultures. You could have one group that just has an extremely pacifistic culture, but maybe if you came back in 200 years, it would be different. So you have to make sure that you are in no way selecting which ones to consider based on anything that in fact is correlated with their violence rate.
Bear Braumoeller: Right. I just want to back up to something you said a minute ago, which is that the rates can be very high and very low. We’re talking about a phenomenon — death in warfare — that’s relatively uncommon, and it’s bounded at the bottom by zero. So mostly what you see in these data errs on the positive side: you see unusually lethal. It may be a lot more difficult to tell a relatively peaceful small society from the others.
But the way that I tried to illustrate this in the book — and it either gets a lot of laughs or it gets a lot of blank stares, but we’ll roll the dice here — is: I took a look at CDC data on deaths by poisoning for US states in 2010. And I broke it down by state: West Virginia, Kentucky, and so on and so forth. And then I broke it down by county. And you see almost exactly the same pattern of lethality coming out of those data that you see coming out of Pinker’s state-versus-non-state society data.
So you see there’s one county in West Virginia that has a terrible opioid problem, and people are dying by poisoning in droves. But you would not conclude on that basis that living in a county is more dangerous than living in a state. Since counties are nested within states and most people do both.
Rob Wiblin: Everyone is in one of both. Yeah.
Bear Braumoeller: Right. Well, the commonwealth people would disagree; they would say that some of them are in commonwealths rather than states. But anyway, that’s the overall point: that you have to account for the difference in size of the groups if you want to compare them.
Rob Wiblin: I got confused with this one a little bit, because it seems like assuming that all states and all counties had an equal population, then the average death rate in the counties has to be the same as the average death rate in the states, doesn’t it? If you’re considering just exactly the same people? And yet it looked like the death rate was higher in the counties than in the average of the states. And I think that could only be explained if there was a correlation between population and the death rates from poisoning?
Bear Braumoeller: No, no. It’s just that, as you say, you get more variation in the counties. They average out to something close. In the same way that you can have a group with people with extreme political views, and on average it’ll come out to be sort of moderate. The variation within groups can be a lot bigger than the variation across groups. And that’s what’s implicitly being compared here.
Rob Wiblin: Right. But if you had a complete accounting, if you took all of the states and all of the counties and compared them on average, then you would have to say that the death rate was the same? Actually, no, not necessarily — because of this issue with correlation with the population. Sorry.
Bear Braumoeller: If you added up all of the counties within a state, then yes, you would have the number of deaths in that state. But essentially this is analogous to what Pinker is doing, because he’s talking about very small non-state societies that then came together to form a larger state. So the reasonable comparison would be the death rate within all of those societies prior to the formation of the state to the death rate following the formation of the state.
It’s funny, I didn’t spend a lot of time on this, and it’s funny that it kind of lit me up to the extent that it does. Because I actually believe he’s right that states do decrease the odds that you’re going to get killed. That’s part of their job: monopoly on the legitimate use of force, right? So he managed to get me disagreeing with him, even though I agree with the fundamental point — and it was all because I looked at the data and I thought, “You can’t say that.”
Rob Wiblin: The methodology is flawed even if the conclusion is probably right.
Bear Braumoeller: Yeah, yeah.
Rob Wiblin: OK, so as we’ve flagged, the two key measures of warlikeness that you talk about in the book are war initiation and then war escalation, and both of them matter. They’re different measures of warlikeness. And plausibly, they could go in different directions, and that’s one reason why we need to look at them separately. You could have a trend that wars start less often, but when they happen they tend to escalate far more. So we’ll consider them one by one. Let’s do initiation first. How do you define war initiation, exactly?
Bear Braumoeller: It depends on the dataset. For the war dataset, you don’t even count the war at all until it reaches a threshold of 1,000 battle deaths. So you have to have been fighting a war for a while for that to count as a war. I’m not sure whether this is true, but University of Michigan lore has it that when the people who founded the Correlates of War project wanted to figure out where to draw the line, they just did a histogram, and they saw a big gap around 1,000. And they looked at the wars that were on the low side of that, and they said, “I don’t really think these count as wars. They’re not what I would call a war.” And then the ones above that did, and so they thought, “Eh, 1,000 looks about right.”
Rob Wiblin: Sounds like a data artefact. Sounds like a coding issue.
Bear Braumoeller: But it’s also being in touch with your data, and not being bound by a priori theoretical commitments. But anyway, that’s where the 1,000 battle deaths, I believe, comes from.
Rob Wiblin: I think you test that threshold and find it doesn’t really make much difference, one way the other.
Bear Braumoeller: Right, right. What’s funny is, for a while, the Falklands War was at like 980 or something like that. And that was just a magnet of criticism for the COW project. And so, I don’t know if they recounted, but now, magically, it’s at 1,000 battle deaths.
Rob Wiblin: God, that is grim. “We’ve got to find 20 more deaths. Come on, let’s scrounge them up.”
Bear Braumoeller: Right, right, right. “Keep looking through the records.”
So that’s the way that the militarised interstate dispute data looks at threats, displays, or uses of force. I only look at uses of force — which is a smaller subset, and that, I think, is the threshold that we’re most interested in.
Now, I presented an early draft of this paper at a conference, and it got written up in a couple of places. And one reporter called Pinker and sent him the paper and said, “What do you think?” And he said, “He uses militarised interstate disputes, and the problem with that is that a militarised interstate dispute doesn’t necessarily contain the threat of escalation. It may be that there’s no realistic chance that it’s going to escalate to a war. It could just be the US with a warship, floating off the coast of some country, lobbing missiles into the interior, and they can never respond.”
I saw that response, and I thought that’s fair.
Rob Wiblin: Yeah. Could be.
Bear Braumoeller: The reason for using militarised disputes is that there are more of them, and they at least potentially contain the seeds of war. They’re wars that just didn’t then escalate. So I actually changed my coding rule at that point, to look at only reciprocated uses of force, because to my mind, that’s the best way that I can think of to argue that this had the potential to blow up into an actual war. So, I deviate from the MID coding for that reason: because Pinker criticised what I was doing, and I thought, actually, that’s a pretty good point.
Rob Wiblin: Helpful.
Bear Braumoeller: Well taken.
Rob Wiblin: Yeah. So, to define a rate of initiation, you both have to say, “What is a war?” And then, you also need to know how many opportunities for war there were. And for this, as we talked about earlier, you’ve got this setup where the Correlates of War says how many states are in the interstate system. In the early period, around 1816, as I understand, there’s only 20-something states, and they were almost all in Europe, within this dataset. And gradually, it expands towards, in the 1920s, I think it has something more like the number of countries that a person might intuitively think exist. Here, we use dyads.
Bear Braumoeller: Two states, yeah.
Rob Wiblin: Basically you have two states. I suppose a dyad would be two people talking, and in this case, two states that might be able to fight one another. For global powers — like the US or China today — you’re saying that they could plausibly go to war with anyone, because they just have the reach to do that. For non-great powers, you have a measure of physical distance across land and across sea. So anyone you border with you could go to war with, and anyone who you can reasonably get to from some distance with your navy, then they could also go to war. But you don’t think it’s plausible that Paraguay might go to war with Myanmar — that’s excluded, or that wasn’t a realistic opportunity for war to break out in that year.
Bear Braumoeller: What you’ve just described is sort of the standard way of coding what are called “politically relevant dyads.” I do define opportunities in that way as a robustness check. The main definition that I use is similar — in that the major powers can reach just about anywhere — but the difference is that it’s a continuous measure, and it’s estimated from the data rather than assumed. So, if you’re a smaller state, your immediate neighbours might be in your politically relevant neighbourhood, as well as potentially a country that’s 1,000 miles offshore. As you get stronger, the radius of politically relevant dyads increases. As you get weaker, it shrinks.
This was a measure that a coauthor and I came up with in a separate paper. When I was thinking about how to define politically relevant dyads, I just looked to see if it did a better job of separating wars from non-wars. One of the criticisms of politically relevant dyads is that a fair number of conflicts occur between theoretically non-relevant states, and our measure does a considerably better job of capturing those conflicts. So that was the measure that I used for opportunities.
Rob Wiblin: Excellent. So in the book, you consider a whole lot of variations in the specifications and different kinds of tests that one could run to see how rates of war initiation have shifted over time. If I recall correctly, I think some of those changes do affect the interpretation that you might make. So to some extent, the answer to this question is going to have to be a simplification, relative to the full amount of content that’s in the book, but what are the patterns in war initiation from 1815 through roughly today?
Bear Braumoeller: The patterns in war initiation are difficult to tease out — and for that reason, I put a lot more faith in the patterns in militarised dispute initiation, just because there are more data. It’s a more granular test, so you can actually see.
Rob Wiblin: And is the issue there just that because it’s a lower threshold, there’s a larger sample?
Bear Braumoeller: Yes. So you don’t see a downward trend in either one, in general, across the last couple hundred years, even accounting for noise. Pinker’s pretty clear that there’s going to be statistical noise, and that’s why we’re running these tests. The militarised interstate dispute data show a gradual increase in the rate of conflict initiation over a couple centuries. We’ve talked about the dataset and inclusion rules, and why there may not be reasons to take that completely literally. It excludes other countries that might have been fighting at the time. But you’re very hard-pressed to believe that there was a decline over that time.
There was, however, at the end of the Cold War, a drop in the global rate of conflict initiation of more than 50%. Which, for someone who studies conflict, this is enormous. We’re used to trying to find causes that produce, I don’t know, a 5% change in the probability of conflict initiation between some dyads. To find a 50% decrease in the rate of conflict initiation across the entire international system, it’s like going out fishing and catching a whale. It is a substantively huge drop. But that’s the only point at which you can point to the data and say, “Yeah, I do see a drop in conflict.”
Rob Wiblin: OK, so that’s a clear breakpoint, which makes a whole tonne of sense. What are some other plausible breakpoints that show up sometimes, but maybe not every time?
Bear Braumoeller: Depending on the exact formulation, sometimes you’ll see an increase in conflict in the 1860s — which makes sense, because that was the period between the Concert of Europe and the Bismarckian system — and then a slight drop after that. Sometimes you see a step down after the very early Cold War period, and it’s right around the time of the Cuban Missile Crisis. And again, this sort of makes sense in terms of what you see substancefully, in terms of global patterns: that the US and Soviets, after the Cuban Missile Crisis, really start to become much more cautious about sabre-rattling, and they initiate arms control agreements, and so on and so forth. So in some formulations, you can see those breakpoints as well.
Rob Wiblin: Right. That makes sense. One thing that just occurred to me is that I think in all of the statistical tests here, as I understand it, the null hypothesis — so the thing that you are comparing your theory of it having changed with — is that war initiation is a stationary process, which is to say that any relevant dyad has an equal probability of initiating a war in any given year.
Bear Braumoeller: No, no, no. Aggregated across the international system, the rate of conflict initiation equals out. So in one year, the countries starting with the letter A might go to war a lot, and in another year, the countries starting with letter B might go to war a lot. Right?
Rob Wiblin: OK. You’re looking at the aggregate level, the total number of conflicts in a year. OK.
Bear Braumoeller: Yeah.
Rob Wiblin: Yeah. Sorry, that’s what I meant to say, but not what I did say. But wouldn’t it be very natural to think that wars cause more wars? I think the technical term for this would be that wars are autocorrelated. That is to say, if you have lots of wars started in 1940, then that conflict might cause more wars to start in 1941, and so that would lead to even more clustering than you might expect from just the illusory clustering of wars. That might cause wars to actually cluster, rather than merely have the illusion of clustering.
Bear Braumoeller: Right, right, right. And you definitely see that in the dyadic data. What you’re calling autocorrelation, as a statistician would describe it, the conflict studies people call “war contagion”: where war spreads across borders. But that’s definitely a phenomenon that happens in some wars — not most militarised disputes, but some.
Rob Wiblin: Yeah. Is it a problem that the statistical models are not considering that? I suppose, for long-term trends, it doesn’t really affect it. That’s going to cancel out pretty clearly. But I suppose it could cause it to see more breakpoints than really exist — because in reality, you have a clustering of lots of war initiations right around World War II, and that might be because of the contagion effect rather than a more underlying change.
Bear Braumoeller: This gets into some complicated weeds, but the problem arises from the fact that the observations are not independent of one another, right? And most statistical tests assume that they are independent of one another. Now I should say, that goes for peace as well as war. You could argue that the fact that the US and the UK were not at war in 1894 is not independent of the fact that they were not at war in 1893. So it cuts both ways, and the impact in the aggregate is difficult to figure out.
The way to deal with that is to talk to the network science people because they’re very good. And I’m not part of it. I do a little bit of it, but not very much. One of the things that they’re really good at is modelling and accounting for the lack of independence of observations. So there are ways to deal with it, but you’re right, a naive statistical test that’s doing something more than just looking at trends would have a hard time dealing with that.
Rob Wiblin: Yeah. So I’m approaching this with the main thing that I want to know is: how worried should I be about a great power conflict, or how worried should I be about a massive war over the next 100 years — or, I suppose, particularly during my lifetime? And for that purpose, perhaps I don’t care about the rate of war initiation per relevant pair of states. I might just care about the absolute number of wars initiated, because a world with a lower rate of war per state, but way more states, might create more wars in total — any of which then could perhaps draw in others and escalate and become a massive problem. Do you know how the absolute number of wars starting has shifted over time?
Bear Braumoeller: Yes. I mentioned earlier that the Correlates of War project has data on extra-state wars and non-state wars — extra-state wars being a war between a state and some non-state entity (like a colony), and non-state wars being wars among non-state entities. And at one point in the book, I just aggregated all of those wars together. And I can’t look at the rate, because I don’t know what the denominator is. But you can look at just the absolute count of wars of all three types, when you add them up together over time. And I threw that into the change-point detection algorithm, and it resolutely refused to find any change points at all.
Rob Wiblin: I see.
Bear Braumoeller: It just said, no matter how you tweak the parameters on this test, you’re getting a flat line over 200 years.
Rob Wiblin: Wow. OK, that’s really helpful to know. Maybe the largest concern that I’ve encountered with this is that the interstate data — to some extent because of who’s counted as a state in the very early periods — to some extent, you’re comparing the rate of wars in Europe in the 19th century with the rates of wars breaking out in the entire world in the 20th century.
Bear Braumoeller: Right.
Rob Wiblin: Now, I’m actually not sure which way that biases things, but it’s not ideally what we would like to be doing. But if we actually have managed to count all of these deadly quarrels according to proper standards across the entire world, over the time period, then we can see if there’s a change in the absolute number. And there is not.
Bear Braumoeller: Right. That’s right.
Rob Wiblin: OK, fantastic. Turning now to an alternative measure of the risk of war: escalation. How do you measure the escalatory propensity of wars in different time periods?
Bear Braumoeller: There’s a simple answer and a complicated answer.
Rob Wiblin: Maybe let’s do the simple one.
Bear Braumoeller: The simple one is that you make a graph where you line up the wars in order of deadliness, and then you take the logarithm of deadliness, and what you get is a straight line. And you do that for two sets of wars, and then you compare the steepness of the two lines.
Rob Wiblin: I see.
Bear Braumoeller: And I realise a lot of people aren’t going to have this graph readily. But imagine the probability of a conflict on the y-axis, going from 1 at the top toward 0 at the bottom, and then the lethality of the conflict on the x-axis, and both of them on a log scale — so 1,000 battle deaths, 10,000 battle deaths, 100,000 battle deaths, so on and so forth — the relationship between those two things looks like a downward-sloping line.
Rob Wiblin: So bigger wars are less common.
Bear Braumoeller: Exactly. Exactly.
Rob Wiblin: Yeah. I want to know how much less common are they.
Bear Braumoeller: But they become less common, in an extremely linear way, in log–log space, which is the signature of a power law relationship. It’s also characteristic of other distributions, but that’s what people look for for a power law distribution. So, the idea is you plot two sets of wars, side by side. And the one that has the shallower line goes farther out on the x-axis at the same level of probability, so that set of wars with the shallower line have more escalatory propensity.
So a friend of mine, a colleague — Lars-Erik Cederman, who’s at ETH in Zurich — wrote an article a while ago, going back hundreds of years — 600, 700 years — and looking at a dataset of lethality of major power wars. And he finds that after the Napoleonic Wars, wars actually got deadlier. And he attributes that to —
Rob Wiblin: More escalatory?
Bear Braumoeller: Yes. He attributes that to the levée en masse: instead of having small, professional armies on the field, all of a sudden, you have mass armies. Other historians have attributed it to the railroad, which allows you to move logistics and troops very quickly from one place to another en masse, large numbers of people in ways that you hadn’t before. Could very well be both. I’m not sure. But the main evidence for this is a plot that he has of pre–Napoleonic Wars and post–Napoleonic Wars, and it’s very clear that the post–Napoleonic Wars are more escalatory.
So basically, what I did was look for changes within the post-Napoleonic period to see whether there was any point at which wars got more or less lethal. And I didn’t find any.
Rob Wiblin: So that makes sense to me, but I have the benefit of reading the book and having looked at all the graphs. So I’m going to try to put this into terms without using the terms “power law” or “log–log,” especially because many listeners won’t be that familiar with logarithms and so on.
So what is the escalatory propensity of war? In this case, we’re going to define it as basically the likelihood that a war that kills 10,000 people goes on to kill 100,000 people — the probability that a war keeps getting bigger by some fixed factor. So, going from 10,000 to 100,000 is a tenfold increase, and then 100,000 to 1,000,000, that’s another tenfold increase. And as it turns out, if you look at a whole bunch of wars over some time period, the probability of a war going from 1,000 deaths to 10,000 deaths is about the same as the probability of it going from 100,000 deaths to 1,000,000 deaths.
Bear Braumoeller: Yes.
Rob Wiblin: That ratio is roughly constant. And that means, for those who do understand logarithms, that if you graph the logarithm of deaths as against the logarithm of the number of wars that get to that point, then you get a straight line.
Now, we want to see: has the escalatory propensity changed? So we need to say, what about all of the wars between 1800 and 1900, and all of the wars between 1900 and 2000? Is the probability of the war going from 10,000 deaths to 100,000 deaths in the first group different than in the second group?
Everything else that we’re saying is fancier versions of that basic idea. And in that case, on the graph where you have the log against the log, that corresponds to the slope changing, basically. OK, fantastic.
Bear Braumoeller: Another way that I think of it is: you break wars up into two groups, and in one of them, you find that 10% of the wars cause 90% of the deaths. And in another, you find that 5% of the wars cause 95% of the deaths. In that second group, you would say that war is more escalatory — because the wars that get big get really, really, really big.
Rob Wiblin: Yeah. Fantastic. That’s great. Why do you think wars follow a power law like this? So they have a constant probability of getting 10 times larger — rather than a natural thing to think would be that they run out of escalatory steam as they get bigger: that as the war gets bigger, people get more reluctant for it to expand, or just it runs out of people to kill for heaven’s sake. But we don’t see that.
Bear Braumoeller: No, you don’t. And that’s one of the questions that I’m in Oslo to try to answer, actually. We don’t really have a compelling answer to that question. We have some hunches. I look at a process that’s called the two-sided gambler’s ruin.
The analogy is something like this: you and I start out at the 50-yard line of an American football field. We flip a coin. If it’s heads, we move one yard toward your goal line. If it’s tails, we move one yard toward my goal line. And regardless of which way we go, we each burn a dollar. How long does it take you to get to one goal line or the other? It turns out that if you simulate that thousands and thousands of times, you get a power law distribution. Most of the time, you get to the one goal line or the other fairly soon, but there are some cases where you just get a hurting stalemate in the middle forever.
Rob Wiblin: And it just goes on forever and ever.
Bear Braumoeller: If you think about World War I, for example, that’s very much what happened, right? They would just sort of go back and forth across the 50-yard line, while people were just bleeding away one after another. So, that’s one model of how it is that war has produced power laws.
I was just talking with another person who’s done a lot of interesting research on conflict and on the decline of war thesis, Aaron Clauset. He was also invited here to Oslo, and he and I are sharing an office. So, we’ve had a chance to talk a bit about these sorts of things. His take — along with his coauthor, Kristian Gleditsch — is that in individual battles, forces are fairly evenly met, and so they tend to cancel out. So the only time you see a spike in war fatalities are chance events, and that process could also produce a power law.
Now, the argument that you made about how you run out of steam: what they’re finding, at least in a preliminary way, is that that does happen in civil wars. Civil wars tend to sort of taper off rather than continue in a strong power law way. But in interstate wars, states have more resources to bring to bear. You can always bring more states into the fight, that sort of thing.
Rob Wiblin: So what have been the trends in war escalations since 1815?
Bear Braumoeller: I have found none. Zero, zip, zilch. No matter how you measure it.
Rob Wiblin: It doesn’t matter what test you run? It doesn’t matter what breaks you consider?
Bear Braumoeller: I threw test after test at measure after measure, and I just found nothing. I did a preliminary test, where I looked for breakpoints. And as a matter of fact, that test did take into account the multiple testing problem. That one I coded myself, and I made sure to do it that way.
Rob Wiblin: Yeah.
Bear Braumoeller: I’m so thrilled. I’m just thrilled that you asked me about that because nobody ever asks about that. People will run 100 tests on a dataset, and they’ll find one significant result, and they’ll just be like, “Look.”
Rob Wiblin: Yeah. You can tell that this is a very unusual podcast that absolutely does not have to turn a profit. That’s why we can ask that question.
Bear Braumoeller: Right. Exactly.
Rob Wiblin: We make an epic loss as a nonprofit. But sorry, go on.
Bear Braumoeller: But anyway, even taking that into account, I just wasn’t able to find anything. Now, interestingly, the people who put me on the grant that brought me here, who offered me the chance to be a part of this, they’ve published a paper in which they argue that there is a slight decline in the lethality of war after Korea. We can get into the differences between those two — it’s actually fairly complicated. But I think it’s really a credit to them that they invited me and invited Aaron, who found the same thing. Aaron and I used completely different methodologies and came to almost exactly the same answer, which is just delightful. So, it’s going to be a really neat year.
Rob Wiblin: To try to get on the same page.
Bear Braumoeller: Yeah.
Rob Wiblin: So you have to have a whole bunch of different conflicts in order to establish the escalatory potential, which comes from the slope. So it takes a lot of data to measure. Does that mean that maybe, even if there were a change in the underlying escalatory nature of war, it could be quite hard to pick up? Because you just wouldn’t have enough wars to be able to get a statistically significant difference?
Bear Braumoeller: Yeah. We’re looking at under 100 wars, right? And you’re dividing those into two sets. So you can actually get consistent measures of the slope, even with a small number of observations. The problem is that, as one of those groups gets smaller, your uncertainty gets bigger and bigger. But what I’d say to that point is that if you take a look at the slopes, in the relationships that I found, they’re shockingly close. So, even if we had 10 times as many wars, we wouldn’t find a significant difference.
Rob Wiblin: Yeah. Realising that, even if there were a difference, we can’t tell, is itself a really important finding — because of all we cannot know, we should not speak. So definitely there’s a lot of things that then you can’t say: it’s going up, it’s going down, it’s staying the same. We just don’t know.
OK, I’ve got some quite technical questions here that were sent in by an extremely smart and informed audience member. I’m fascinated to hear what you have to say about them.
Bear Braumoeller: Sure.
Rob Wiblin: I’m super curious to hear the response. The first one here is going to take a minute to read, but I think it’s a good one:
“This one is very in the weeds, but I was very confused about some conflicting results Pinker and Braumoeller get in testing for the hypothesis of a break in war incidents after 1945. Pinker writes, ‘Taking the frequency of wars between great powers from 1495 to 1945 as a baseline, the chance that there would be a 65-year stretch with only a single great power war, the marginal case of the Korean War, is one in a thousand. Even if we take 1815 as our starting point, which biases the test against us by letting the peaceful post-Napoleonic 19th century dominate the base rate, we find that the probability that the postwar era would have at most four wars involving a great power is less than 0.4%. And the probability that it would have at most one war between European states, that is the Soviet invasion of Hungary in 1956, is 0.08%.’ Braumoeller gets different results by modelling the onset of great power war in a given year as a binomial distribution with P = 0.02, based on the rate of great power war over the last five centuries: ‘The probability of observing seven continuous decades of peace is 24.3%.’ He also writes, ‘It would still take about 150 years of uninterrupted peace for us to reject conclusively the claim that the underlying probability of systemic war remains unchanged.’ Both Pinker and Braumoeller are relying primarily on Levy’s War in the Modern Great Power System to estimate the rate of great power war, so I don’t understand why they’re getting such radically different results.”
What’s going on?
Bear Braumoeller: Well, what’s going on, first of all, is you have a phenomenally interesting audience. That really is in the weeds, and it’s a neat question. The answer is actually fairly simple. So sometimes when Pinker’s writing about great power wars, he’s talking about wars involving a single great power, and Levy does measure those. Sometimes he writes about wars that involve most great powers — that goes by the name of a “systemic war” — and Levy codes those as well.
So this discussion was in the context of John Lewis Gaddis’s discussion of the long peace after World War II. And Gaddis’s observation was that it’s really striking that we haven’t had a World War III, given that systemic wars have actually been fairly common up to this point. So at different points in the book, Pinker writes about wars involving a single great power or wars involving most great powers. This quote is from a section where he is talking about wars involving a single great power. Since the context was the discussion of the long peace, I thought the appropriate test would be: how often have we had systemic wars? So we’re actually looking at subtly different things, and that explains the difference in the finding.
Rob Wiblin: I see.
Bear Braumoeller: What’s funny — just as a quick footnote — I mentioned that Aaron Clauset and I had come to similar findings by different means. One of the things I loved when I first read Aaron’s paper was that he was using a completely different test, just looking at things in a totally different way. And at one point he says, “We’d have to go 150 years before we could conclusively say that there’s been a change.”
But the fact that we not only got the same finding, but we also came up with the same number, really is reassuring.
Rob Wiblin: Yeah, yeah. It’s a complete data replication. I guess to people in this field, these might seem like very different questions. To me, as someone who’s not an expert in this field, these both seem like relevant questions that one might ask, and yet they produce radically different conclusions. What should I take away from that?
Bear Braumoeller: If you look in the post–World War II period, you do find fewer wars involving great powers on great powers, right? Part of the reason is that the considerable majority of the great powers were part of an alliance. I do think it’s an interesting thing to look at, but I think it’s very difficult to know what to say about the results. Whereas if you’re looking at big systemic wars like World War I and World War II and the Napoleonic Wars — these sort of huge catastrophic wars that happen on average about once every 50 years — then it’s a little bit easier analytically to ask the question: how unusual is it that we haven’t seen one since World War II?
Rob Wiblin: Not shocking.
Bear Braumoeller: And the answer is, it’s a little surprising, but it’s not shocking. It’s not to the point where we can say there’s been a fundamental change.
Rob Wiblin: OK, we’re going to push on. I should say, there’s another chapter where you consider another measure of warlikeness: the prevalence of different potential sparks for war, different kinds of causes of war. So you could have, say, how many territorial disputes are there active at any given point in time?
Basically, in brief, you find that there’s many different reasons that countries fight, and some of them are becoming more common over time, and some of them are becoming less common over time, and there doesn’t seem to be any great trend there. So it’s very hard to say very much about whether the sparks for war are increasing in prevalence or decreasing. It really just seems like a total mess.
But to wrap up this data analysis section, how can we aggregate all of the above trends to give a sense of how grave the risk of war is overall? How worried should we be?
Bear Braumoeller: So it’s sort of a mixed answer. Like I said, interstate war initiation has gotten far less common since the end of the Cold War, which is great news. From the perspective of making the world a better place, we need to understand that better. This is basically the whole reason that I formed a research lab. We’re doing things like coming up with measures of order and figuring out the logic of the relationship between order and war, and how the tensions between domestic political order and international order play out, and even looking at Chinese understandings of international order to see what kind of traction we can get on that.
So I think one takeaway is that we do sometimes see a big drop in the rate of conflict initiation that I absolutely believe in, and we need to understand what happened there. A second takeaway is that the lethality of war, like I said, no matter what I threw at it, just stubbornly refused to give me the answer that war is increasing or decreasing. People think that if a war starts now, it’s less likely to escalate into an incredibly bloody conflagration than it was in 1939 or 1913, and there’s no evidence to support that conclusion.
The final thing I’d say is that wars are so incredibly escalatory that we should really be worried about them. The analogy that I use is: imagine I have a deck of 96 cards. The most common card has 1,000 battle deaths, but one of the cards is World War I, and one of the cards is World War II. How worried should you be about drawing a card from that deck? You could say, “Well, most of them are 1,000 battle deaths, so I shouldn’t be too worried.” But at the same time, World War I and World War II are in there, and if the deck hasn’t changed, we really need to be thoughtful about when it is we’re going to draw another card.
Rob Wiblin: If we came back in 10 or 20 years, and by then you decided that at least some of the bottom lines in Only the Dead had turned out to be more wrong than right, what would be the most likely reason for the analysis to have drawn some mistaken conclusions?
Bear Braumoeller: I can imagine two reasons. One reason would be the statistical tests. It’s just very difficult to do statistical tests on some of the data that I’m looking at. Aaron and I got the same result, Nassim Taleb and I got the same result — all using different tests, which is really encouraging to me. Nils Hjort, who invited me here, thinks we’re using the wrong statistical tests, and he’s a statistician. Now we’re sort of united in disagreeing, in that we think about how you design a statistical test in different ways. I’m not currently convinced, but it’s possible by the time I get done with this year that I will be. So that would be one reason that I could say: maybe we would come up with a statistical test that is based on a more nuanced understanding of why war happens that would give a different result. That’s very possible.
Another possibility would be data. As we’ve mentioned, there is controversy over the coverage of the COW data in the early 1800s, increasingly. It arguably omits states that belong in the interstate system, in particular non-European states. One of the points that’s been raised is that leaves out wars that happened. It also leaves out opportunities for war. So I don’t actually know what the net result would be. To find out, I’d actually have to have the data. That said, the rate of conflict initiation outside the current system would have to be pretty big for there to be a decline over time.
Rob Wiblin: I see. So you’re saying, say that we got much better coverage of non-European states in some time period. It’s possible that that could then make it look like wars declined or increased. But I suppose, given that you already have a meaningful fraction of the data in the current dataset, those states would have to be following a pretty different trend for it to shift the overall trend for everyone.
Bear Braumoeller: Right. I guess the point I’m making is that people point out that there are wars outside the European system during this period. They’re absolutely right, but what they don’t do is take into account the number of opportunities for war. So if you include colonial wars: there are a fair number of colonial wars, there are a huge number of opportunities for war. So you could actually theoretically drive down the rate of conflict initiation in the 1800s by including those. I just don’t know one way or the other. But to get it up to the level of the Cold War would require a huge lift.
Rob Wiblin: I see. Just out of curiosity, where would the wars of independence fought in Latin America against Spain go? Do they count as civil wars in this dataset?
Bear Braumoeller: No, actually, I think they count as interstate wars. They tended to happen a little later in the 1800s though.
Rob Wiblin: So they get counted.
Bear Braumoeller: Yeah.
Rob Wiblin: Nice. Are there any other issues? I guess those are the main ones?
Bear Braumoeller: No, I was going to say, I want to be clear that what I tried to do in this book was give the best possible answer you can give based on the data. I recognise that there’s a lot of uncertainty over which data to use, and how to code the data, and so on and so forth. And as you saw, I tried to be pretty exhaustive and look at all possibilities. And what’s striking to me is how hard it is to find causes for optimism across all those tests, across all those specifications. It’s extremely difficult to find reason to believe that over the past couple hundred years we’ve seen any kind of secular decline in conflict. I do believe the end of the Cold War was a big decline, and that’s really cool and really interesting. The broader thesis about a longer-term decline is very difficult to support.
Rob Wiblin: Yeah. OK I appreciate you being willing to dive so deep into the details with me there, I hope the audience found all that pretty instructive and it leaves people appropriately calibrated about how confident to feel about various different ideas, appreciating both the strengths of the work and its weaknesses that we’d love to improve with future work. It’s almost 11pm for you over there in Norway so we should have some mercy and let you get home.
Normally, we try to end the show with a more lighthearted question. I’m not sure that we can possibly call this a lighthearted question, but what’s a war few listeners will have heard of that is astonishing in one way or another, that maybe people should go read the Wikipedia article about?
Bear Braumoeller: It’s funny. When you think of wars, you tend to think of the big wars that everybody knows. But there are a lot of smaller wars that just, in some way or another, are so odd and quirky that they surprise you, and in some ways they sort of restore your curiosity about people. The Soccer War is one good example.
Rob Wiblin: I don’t know about that one. I’m so curious.
Bear Braumoeller: It was between El Salvador and Honduras in 1969. Really, the issues at stake were things like land reform and immigration issues — there were already tensions between El Salvador and Honduras — but what sparked the war was mistreatment of one another’s fans during soccer games. The soccer games spawned a crisis. The crisis spawned a war. It was a spark in a powder keg. The powder keg already existed, but the thing that set it off was soccer.
The Pig War is another one that most people have never heard of, in part because the only fatality was the pig. But it was the US and the UK in the 1850s. It was a border dispute that was dormant until an American farmer shot an Irish farmer’s pig. Then both sides mobilised, and they managed not to go to war.
Rob Wiblin: Right. Wow.
Bear Braumoeller: I think my favourite though is the War of Jenkins’ Ear. You cannot hear a name like the War of Jenkins’ Ear and not go to Wikipedia and say, “What the hell happened? Where did this come from?” The answer is, there was a British sea captain named Robert Jenkins, who was caught by the Spanish Coast Guard in the Caribbean. They accused him of piracy, and they were probably right. They cut his ear off. I think it was almost a decade later that British trade interests were lobbying for war against Spain. As part of that lobbying, they paraded the earless Jenkins in front of Parliament, and Parliament was shocked and outraged that Spain had cut this guy’s ear off 10 years ago or whatever it was. That became part of the justification, and that’s how Jenkins’s ear ended up being immortalised in a war. Those are always the wars that I find fascinating for purely storytelling reasons.
Rob Wiblin: I will stick up links to Wikipedia articles on all of those wars and a couple of other ones that are shocking to me. The war between an alliance of Argentina, Uruguay, and Brazil against Paraguay is an unbelievable story.
Bear Braumoeller: Oh, God.
Rob Wiblin: Then the Chaco War, the war between Paraguay and Bolivia, over a place where they thought that there was oil. I don’t know that they’ve gotten almost any oil out of that, but they did manage to lose a shocking fraction of their populations.
Bear Braumoeller: They found oil two years ago or something. But yeah, those two wars are actually, in per capita terms, the deadliest in the last two centuries — deadlier even than World War I and World War II. It’s just amazing. Those are great stories. The Paraguayan president who led his people to the horrendous defeat in the Paraguayan War is now immortalised in a statue, and he’s on the money. You’re right, it’s a terrific story. It’s very interesting. But spectacularly deadly wars.
Rob Wiblin: My guest today has been Bear Braumoeller. Thanks so much for coming on The 80,000 Hours Podcast, Bear.
Bear Braumoeller: Thank you so much. It’s been a real pleasure.
Rob Wiblin: If you liked this episode, I think it’s very likely that you’ll enjoy two other episodes from this year: #128 – Chris Blattman on the five reasons wars happen, and #134 – Ian Morris on what big picture history teaches us.
I’d also like to plug the fact that our job board has just gotten a big update. You can find it at jobs.80000hours.org.
It now has better filters to help you narrow down on roles that are suitable to you, in terms of their location, skill type, and seniority. You can also search for specific terms within the job descriptions themselves.
But really the biggest improvement is that you can now set up email alerts to be told right away about new roles that meet your specific job search requirements.
Let’s say that hypothetically you are looking for early- or mid-career roles related to pandemic preparedness in either London; Washington, DC; or the San Francisco Bay Area.
You can go to the job board, select those filters to see what roles are currently available meeting those criteria, and then click the ‘Set up alerts’ button and put in your email.
Now if any jobs ever come up that meet those needs, we’ll email you to let you know about them as they go up on the job board. This way you can’t accidentally miss any, or find out about them too late to apply.
Of course you can do this for any combination of problem area, location, skill type, organisation, and so on.
This is a really neat feature in my view, so nice work to the people who’ve been involved in this update, which I think is Kush Kansagra, Victor Yunenko, Maria Bækkelie, and Conor Barnes.
All right, The 80,000 Hours Podcast is produced and edited by Keiran Harris.
Audio mastering and technical editing for this episode by Ryan Kessler.
Full transcripts and an extensive collection of links to learn more are available on our site and put together by Katy Moore.
Thanks for joining, talk to you again soon.