Jacquelyn Schneider, the Hargrove Hoover Fellow at the Hoover Institution, directs the Hoover Wargaming and Crisis Simulation Initiative.
Jonathan Movroydis: Let’s begin with this: what is a wargame, and what does it ask of its players?
Jacquelyn Schneider: Well, you actually asked the most complicated question. As somebody who uses or designs them and has worked on creating wargames as a methodology, I had a very regimented idea about what I thought constituted a wargame. I published a paper with Erik Lin-Greenberg at MIT and Reid Pauly at Brown, and it identifies the characteristics of a wargame: it has human players, it has rules, there are consequences, and there’s some sort of immersive scenario. So, great—all these ingredients turn into a wargame.
But then, as I started bringing games into Hoover’s new archival collection, we realized that’s a beautiful academic definition that in no way corresponds to the way we’ve thought historically. “What is a wargame?” includes things that look like political-military games, where people sit around a table discussing strategic crisis situations. It can look more like a board game that simulates campaign tactics or operational concepts. Sometimes they’re video games. And sometimes they’re exercises that have a game component inside them.
Wargames are really an amorphous concept. When I’m designing them, I’m using my very precise definition, but when I’m bringing things into the collection, I’m looking at it more broadly. Is there some element of chance? Is there a scenario that was developed? Is there a game aspect, whether they call it a simulation, a tabletop, an exercise? It’s a more expansive view.
Movroydis: Last year, during your launch of the Wargaming and Crisis Simulation Initiative in Washington, you said wargames are having their moment. Why now?
Schneider: It is a confluence of factors. On the academic side, scholars are starting to think about how wargames are similar to experimental or behavioral methods. It seems that maybe games can help us better understand not just war but also democratic processes, economics, disciplines beyond political or strategic security studies—anything that requires understanding human behavior. There’s also been a proliferation of public wargames publicized by MSNBC, Bloomberg, and the Wall Street Journal, games run by Congress members in order to discuss, say, the implications of a Chinese invasion or blockade of Taiwan.
You also have a movement started under Bob Work when he was deputy secretary of defense, and this has fostered innovation and experimentation in wargames within the Department of Defense.
All these things coming together have led to a real renaissance of wargames.
Movroydis: You’ve recently talked about the Sigma Wargames, parts of which are in the Hoover wargaming archive, and how they were prescient about the events that unfolded during the Vietnam War of the 1960s and the early 1970s. How was that game designed, and were the officials who played it able to anticipate events?
Schneider: The Sigma games are really interesting. They started under the Kennedy administration in 1962 and continued through the Johnson administration until 1967. A few of those games—in particular Sigma I in 1964 and Sigma II in 1964—featured very influential players. At the time they were called political-military games. The reports have a preface saying this is not a wargame in the typical sense—which just goes to show how the definition is contested—and that we’re not going to test out an operational campaign. Instead, the game is just as much about diplomacy, information, intelligence, and economics, the big strategic elements of foreign policy, as it is about the military. This all started in an office called International Security Affairs (ISA), which was where President Kennedy put a lot of his civilian whiz kids. A lot of relatively young academics started and were running the office that included the joint wargames agency.
These games have an action group: brigadier generals, colonels, senior civilians, but not cabinet-level. They routinely meet every day for two weeks, and they’re making policy recommendations in the game to a cabinet of principals who meet less frequently. These are very senior people in the cases of the two Sigma series in 1964. We’re talking about Curtis LeMay, Air Force chief of staff; Chairman of the Joint Chiefs of Staff Maxwell Taylor; and later, Earle Wheeler, Taylor’s successor as Joint Chiefs chairman; national security adviser McGeorge Bundy; his brother, William Bundy, an assistant secretary of state; and John McCone, who’s the director of the CIA. So, the most senior people in the Johnson administration are playing these games. And then they have a control group made up of both regional experts and people considered military experts.
These are exercises in depth. They’re not just saying here’s what happens, now make a decision. You have this rich scenario. Meanwhile, as the moves are being made, the control team is making decisions about the outcomes of these different choices the players make.
In both of the Sigma 1964 series, they make what’s called tit-for-tat escalation choices: you do this, I do something a little more escalatory. You do this, I do something slightly more escalatory. And the control is making decisions about, for instance, how effective is strategic bombing? What is US domestic opinion? What’s happening in international opinion? And then, most important, what’s happening with the Viet Cong and the North Vietnamese government? Are the attacks by the United States and to some extent the South Vietnamese making any difference to the outcome of the game?
What makes Sigma so interesting is how the games foretold what was going to happen in Vietnam. They showed that any military involvement in the region was going to take a very long time to be successful and be very expensive. The tit-for-tat strategies—which in the second Sigma ’64 series they call “nibbling strategies”—end up being not very effective at taking down the Viet Cong. And the concerns that the control team has about domestic public opinion, and to some extent international public opinion, turn out to be completely valid.
So, here are games that predicted the future. They were played with the top players in the Johnson Cabinet, and yet these figures ignored the results and advocated for the same policies that failed. Why? It’s a puzzle for scholars.
Movroydis: How is the data processed, analyzed, and put into some kind of product that’s useful for decision makers?
Schneider: As academics, we have peer review, so our data is generally extremely transparent. When I published the game series that I ran for three years, not only did I have to publish all the details about the game design, but I also had to publish all the details of my data so people could replicate it. That’s not what generally happens in the Department of Defense.
If you look in the wargaming archive, some of the games are really good about giving you all the data that came with the game—the outcomes, the moves, the actions. These are the types of games that give great insight into why players took specific actions. Generally, we only get this kind of data when there’s a stenographer (usually female) in the game control team. So, as you move into later games in the 1980s, 1990s, and definitely today, most of what you get does not include the raw data or transcription of conversations. Instead, it is after-action reports generated from someone who looked at the data and then summarized it in the report. These reports are polished, but you don’t get to see a lot of where they came up with their answers and any flaws that might exist in the data.
In the 1960s reports, they were taking verbatim notes. You really get to feel the emotional responses players have to the game. You have the notorious quotes such as “bomb them back into the Stone Age,” right? You feel how frustrated General LeMay is in that quote. When you read today’s game reports, they’re summarized, so you lose the evocative parts. You’re missing some of that special sauce.
Movroydis: Is that because they’re classified?
Schneider: No, I think it’s because they got rid of stenographers. But also, some of those candid remarks ended up becoming very embarrassing to people who were playing the game.
As we move into the 2000s, games are becoming a lot more structured and are sponsored by organizations. So, if you’re the Naval War College, you run maybe a few war college games, but mostly you’re running games sponsored by big organizations like the Department of the Navy or Strategic Command. Those institutions all have things they want to get out of the game, so the report is written for the sponsor’s eyes. If something embarrassing is happening—like somebody saying they want to bomb Vietnam back to the Stone Age—that’s probably not going to end up in the after-action report. With the more contemporary materials, we’re not finding a lot of that unprocesssed data is being preserved for the public record.
Movroydis: You’ve been conducting an international-crisis wargame over the past three years. What have you learned from this and other games?
Schneider: The first big series I led was the International Crisis Wargame Series, which looked at cyber vulnerabilities and nuclear command: control and communications. It ran over three years and had 580 players. The players were in teams of about four to six, about 115 of them. We published those results in scholarly journals. Those games found that cyber operations, interestingly enough, do not create uncertainty and fear; instead, it’s the way in which they may imbue misperceived confidence that creates escalatory dynamics. Another game we’re writing up right now is on network structure and how different structures for early warning affect strategic stability. That game ran over two years, with more than 200 players.
The game we just finished is on artificial intelligence and how it affects the US-China crisis. For example, if we introduce a new AI-enabled weapons system, how do people choose to use it in a crisis? And then, let’s say something goes wrong with it: how does that affect how they continue to use the weapon system, and what does that mean for US-China crisis options? We’ve been running that game—it has about 300 participants at this point—for two or three years.
These games do take a lot of time, and one of the things we can do at Hoover that you can’t do at a place like the Department of Defense is stare at a problem for a long time. Most Department of Defense games are one-off iterations, or they do one every year, but we’ll run games over and over and over again with different samples. That way, we understand whether these are the results we’re getting today or whether there’s something more pervasive that helps us generalize beyond any one game iteration.
In social science, we call that generalizability. Our games are trying to find generalizable patterns of human behavior that help understand how emerging technology might affect strategic stability.
Movroydis: What were some of the preliminary findings of the artificial-intelligence wargame involving China?
Schneider: Well, in general, people don’t trust the AI. So, there’s a lot of hedging that comes with the adoption of the technology. And we played this game not just with humans but also with LLMs [large-language models], and even the LLMs also primarily choose human control, or at least some level of human control. Nobody wants just a “dumb” human with no machine augmentation, but they do prefer some element of human oversight in the automated processes, even if that means you are less efficient or effective. It’s just a trust element.
Movroydis: You’ve talked about how wargames have been brought to people in Congress and the business community and other groups. Do these groups react differently to a certain scenario?
Schneider: Yes, it’s interesting. Maybe all Americans play the same first strategy. It doesn’t matter if you’re an undergrad, a mom who works at the bakery, or Henry Kissinger—maybe not Henry Kissinger, that might be an outlier, but a senior practitioner—but you’re hedging in the first move. You’re going to do intel, covert operations, you might do a little cyber thing here or there, and then hedge for time. I don’t know why, but they have to do it. I call it the throat-clearing move. Maybe it’s our national gift of having these oceans and borders with relatively friendly countries. Some of my foreign audiences, on the other hand, are more aggressive in that first move.
In games we ran with NATO players, they’re generally sophisticated about their understanding of nuclear stability and the effectiveness and risks of counterforce operations. But when we ran the same game in South America, a nonnuclear continent, we found that military officers were more likely to use counterforce and early, aggressive counternuclear operations. That might be because they haven’t had fifty years of people talking about nuclear stability the way NATO officers have.
In the AI project, we’re looking into whether demographics and expertise play a role. Do you work in tech? Do you work in the military? Are you a student? We’re studying whether those types of expertise variables influence the way you play the game. In general, I find that the more technical or tactical a game is, the more expertise matters. But the more “big strategy” is involved, the less expertise matters.
Which is a bit sad, right? It’s an affront to all of us who think we’re fantastic national security decision makers because of our years of experience. But a lot of our undergrads are playing the game very similarly.
