The world lives in fear of an all-out nuclear war, but we don’t know much about how one might go down. The only wartime use of nuclear weapons happened more than 70 years ago, and technology has changed dramatically since then. What can we do to learn about, and prepare for, how world leaders might react to a nuke in a world with cyberwar, artificial intelligence, and advanced surveillance?
What if we played a video game to find out?
This approach is unconventional, but not unprecedented. During the Cold War, military and political leaders played hundreds of games to predict how a nuclear war might play out. With that in mind, the Project on Nuclear Gaming (PoNG)—helmed by a team of researchers from the University of California at Berkeley, Lawrence Livermore National Laboratory, and Sandia National Laboratory—created an online multiplayer strategy game called SIGNAL, or Strategic Interaction Game between Nuclear Armed Lands.
SIGNAL is part video game, part experiment, and part data collection tool. The hope is that, by observing people playing the game and collecting the data it generates, PoNG can learn about human decision-making during nuclear conflicts. They’d love for you to play a few rounds and see how you fare.
SIGNAL is fascinating, but flawed. It’s attempt to study something terrible—nuclear war—in a controlled environment. I didn’t have a good time playing the game, but PoNG didn’t design it as a commercial product. But it also felt like the in-game consequences of nuclear war were low, and the games so short that I had to wonder: Is a game really a viable medium for learning about something as important and complicated as nuclear war?
“The promise of experimental wargaming is that it offers an additional tool in the toolkit for scholars to consider questions that can’t be answered using ‘real-world’ data,” the researchers of PoNG told me in an email. “The data collected may serve to generate theory for further study.”
SIGNAL began life as a board game, but the PoNG team concluded its use was limited, though they still use the analog version to gather data. One of the problems with the board game was that the participant pool was limited. PoNG offered the game up at conferences and got a certain kind of player, what it calls its “elite” pool—government functionaries, politicians, military leadership, and defense journalists.
This kind of narrow participant group is a historical problem with war games. RAND—a Pentagon-backed think tank—ran the bulk of the Cold War-era wargames, and felt that data generated by wargames only mattered if elites played the games.
Putting people from all walks of life in charge of nukes was important to PoNG. They wanted as broad a data pool as possible, and so they decided to make SIGNAL an online multiplayer game that anyone can join from home.
“ SIGNAL offers the opportunity to examine whether and how players from different backgrounds and with different political-military experiences might play the game differently,” PoNG told me. “We collect demographic data on players, including their level of knowledge on nuclear issues, and our pool includes a number of players with extensive knowledge of nuclear weapons, weapon effects, military strategy, and diplomacy.”
Can a game really mirror nuclear war?
SIGNAL plays like a browser-based, multiplayer game of Civilization distilled down to a few quick rounds. It takes place on a world map made of hexes, everyone can see what everyone else is doing. Players take control of one of three fictional and nameless countries abutting each other. The goal is to score points by expanding your country’s infrastructure, gathering resources, and defending yourself from assault.
The game introduces some wrinkles to keep things interesting. Occasionally, for example, one player may spawn without nuclear abilities in order to act as a potential neutral ally. There’s also a chat window so everyone can talk openly with each other, or negotiate in private messages.
Each round begins with five minutes of “signalling.” The “signalling” phase of the game is the most crucial, the player has markers they can put on the board to show that they’re interested in a hex. They don’t have to explain what they’re going to do on that hex—they may build a farm, they may deploy troops, or even launch a nuke—just mark that they might to do something there; or, a player may do nothing at all.
“Signalling” takes five minutes and happens in real time. During this time, players can negotiate with each other; money, resources, and territory can change hands. You could, for example, drop a signal flag on an opponent's city, and then claim in the chat that you’re going to nuke that space unless the opponent pays a ransom. After five minutes, players take turns executing actions on the hexes that they’ve marked. Each of these action phases only lasts 45 seconds and there’s pressure to play well or lose big.
Purely in terms of game mechanics, the built-in consequences of using nuclear weapons are low. According to PoNG, that’s part of the point of the game. “Players control a good deal of the costs of military action, including nuclear use,” they said. In effect, players create the costs of nuclear actions in the game.
“We have...seen players make all sorts of agreements concerning nuclear use from ‘no first use’ policies to, in [the board game], disarmament as players discard their nuclear capabilities altogether," the PoNG team said.
PoNG pointed out that, in the real world, nuclear taboos and treaties are only as good as long as they’re agreed upon by all parties. For decades, Russia and the United States had a treaty against intermediate-range ballistic missiles. As of August 2, that treaty is dead. Another Obama-era treaty may be next on the chopping block. “After a lot of debate, the team agreed that for this version of the game, an attempt to manufacture either deterrence or normative costs would have been artificial and biased the experiment,” PoNG said.
Wargaming has a long tradition among the American military. Large portions of America’s Cold War nuclear strategy and Vietnam plans were gamed out in huge simulations. Outside of a military context, epidemiologists once used a World of Warcraft bug to study the spread of disease. But there are limits to what you can learn from a game.
The PoNG was quick to point out the World of Warcraft study worked so well because the epidemiologists had plenty of real-world examples to compare it to. “For questions concerning nuclear weapons use, researchers face the fortunate but difficult problem of having little by way of empirical data to draw on,” PoNG said.
SIGNAL is an admittedly imperfect attempt to fill in the data gaps of nuclear war research. But there are other, more practical, problems with the game. For starters, SIGNAL is remarkably complicated for a game that takes roughly 20 minutes to play from start to finish. The user interface is clunky, non-reactive, and buggy. It launches in Firefox just fine, but often crashes in Chrome. And that’s before I even tried to get into a game.
Yet another practical problem is that, as an academic research tool, not many people are playing SIGNAL. The few times I did manage to get a game together with random folks, players would often drop out halfway through the game. The only full games I experienced happened when I convinced friends to log on. “It is important to bear in mind that this is an academic study and not a commercial game,” the PoNG said.
Despite the low player count from the user side, PoNG said that more than 400 games of SIGNAL have been played. “At the time of writing (early August), we have conducted over 400 games and the data pool continues to grow,” they said. “As far as we’re aware, our team—with just the games played thus far—has the largest dataset of wargames designed for academic inquiry by a considerable margin.”
In the real world, we’ve gotten close to nuclear war but haven’t seen a wartime detonation since 1945. The consequences were so extreme that we’ve restrained ourselves since then. Those consequences aren’t built into SIGNAL, which bugged me despite the PoNG team's intention for players to create their own consequences. This lack of consequence—either on an emotional level driven by a narrative or a mechanical level driven by in-game consequences—points to one of the fundamental issues with using games (whether of the analog or digital varieties) as a research tool.
“Wargaming has a long history in trying to help people figure out what could happen. The tricky part is, of course, that a game isn't reality—not in the slightest,” Alex Wellerstein, a professor at Stevens Institute of Technology and creator of Nukeamp (a tool that lets users see the effects of nuclear detonations), said. “Even in situations where the game's choice of dynamics to model are ones that correspond with reality (which is arguably rarely the case), the fact that the player knows it is a game means that you aren't going to get realistic outcomes.”
According to Wellerstein, the fact that the people playing war games for research wouldn't have a relevant stake in real-world decisions can limit their explanatory potential. On the other hand, institutional review boards are likely to view simulations that make an average person believe that they're actually launching a nuke would be unethical.
PoNG isn’t blind to this problem. “It’s worth noting that this is a challenge for all wargames or models of conflict that include human decision making—the costs in the model are typically nowhere near what they would be in an actual conflict,” The alternatives aren't much better, however. "Do survey experiments, for example, do a better job of engaging an [military commanders] than an experimental wargame? Do formal models?” the team asked.
Wellerstein still sees opportunities for games to teach and enlighten. ““Video games are another type of media, one whose active-learning potential is probably higher than the others,” he said “Certainly I've seen people talk about the impact of the Fallout series on their understanding nuclear outcomes.”
Wellerstein even runs his own game-like simulations when teaching nuclear issues. Wellerstein simulates the negotiations around the 1968 Nuclear Non-Proliferation treaty. The students form a kind of model United Nations, each representing real country, and they can make their own choices including spying.
PoNG believes that SIGNAL has value, despite its issues. No game is perfect and no game designed to collect data about nuclear war can ever replicate the real thing. “But we cannot simply say that the problem is too hard to study," the PoNG team said. "Moreover, it is too important not to study. So we’re moving forward with our work while trying to be ever mindful of its many limitations.”