Can war games really help us predict who will win a conflict?

The writer is director of the Hoover Institution’s war gaming and simulations initiative and a Hoover fellow at Stanford University

War games, once niche and highly secretive, are coming in from the cold. Two recent unclassified games run by US think-tanks found that a future conflict over Taiwan would lead to a bloody stalemate, predicting “a huge cost” for all countries involved. There have also been reports of classified Nato-Ukraine war games and declassified Air Force war games that find (unsurprisingly) that in order to defeat China, military personnel need new fighter jets and bombers.

These games have received a disproportionate amount of attention — but what do they really mean and why do they matter? While often called “simulations” or “exercises”, war games are distinct from computer imitations of combat, field exercises or organised brainstorming sessions. Instead, they are interactive events that display four characteristics: expert players, immersed in scenarios, bounded by rules and motivated by consequence-based outcomes.

War games go back millennia to ancient Rome, early Iraq, and China. They took on a central role in the modern conduct of war with the Prussian development of Kriegsspiel, a board game that simulated combat to train officers in the early 19th century. During the cold war, the US and Nato turned to war games to understand the impact of the nuclear revolution. Defence war gaming continued after the Berlin Wall fell, with games designed to test new ideas about “information age” warfare.

Due to their close resemblance to real conflict experience, games offer compelling evidence for policymakers facing difficult choices. Indeed, they can be such powerful devices of influence that organisations have been known to change rules to sway outcomes and leak the results when it benefits their cause.

A game designed to answer a question — who would win a war over Taiwan? How would American support for Ukraine affect the war? — needs five qualities. It needs to be believable, it needs the right players with the correct expertise and demographics, and there must be enough players and game iterations to come to realistic conclusions. The best war games control for bias within their scenarios and rules. Good data collection is vital.

Satisfying all these qualities is hard. War games are not crystal balls that tell us what will happen in conflict or crisis, but instead reveal the possible outcomes. US Admiral Chester W. Nimitz famously concluded that the interwar games run at the Naval War College had been run “by so many people and in so many different ways, that nothing that happened during the war was a surprise”.

The lesson is that we shouldn’t draw too much certainty from any one game and should instead look for insights across multiple games. The US think-tanks’ games suggested there would be no clear winner in a Taiwan fight. That fits with other analysis.

However, variables such as individual leadership styles, weapons capabilities or campaign choices can lead to different results. This is why we must evaluate not just the outcomes of a game, but its design: the rules, assumptions, scenarios and players. These Taiwan games, for instance, used only unclassified information about weapons capabilities and did not include Chinese players.

We should also be aware of an organisation’s incentives for running a war game and publicising the results. The think-tank games were designed by respected scholars who revealed their rules and assumptions, suggesting they are less biased than those which are “sponsored” by interested parties such as defence contractors or even organisations within the US Defense Department.

Leaked classified games should always be questioned. Does the organisation running things have a bias towards certain outcomes? Understanding the motivation behind the leaks is perhaps more illuminating than the games themselves.

Finally, remember that even the best games are not predictions of the future. A recent series I ran over three years with 580 players found that cyber threats to nuclear weapons arsenals did not lead to nuclear escalation. However, it revealed that the games which did go nuclear always occurred when I gave players cyber weapons that could target nuclear command and control. These weapons were all the more dangerous in the hands of those with limited experience in nuclear strategy. The power of this game series, therefore, was what it revealed about when and why things went awry.

This is the value of war games. They can help us understand human behaviours in unpredictable and rare scenarios but not necessarily to predict the future.