CNBC International employed quantitative game theory to generate a digital feature that forecast the future of the “Quad”—the strategic security grouping formed by the United States, Japan, India and Australia—and what it means for the economic and political future of the Indo-Pacific. To our knowledge, the Quad Project marks the most ambitious, data-intensive use of game theory by any major news organization in history. CNBC worked on the Project for seven months, publishing two days before Joe Biden hosted the first heads-of-state Quad meeting at the White House with Prime Ministers Narendra Modi, Yoshihide Suga and Scott Morrison.
The future of the Quad has tremendous and growing significance as China has become an increasingly assertive global power.
The Quad digital feature, entitled “The ‘Quad’ is on the rise in Asia-Pacific: Game theory has a prediction about its future,” was formally distributed within the U.S. State Department and read widely in the capitals of Japan, India, Australia and other countries in the region.
The project’s findings sparked intense interest at institutions that helped to “populate” the game theory model, including the Hoover Institution, Observer Research Foundation America, Eurasia Group, the Council on Foreign Relations, Brookings Institution, Asia Society, CSIS, Stanford University, New York University, the University of Pennsylvania, Princeton University and others. Especially noteworthy is that the piece was circulated and referenced among academics within China.
The digital-first project prompted several live international TV interviews on CNBC with its author, Ted Kemp, and the creation of a feature TV documentary about the project that began airing the week of 3 January 2022 throughout Asia, Europe, Africa, Latin America and Australia. Jonathan Grady, the applied game theorist who worked closest with Kemp on the project, discussed findings on live radio in Seoul, South Korea.
Policymakers, risk managers, investors, CEOs, and regular citizens in Asia, the United States and beyond are increasingly aware of rising stakes in the Indo-Pacific region. CNBC International’s Quad feature has attracted 150,000 unique readers online, with an impressive average engagement time of almost four minutes.
Even before project was published, it generated great excitement within policy circles. Among the 37 individuals who provided data to construct CNBC’s Quad game theory model were two former Australian foreign ministers, a former U.S. Defense official, former officials from France and Taiwan, and senior policy experts at several think tanks and top universities.
The Quad Project was based on a proprietary game theory architecture designed by Hoover Institution’s Bruce Bueno de Mesquita and his protégé Jonathan Grady of New York University. The Bueno de Mesquita model is well known within the highly specialized world of game theory.
In short, game theory uses computing power and logic to predict what individuals will do when they’re competing against each other. It creates a model that forecasts the decisions and counter-decisions within a scenario or “game” between those people, who are called “players.” The Quad Project crunched inputs on almost 300 individual “players”—most of them policymakers at the highest levels of government—in 15 countries and territories. The model ran millions of individual calculations to generate its final forecast.
The game theory architecture designed by Bueno de Mesquita has been used internally by the CIA on more than 1,200 intelligence projects, and it was found by the agency to have an aggregate 90% accuracy rate, according to unclassified CIA documentation that was published Yale University Press. The model has also been used commercially by Fortune 500 companies, and it has been cited by the Wall Street Journal and New York Times Magazine.
CNBC International’s Quad Project model is the largest computation ever run by the Bueno de Mesquita model in its history—more complex than anything the CIA did. CNBC “populated” the Quad model with input data from 37 of the world’s top experts on policymaking in China, the United States, India, Japan, Australia and 10 other countries and territories.
What was the hardest part of this project?
The first challenge was constructing the Quad model itself—what’s called “populating” the data that fed the model. It took seven months to find experts who could provide the data, to interview them in depth, to cross-check data points from experts against one another, and to format the data for the model. Running the model itself was by far the easiest part.
Second, most people are not familiar with game theory. Therefore, the feature article had to educate readers on game theory, including its shortcomings and criticisms, in addition to explaining the results pertaining to the Quad itself. This made writing a serious challenge, because it was necessary to balance introductory explanations of game theory and the Quad as a political grouping with a brisk, interesting narrative that delivered the actual predictions.
Third, almost as soon as we ran the model, some of its predictions began to actually take place in the real world. The final phases of publication were a race to incorporate those developments into the feature itself quickly so that events wouldn’t outrun the predictions we generated. More on that below.
This project was a first, not just for CNBC International, but on this scale a definite first for the Bueno de Mesquita model architecture, and we believe a first for any media organization anywhere. We strongly believe this project broke new ground for game theory and for an underexplored, data-driven journalistic practice.
What can others learn from this project?
Other journalists should learn from the Quad Project that quantitative methods, applied with rigor, can be used to render accurate predictions of the future, not just to analyze or assess events that have already happened. At CNBC, we came to see game theory as a new adjunct to the qualitative predictions—from analysts, strategists and others—for which we are well known on TV and online.
Specific predictions made by the Quad Project began happening almost immediately after we ran the model.
The model said other, non-Quad countries would align with the Quad or come close to its position on security, specifically naming the United Kingdom among others. After CNBC ran the model and generated that prediction—but just days prior to the Quad Project’s publication online—the United Kingdom made a surprise announcement that it would join a new security partnership with the United States and Australia that will, among other things, equip Australia with nuclear-powered submarines. The UK is not the only non-Quad country that is stepping up maritime security work in the region since then.
The model said that leaders in Australia, India, Japan and the United States would become more heavily focused on Indo-Pacific security, and the countries would act in an increasingly coordinated way. The joint statement by the Quad heads of state following their meeting at the White House testified to this new coordination, as have other events.
The model also singled out Vice President Kamala Harris as taking an increasingly sharp interest in the Quad, and it did so a week before she delivered pointed remarks about Indo-Pacific security while in Singapore and Vietnam.
There are early, tantalizing indications that the model’s most controversial prediction—that factional politics would develop within China in response to the Quad—may be taking shape as well.