When a group of about 40 players first tested out a live game called the Machine Learning President at a private event in San Francisco this February, they were unaware that the game would end up memorialized in the pages of The New Yorker.
But during a ski vacation in March, the Republican mega-donor Rebekah Mercer gathered her friends together to play several rounds of the game, which pits special interest groups, political candidates, and activist organizations against each other in a simulated presidential election, aided by cash and artificial intelligence. A lawyer for Mercer told The New Yorker that she owned a copy of the Machine Learning President but had not created it and that it did not reflect her family’s views.
It’s not hard to draw comparisons between the rules of the game, with its reliance on big cash and tech capabilities, and the actions of the Mercer-backed Cambridge Analytica during the 2016 U.S. presidential election. But, as Mercer’s lawyer stated, she had nothing to do with creating the game—in fact, it was conceptualized by one of her vocal critics.
Brett Horvath and Berit Anderson are the co-founders of Scout AI and the creators of the Machine Learning President. In 2017, the pair published a scathing critique of Cambridge Analytica, the now-shuttered political consultancy that misused the data of tens of millions of Facebook users and sat at the center of the social network’s largest scandal in years. “By leveraging automated emotional manipulation alongside swarms of bots, Facebook dark posts, A/B testing, and fake news networks, a company called Cambridge Analytica has activated an invisible machine that preys on the personalities of individual voters to create large shifts in public opinion,” the duo wrote.
That invisible machine—and the lack of preparedness for it in the 2016 election—provided inspiration for the Machine Learning President. The goal of the game is to get players thinking about ways tech and money could be manipulated to influence the 2020 election. (It also inspired Scout AI to spin out another group, Guardians AI, that’s focused on protecting pro-democracy groups from information warfare and cyber attacks.)
“This is an experience we created to help pro-democracy groups and strengthen democracy against some of the ways technology might interfere with fair elections,” explained Randy Lubin, one of the game’s designers and the leader of a design studio called Diegetic Games. “We knew that some sort of game or simulation or exercise was a really great way to understand the incentives and systems at play.”
On the only night the Machine Learning President had ever been played—at least, as far as its creators knew until The New Yorker revealed today that it had become part of a Mercer ski trip—the players included tech investors and leaders, as well as policymakers. The group of about 40 played it at an invite-only space in San Francisco’s ritzy Marina District, and they were instructed not to talk about the game on social media.
Players divided into “factions” during gameplay. There were two Republican, two Democratic candidates, and a bevy of special interest groups. Players representing Wall Street, Russia, and even Robert Mercer himself worked to convince the ‘candidates’ to publicly endorse policy positions on topics like America’s economic safety net, tech regulation, and privacy. In exchange for their endorsements, candidates could earn cash, technology, or direct influence over voting blocs. Players could also purchase technology to help push their messages to the public.
After heated primaries between Mike Pence and Condoleezza Rice on the Republican side, and Elizabeth Warren and Kamala Harris on the Democratic side, Harris won the mock election.
“People were really engaged. You can always tell how these things will go by how loud the room gets after you explain the rules,” Horvath said.
Horvath and Lubin hoped that the game would help bridge knowledge gaps between Washington, D.C., and Silicon Valley, and help leaders become better prepared for the next election.
“These problems have not really been solved, they’ve just gotten worse,” Horvath said of the botnets and shady ad buys that pervaded the 2016 election cycle. “Let’s get policymakers in the room so we can protect democracy, the integrity of elections, and the integrity of civil discourse.”
Horvath and Anderson have designed and created several other scenario-planning games. One, called Climate Casino, asks players to organize themselves into hedge funds and place bets on climate volatility. Another Scout AI brainchild lets players step into the role of a local mayor and decide on policies for self-driving cars. Lubin, in partnership with Techdirt editor Mike Masnick, is currently re-creating a game used by the Central Intelligence Agency to train new recruits, based on information about the game obtained in a freedom of information request. Techdirt also co-sponsored the one-night-only Machine Learning President event.
In order to determine the outcome of the game, Lubin based scoring on real-world voter models—and although the Mercer family apparently ended with a copy of the rules to the Machine Learning President, they likely don’t have all of the detail they would need to properly play the game. Lubin didn’t learn that the game had fallen into Mercer’s hands until a friend sent him the New Yorker story this morning, joking that he should launch a Kickstarter to create it. As he read the article, he recognized sentences from his own game design.
“The fact that they went to the trouble of getting the rules and running the game—it’s a bizarre and ironic form of validation, I suppose,” Horvath said. “I am fascinated to know what they learned and how that affected their plans in upcoming elections.”
And although he’s shocked that the game made its way into the hands of the Mercer family so quickly, he hopes to let more pro-democracy groups play it in the future.
“If anything, democracy is a giant act of scenario planning,” he added. “We might as well get good at it.”