Modeling the complex and diverse threats that are targeting the world’s most important election
For all the talk about the revenge of the bipolar or multipolar world, the honest truth is that America’s presidential election is the definitive event in international relations. Every four years, global leaders wait with bated breath to see how the world’s sole remaining superpower tacks its sails.
That decisiveness also means that more and more of those leaders are no longer passively holding their breath for Election Day results, but are instead using the full spectrum of tools available to them to actively influence the election’s outcome. That can mean propping up one side or the other; fomenting chaos, nihilism or cynicism; or just extracting cash from the ever-burgeoning economic scope of U.S. elections, which this year has breached $14 billion across 11,000 political groups — and that’s just what’s disclosed.
Worse, the complexity of threats to U.S. elections continues to deepen. Artificial intelligence and big data combine to offer influence operations a level of personalization that was non-existent in previous cycles. The lives of Americans have rapidly digitalized since Covid-19 and the 2020 cycle, offering more surface area for hackers, cybercriminals and others to endanger the vote. Then there is the inexorable rise of the influencer industry and TikTok concomitant with the decline of traditional news sources.
Appreciating this complexity and expansiveness is hard with policy memos — which is where Riskgaming comes in. With just days before Election Day, I’m excited to publish our second scenario, DeepFaked and DeepSixed: AI Election Security and the Future of Democracy, which incorporates the diverse threats that confront U.S. security officials with the dizzying diversity of actors who can potentially offer clues to solve them.
DeepFaked and DeepSixed is a bit different from our previous political and economic simulations, which tend toward groups of 4-8 people negotiating, haggling and cajoling over the course of several hours. Instead, this game centers on an intelligence fusion center at the White House where 54 people come together to offer information and to seek out patterns of threats against American democracy. Player roles come from across government, international organizations, the private sector and non-profits, and they are all designed to offer both a crisp backstory as well as essential clues relevant to that character’s background.
During each of two scenes — one representing the events before Election Day and the other during and after the ballot — players walk around the fusion center trying to connect their clues to those from other players. When they identify a potential threat, they submit the details to the “center director” for processing. After a debriefing at the end of the scene, points are offered for correctly identified threats and lost for threats that didn’t pan out. It’s a cooperative game, with everyone racing against time and the chaos to seek signal in the noise.
We played the game earlier this year with our friends at TheFuture.us, a “policy studio that forecasts the coming tech disruptions to future-proof America,” which inspired the game’s design. That event included secretaries of state, Senate candidates, policymakers from across local, state and federal government as well as a number of tech CEOs and venture capitalists.
This week, we hosted two parallel runthroughs in New York and Washington with a collective 100 players spanning politics, tech and finance. Our special guest was Senator Mark Warner of Virginia, who spoke about his efforts to secure American elections, and then proceeded to role-play his character, seeking out help on a threat the country faced before the election (he succeeded).
Compared to our other riskgaming scenarios, where it can take a good 15-20 minutes for some players to fully inhabit their characters and begin making strategic decisions, this game’s simpler design helped everyone dive right in. There were agency turf wars, arguments over evidence, and even a potential threat identified all within the first minute of the experience kicking off.
However, when all the points were tabulated for both scenes, New York scored 10.5 and our Washington crew scored 5.5. Why the gap? A few observations, which you can hear more about on the Riskgaming podcast this week between me and our director of programming Laurence Pevsner.
🔊 Listen to “Introducing our new scenario, DeepFaked and DeepSixed”
First, we noticed in New York with the tech- and finance-centric crowd that there were more of what Laurence dubbed “super-connectors” — people who tried to connect together others’ clues and not just their own. While playing, they would pick up on a pattern and would actively work to bring those people together to trade notes. This helped to build signal quickly under the tight conditions of the game. There were less of these super-connectors in DC, where players tended to be more goal-driven to seek information directly relevant to their own character.
Second, we observed that in Washington, some people dismissed their information out of hand and didn’t share it with others. They had read their clue and decided that it wasn’t likely to be relevant, and so they had decided to withhold it to make it easier for everyone else to find information that might be more impactful. The downside to that analytical filtering though is that crucial facts about some threats to the U.S. elections were entirely lost. In New York, we saw less of this filtering and more of a perspective that any information might be useful to someone somewhere to connect a stray dot. That helped players discover threats serendipitously.
Third and less surprisingly, both groups performed better in their second scenes than in their first ones, as they got used to the game and the group. It’s an important pattern we see in the real world: groups that have previously worked together are going to perform better than people who have never met each other. It’s critical to build relationships and a working culture before the security threats begin.
That’s the game — it was a blast. And now all the materials needed to play it are available on the Lux Riskgaming website.
Are American elections safe? Yes, and no. As Senator Warner was describing in his remarks, federal agencies are working tirelessly to protect the vote from foreign interference and emerging technologies. For instance, just yesterday the FBI, ODNI, and CISA determined that several deepfake videos circulating online were planted by Russia to foment disenchantment with the democratic process. These agencies are doing the best they can under the circumstances.
What’s missing is that the private sector is weakly engaged in these national security conversations. It’s hard for dozens of companies with relevant information to coordinate with local, state and federal election officials plus the intelligence community plus international counterparts. There is just too much surface area.
Second and more challengingly, elections are held infrequently. Unlike, say, financial transactions which are rigorously and continuously monitored by well-staffed security operations centers — no such infrastructure exists for U.S. elections, which are organized by thousands of different local counties. Every four years, these election agencies have to rebuild their security playbooks for new technologies and new threats with help from federal institutions like the Cybersecurity and Infrastructure Security Agency. Elections need surge capacity to protect them.
Third, while nearly all counties lack sufficient resources to protect elections, the reality is that a small percentage of counties in critical swing states will determine the outcome. For those who wish to threaten American elections, these are ripe targets. Not only should we have a reserve force of cybersecurity and other professionals prepared for each election cycle, but those resources should be more heavily deployed to the greatest hotspots on the map.
Those are just some of our findings with this new Riskgaming scenario. My hope is that as more people play this game and internalize the complexity of election security and emerging technology, they will come out energized to confront these challenges to American democracy. Security is always an arms race even for a superpower — but that doesn’t mean we’re DeepSixed.
Links
- DeepFaked and DeepSixed Homepage
- Game Scenario and Rulebook
- Lux Riskgaming Homepage
- This week’s podcast discussing the game
Sign up for Riskgaming!
Excited to join a live Riskgaming event with us here at Lux? We have a proper signup form now, and we’d love to have you come to our next installment.
✍️ Sign up for future Riskgaming events now
The Orthogonal Bet: Exploring the history of intelligence
In this episode, Lux’s scientist-in-residence Sam Arbesman speaks with writer, researcher, and entrepreneur Max Bennett. Max is the co-founder of multiple AI companies and the author of the fascinating book, A Brief History of Intelligence: Evolution, AI, and the Five Breakthroughs That Made Our Brains. This book offers a deeply researched look at the nature of intelligence and how biological history has led to this phenomenon. It explores aspects of evolution, the similarities and differences between AI and human intelligence, many features of neuroscience, and more.
🔊 Listen to “Exploring the history of intelligence”
Introducing our newest initiative, Lux Labs
Before we close out, we have an audacious vision at Lux to fund the most pathbreaking companies in the world. Sitting parallel to our traditional venture funding is a new initiative we launched this week called Lux Labs. As we described in the announcement:
Lux Labs formalizes a pattern that we have used repeatedly and to great success across Lux’s funds, where we partner with ambitious people with brilliant and proven ideas and offer them an entrepreneurial pathway to transition from the intellectual ferment of the research lab into the competitive free-for-all of the marketplace.
Our own Josh Wolfe discussed Lux Labs with Allie Garfinkle at Fortune Term Sheet:
Venture is in a pullback, Wolfe says, making this the ideal time to pursue a differentiated approach. Lux Labs will offer hands-on support, like IP rights, company formation, board member recruitment, and building networks.
“Sometimes it starts with a thesis,” said Wolfe. “Sometimes it starts with a specific breakthrough. Sometimes it starts with a breakthrough scientist, an individual. We’ll find a team and actually convince them that they should spin out.”
Come join us! And check out our one-minute teaser on YouTube.
Lux Recommends
- One of our friends at Lux, Valentin Pitarque, sent us Jared Cohen’s essay in Foreign Policy on “The Next AI Debate Is About Geopolitics.” “America cannot achieve AI autarky, especially when it comes to data centers. AI software must run on AI hardware somewhere—the question is where. The United States needs to develop a list of partners with the capacity, will, and aligned interests for a secure global data center buildout.”
- In a stunning analysis, Dylan Patel and colleagues at Semianalysis track down “Fab Whack-A-Mole: Chinese Companies are Evading U.S. Sanctions.” “Sanctions violations are egregious. SMIC produces 7nm-class chips including the Kirin 9000S mobile SoC and Ascend 910B AI accelerator. Two of their fabs are connected via wafer bridge, such that an automated overhead track can move wafers between them. For production purposes, this forms a continuous cleanroom and effectively one fab. But for regulatory purposes, they are separate! One building is entity-listed by the U.S. and working on advanced logic for AI chips, a clear national security concern. The other is free to import ‘dual use’ tools as it runs only ‘legacy processes.’ Do you believe they aren’t sharing anything over the wafer bridge?”
- We had Fluke author Brian Klaas on the RIskgaming podcast earlier this year, and his new essay in Aeon is a great dive on “Without chaos theory, social science will never understand the world.” “By smoothing over near-infinite complexity, linear regressions make our nonlinear world appear to follow the comforting progression of a single ordered line. This is a conjuring trick. And to complete it successfully, scientists need to purge whatever doesn’t fit. They need to detect the ‘signal’ and delete the ‘noise’. But in chaotic systems, the noise matters. Do we really care that 99.8 per cent of the Titanic’s voyage went off without a hitch, or that Abraham Lincoln enjoyed most of the play before he was shot?”
- Sam recommends “Retro 80s Versions of Tech Company Logos” as well as “The ABC of Fonts,” a three-part series by Simon Garfield on “Albertus, Baskerville and Comic Sans.” “In a now-famous thought experiment, the type designer Cyrus Highsmith once tried to plan a day in New York without Helvetica. A tough assignment, for it was everywhere: dollar bills, breakfast cereals, morning newspapers, subway lettering.”
- Finally, wargaming (the far less interesting variant of Riskgaming if I dare say so myself) got a nice profile in The Wall Street Journal this past week in “A Million People Play This Video Wargame. So Does the Pentagon.” “Whether AI and advanced software actually improve wargaming and preparations for war is a question sparking battles of its own. Warfare is so complex—buffeted by factors ranging from equipment and strategy to politics, weather and corruption—that modeling all the inputs entails parsing an almost infinite number of variables. Quantifying unquantifiables such as military morale requires arbitrary decisions.”
That’s it, folks. Have questions, comments, or ideas? This newsletter is sent from my email, so you can just click reply.
