Chaos is a Ladder: Media Literacy in an Age of Unlimited Fake News

You are a rising star inside Chaos Corp, an underground disinformation operation. Your bosses have handed you the biggest job of your young career: get Teodoro “Teddy” Bautista elected president of the Philippines. Teddy is a former child game-show sensation and karaoke cosplay champion, the man who gave the nation Teddy Power Energy Drinks and Love Finder Pro Max. His promises are impossible. His claims are absurd. His hair is clearly dyed. You have seven days to convince voters he is a man of the people. Teddy’s principal interest, however, is Teddy.

That is the premise of Chaos Corp: Troll Farm Simulator, a game I developed with Katrina Alvarez, a media and politics researcher at De La Salle University in Manila, and our production team. Littlefinger explains the strategy in Game of Thrones: chaos is a ladder. You run the troll farm and manufacture chaos. Teddy climbs.

As a production professor who builds games and interactive media, I wanted players to understand disinformation from the operator’s side of the desk. You encounter a news event, choose an attack, and watch how it serves your client. Your goal is to get Teddy elected. Nobody in the game sets out to damage public trust, but every attack does, and the player experiences that damage firsthand. That is the lesson.

Generative artificial intelligence (AI) makes convincing false content cheaper and easier to produce, so an operation that once needed a building full of trolls now needs a laptop and a subscription. The game exists to fight the resulting infodemic: a pandemic of nonsense, accelerated by AI slop, amplified by the engagement algorithms that run social media, and weaponized by bad actors. The messages get more polished and more plausible every month, and they arrive in unlimited quantity. The tactics behind them do not change. Learning to recognize the tactics still works.

Why We Let Players Be the Bad Guy

During development we drew up the opposite design: a crusader for truth who chases lies and defends the public. We wrote the design documents and looked at it both ways. Playing the paladin, the white knight, made for the more boring game. Running the troll farm gave players richer strategic choices, and it was edgier and more transgressive, which matters when you want young people with high game literacy to keep playing.

We trust the gamer. James Paul Gee has described how players hold a real-world identity and a virtual identity at the same time, and try out new ones because the game carries no real-world consequences. Anyone who has played as the Empire in a Star Wars game knows the difference between the character and the person holding the controller. We trust that players can inhabit the role for seven days and still understand the difference.

I also wanted to avoid chocolate-covered broccoli: textbook instruction with a few weak game elements sprinkled on top. Students who play games can see right through that. In Chaos Corp you have to think through how an attack works to decide whether it will help you win, and that reflection is where the learning occurs.

Blame the Intern

Under the hood, Chaos Corp is a card game, and every card is a troll: bot operator, influencer, provocateur, celebrity, troubleshooter, crisis manager, each with its own strengths, costs, and cooldowns. Plan your pandemonium accordingly, the tutorial advises. News events pop up on a map of the Philippines, and you drag a troll onto an event to spin it in Teddy’s favor. The first time you use a tactic, it enters your glossary. There are twenty: ad hominem, astroturfing, whataboutism, gaslighting, sealioning, clickbait, context twisting, fear-mongering, imposter infiltration, conspiracy theories, deepfakes, and more, spread across ten storylines.

Take one. House Speaker Andres Moreno, Teddy’s golf buddy and brother-in-law, gets caught pocketing millions in pork-barrel kickbacks. Inconvenient. You deploy a bot operator, who pins the whole thing on an intern named Ricky “the Ringer” Vasquez. From intern to infamy, reads the headline. The scandal of a powerful man becomes the scandal of a nobody, and the powerful man walks. Poor Ricky. You have just run an ad hominem attack and watched it work.

In another event, Teddy forgets his own education policy in a live debate. You answer with astroturfing, fake grassroots support, and within hours posts from invented educators praise Teddy for thinking on his feet and valuing input over rigid plans. His failure to prepare becomes proof of his humility. The educators do not exist. Astroturfing is one of the most compelling tactics in the game because it hits close to home, and because generative AI is so good at it. A thousand supportive educators, each with a name, a face, and a writing style, now cost nothing to invent.

When a former employee exposes Chaos Corp’s illegal surveillance of political figures, your troubleshooter answers with fear-mongering, warning that releasing such information endangers national security and urging the public to focus on external threats instead. Investigation closed. The public now worries about something else.

And when celebrities organize Unmask the Trolls concerts against Chaos Corp, your troubleshooter unleashes a whataboutism whirlwind, dredging up the celebrities’ old controversies to question their moral authority. What about this, what about that. Nobody is talking about the trolls anymore.

When Winning Means Everyone Loses

Every attack serves the client and pollutes the information environment on which everyone else depends. Fabricated endorsements make real support indistinguishable from paid campaigns. Misleading frames hide what a candidate did. Attacks on critics bury the evidence. The player watches this happen, event by event, and the game keeps score.

Take enough risk and an investigation spawns. The music turns paranoid. Investigations block events and drain chaos every day until you close them with defensive trolls, which cost chaos of their own. Gabriela Santos, a relentless investigative reporter, files a report at the end of every day. She knows you are out there, and she is closing in on you. She is the game’s moral compass, and within the game logic she serves a specific purpose: she shows the player, day after day, how decent people respond to what Chaos Corp is doing.

Then there is the Chaos Meter. Reach fifty-one percent by day seven and Teddy wins. Push too far and you unlock the secret third ending, Collapse. The information sphere gives out. Nobody believes anything, including the lies, because a lie only works while some of the news is still credible. Information bankruptcy ruins the public sphere for everybody, the client and the trolls included.

Our hypothesis, in terms of learning design, is that if you have run an attack, you will recognize it in your feed. Someone who has manufactured grassroots support has a reason to ask whether a wave of praise comes from real people. Someone who has pinned a scandal on an intern can spot the diversion and return to the original allegation. This is situated learning, doing as a way of learning. A player who has used these tactics in context, inside a simulation, will recognize them in context, out in the world.

Generative AI will keep making the material more polished, more plausible, and more plentiful, and no fact-checking operation will keep pace with a machine that produces nonsense at scale. The tactics will not change. Recognizing them is a skill, and skills take practice. Chaos Corp turns that practice into a game.

Chaos Corp is free on iOS and Android. More information is available on the game’s project page.