Looking Back at a Pandemic Simulator

Bouncy-circle COVID game reignites model fights — and the comments go feral

TLDR: A designer revisits his COVID math and pitches a simple simulator to teach how outbreaks spread. The comments explode into a model smackdown: some say simulations are a nerd trap and even major code was buggy; others argue imperfect models still beat guesswork—because policy depends on them.

A veteran designer revisits his early-pandemic math and a simple “bouncing circles” simulator meant to teach how infection spreads, why testing matters, and why social distancing helps. He cites grim numbers and says his math wasn’t far off: over a million U.S. deaths. He explains terms in plain English—IFR is the chance of dying if you catch it, and R is how many people each sick person infects—and argues that aggressive, united action could’ve saved more lives. Then the comments crash the party. One camp calls the whole thing a nerd trap, dunking on models that look smart but miss messy reality. Another camp fires back: models aren’t perfect, but they’re better than vibes. The spiciest post? A takedown of the UK’s Imperial College (ICL) code, alleged to have a race condition so bad timelines could be off by a week. Cue memes: “screensaver of doom,” “Agar.io but for anxiety,” and “name every dot from a baby book—Donna’s got diabetes!” The thread devolves into “trust the math vs trust your gut,” with eye-rolls at spreadsheets and dark humor about policy made from buggy code. The article lives here: link. The real show? The roast fest in the replies.

Key Points

  • The author began analyzing COVID-19 in late 2019, sharing posts on Facebook and explaining IFR and R values.
  • A sample calculation using a 0.4% IFR and 60% infection rate estimated ~780,000 U.S. deaths from a 328 million population.
  • The article states herd immunity was unrealistic; vaccines were deployed rapidly, and viral evolution lowered death rates.
  • Reported estimates cited are 1.2–1.4 million deaths in the U.S. and 15–28.5 million globally.
  • A simple agent-based pandemic simulator design is outlined, with agents moving between defined health states and a diagnosed flag affecting visibility.

Hottest takes

"My model was right up until… so it was pretty good." — groundzeros2015
"Having access to data can give one an illusion of understanding" — groundzeros2015
"They're all awful." — rob_c
Made with <3 by @siedrix and @shesho from CDMX. Powered by Forge&Hive.