Red vs. blue button: TBPN audience skews red in the viral thought experiment that stumped the internet

Red Button Voters Dominate TBPN Chat in Viral Thought Experiment

The red-blue button dilemma went viral this month—Tim Urban's Wait But Why post hit 24 million views on April 24, followed four days later by MrBeast's 10 million view riff on the same question. The premise is simple: everyone votes privately. If more than 50% press blue, everyone survives. If fewer than 50% press blue, only red voters survive. Which do you choose?

The TBPN audience skewed decisively red, a sharp reversal from both the Urban and MrBeast polls, where roughly 58% and 55.7% chose blue respectively. That split matters because it exposes a real tension in how people reason about the problem.

The blue case rests on coordination. If you assume most people will act rationally and press blue, you need only 51% compliance to save everyone. The red case rests on dominance. Pressing red guarantees your survival no matter what anyone else does—if more than 50% press blue, red voters live anyway; if fewer press blue, red voters are the only survivors. The math is unambiguous: there is no scenario where blue produces a better outcome for you individually than red. That's why Claude, when asked directly, called red "the strictly dominant strategy."

Yet the original polls showed blue winning. Why the gap between TBPN and the mass audience? The hosts zeroed in on selection bias. Millions of people who voted in the Urban and MrBeast polls had never seen the thought experiment before. They answered quickly, on social media scrolling, with no real stakes. The TBPN audience, by contrast, consists of people who follow a tech podcast—they're more likely to have encountered the dilemma before, to have thought through the game theory, to have been persuaded by the rational argument for red.

The frame you use determines the answer. One creator reframed the buttons with honest labels: red says "nothing happens," blue says "the ultimate death gamble." Suddenly red looked obvious. Another inverted it: blue says "nothing happens," red says "the ultimate murder gamble" because pressing red guarantees blue voters die unless enough people defect. The emotional weight shifts depending on which button you're responsible for.

ChatGPT and Claude gave different answers when asked directly. ChatGPT pressed blue, framing it as "a cooperative choice that maximizes the chance everyone survives" and acknowledging moral weight in voting. Claude pressed red: "Red is individually tempting as insurance, but if everyone reasons that way, everyone except red voters dies." Claude's reasoning was pure game theory; ChatGPT's was moral hedging.

A separate poll of 14,000 nationally representative Americans showed blue winning by three to one. The trait most predictive of blue selection was honesty. People who agreed strongly with "I tell the truth" chose blue at higher rates.

The real world analogy is coordination under uncertainty. If people had time to educate, campaign, and persuade each other—as they would in a genuine global scenario—the outcome could flip. Mass ignorance of the dilemma before encountering it in a social media post is not the same as rational deliberation with full information.

The thought experiment works as a personality test because it exposes whether someone defaults to individual optimization or collective benefit, and whether they trust the world to cooperate or assume defection.