Twitch Plays Pokémon began with a wonderfully bad idea: let an entire Twitch chat control one copy of Pokémon Red by typing button commands. The anonymous experiment launched in February 2014, and the obvious prediction was that thousands of competing inputs would make basic movement impossible.
For long stretches, that prediction looked correct. Viewers could send directions and button presses into chat, with the game trying to execute the crowd's instructions. A doorway could become a project. Menus became hazards. Pokémon were released by mistake. Yet the same disorder that made progress painful also gave the stream its personality.
PC Gamer's later history of Twitch Plays describes the first Pokémon Red run drawing more than 100,000 simultaneous viewers and about 1.16 million participants over roughly 255 hours. The point was no longer simply whether a Game Boy game could be finished. People were watching a crowd invent its own way of playing.
Anarchy created the story
The original control system became known as anarchy: commands were executed as they arrived. It was fast, democratic in the loosest possible sense, and spectacularly unreliable.
That unreliability produced the moments people remembered. Viewers gave accidental nicknames to Pokémon, mourned party members that were released, and built elaborate jokes and mythology around ordinary items and characters. A Helix Fossil became a symbol. A Flareon became a villain. None of that was written into Pokémon Red; it was a layer of culture built by the people trying to steer it.
The practical problem was that some sections demanded more precision than raw chat could provide. The experiment introduced a democracy system that let viewers vote on the next input, creating a slower but more controlled alternative. The stream then developed another argument on top of the game itself: whether order was worth the loss of anarchy's chaos.
That tension mattered because it made spectators part of the event even when they were not issuing useful commands. People could argue about strategy, spam an instruction, make art, repeat a joke or simply watch the crowd fail at walking around a corner.
Somehow, the crowd finished it
After more than 16 days, the players defeated the Elite Four and the final rival battle. PC Gamer's retrospective records about 1.16 million participants across the run.
The achievement sounds more coordinated in a sentence than it looked in motion. Progress came through enormous amounts of wasted input, reversals and accidental decisions. That was precisely why the finish worked as a communal payoff. The crowd had not found an efficient way to play Pokémon. It had found a way to make inefficiency entertaining enough that people kept returning until the job was done.
The experiment also proved that audience interaction could be more than a poll beside a livestream. Chat could become part of the game system itself.
The format outlived the original run
Twitch Plays Pokémon continued beyond Pokémon Red, and the basic idea escaped into other games and streams. PC Gamer later documented Twitch-controlled attempts at games including Dark Souls and PUBG, while developers and publishers have continued borrowing the format.
A particularly neat example arrived in 2024, when Capcom used a Twitch Plays setup to let viewers collectively control Akuma around the launch of the character in Street Fighter 6. PC Gamer explicitly traced that stunt back to the Pokémon experiment a decade earlier.
That is the useful legacy of Twitch Plays Pokémon. It was not important because a million people discovered the optimal way to beat a children's RPG. They did almost the opposite.
They turned control itself into the spectacle. The mistakes became lore, the arguments became mechanics, and a solitary Game Boy adventure became something a huge online crowd could claim as its own. Twelve years later, that still feels like one of livestreaming's stranger and better tricks.
Reporting notes