1-Slack

Stewart Butterfield wanted to build a video game. He wanted it badly enough to do it twice. In 2002 he co-founded Ludicorp to build Game Neverending, a browser-based multiplayer world with no combat and no winning — just a strange, sociable sandbox. The game never worked. But to build it, the team had constructed a small feature for sharing images inside the game, and when Ludicorp pulled that feature out and shipped it on its own, it became Flickr. Yahoo bought it in 2005. The game was quietly killed. Most founders would take the lesson. Butterfield took the money and went back to the game. In 2009 he started Tiny Speck to build Glitch — again a browser-based multiplayer world, again no combat, again beautiful and strange. Announced in 2009, in beta by February 2011, formally launched September 27, 2011. Two months later the team “unlaunched” it back to beta, admitting it was too hard for new players and too shallow for committed ones. They kept working. On December 9, 2012, Glitch shut down for good. Tiny Speck’s team was scattered across San Francisco, Vancouver, and New York. To coordinate, they had wired up an internal messaging system on top of IRC and then kept adding to it — persistent archives so nothing was lost when you logged off, full-text search across every conversation, file sharing, and a habit of piping automated updates from other systems into the channels. It was never a product. It was infrastructure for building a game, the way a scaffold is infrastructure for building a wall. When the game died, they looked at what was left. Three years of work had produced no viable game and one internal tool the team could no longer imagine working without. The insight was not “workplace chat is a big market.” Plenty of people knew that; HipChat, Campfire, and IRC itself already existed. The insight was that Tiny Speck had accidentally solved the part everyone else treated as secondary — the archive. Their tool assumed that a conversation was a permanent, searchable company asset rather than a stream that evaporated. That single assumption is what made Slack feel less like a chat room and more like an organization’s memory. Slack went into preview in August 2013 and launched publicly in February 2014. It became one of the fastest-growing business software products ever measured. In December 2020, Salesforce bought it for $27.7 billion. Butterfield’s record is now the strongest evidence for the whole pattern. He set out twice to make a game. He failed twice. Both failures produced a company — one sold to Yahoo, one sold to Salesforce for nearly twenty-eight billion dollars. In neither case was the valuable thing visible at the start. In both cases it was sitting inside the wreckage of something else, and it was only visible because he had spent years building the wrong thing well. There is no version of this story where Butterfield skips the games and goes straight to Slack. The tool did not exist as an idea. It existed as a resid