Flickr Was a Feature Inside a Game Nobody Played

At three in the morning in a New York hotel room, Stewart Butterfield was on his knees in front of the toilet, unable to keep down ginger ale. He had gotten food poisoning on the flight from Vancouver. His company, Ludicorp, was almost out of money. He was supposed to be at a video game conference the next day, but he had no idea what he would tell his team when he got back. The game they had been building for two years was going nowhere.

Somewhere between bouts of nausea, the entire idea for Flickr came to him.

“And this sounds almost made up,” Butterfield would tell Reid Hoffman years later on the Masters of Scale podcast. But it happened. The product that would define an era of the internet, pioneer social photo sharing, and sell to Yahoo within thirteen months was conceived by a man with food poisoning in the middle of the night, out of pure desperation.

A Log Cabin, a Philosopher, and a Game That Would Never End

Butterfield’s origin story reads like it was written by someone trying too hard to be interesting, except every detail is true. He was born Dharma Jeremy Butterfield in 1973 in Lund, British Columbia. His father had fled the United States to avoid the Vietnam War draft, and the family lived on a commune in remote Canada. For the first five years of his life, Butterfield grew up in a log cabin without running water or electricity.

He taught himself to code as a child, changed his first name to Stewart when he was twelve, and eventually earned a philosophy degree from the University of Victoria followed by a Master’s in Philosophy from Cambridge. His academic focus was cognitive science and the philosophy of mind. None of this was a traditional path to founding a tech company.

In 2000, he met Caterina Fake in San Francisco. She initially refused to date him. They married in 2001 anyway. Two weeks after their honeymoon, they founded Ludicorp together with Jason Classon.

The company’s purpose was to build a massively multiplayer online game called Game Neverending. It launched in 2002, during the darkest days of the dot-com bust. The concept was unusual: the game was designed around social interaction rather than combat. Players collaborated to build a world. They could chat, trade objects, manage inventory, and, crucially, share photographs with each other.

Game Neverending attracted a small, devoted community. But “small and devoted” doesn’t pay the bills. The company was burning through its modest funding, and the game wasn’t generating enough traction to justify continued investment. By late 2003, Ludicorp was running out of money.

The Tie-Breaking Vote

What happened next is one of the most human pivot stories in tech history. After Butterfield’s food-poisoned epiphany in New York, he came back to Vancouver with a proposal: stop building the game. Take the photo-sharing feature, the part of Game Neverending that players actually loved, and spin it off into its own product.

The team was not enthusiastic. These were game designers and programmers. They had signed up to build a game, not a photo website. On December 8th, 2003, Butterfield called a vote.

It was a tie.

The holdout was Eric Costello, who handled all the front-end code. He was also the only person on the team who was getting a paycheck. Everyone else was working for free, friends who believed in the project enough to go without pay. Costello had three kids. He needed the income.

Butterfield went to talk to him privately. “If you’d like to continue to get paid, there’s only one way we’re going to do this,” he said. He didn’t mean it as a threat, he later clarified. It was a statement of financial reality. There was no money to continue building the game. The photo project was the only path to survival.

Costello changed his vote. The team revoted. Flickr won, reluctantly.

Caterina Fake described the moment with characteristic bluntness: “We only had money for one last shot, and we knew we could deploy Flickr faster than the game. But first, we had to convince the team.”

Eight Weeks, Then Everything Changed

The team built Flickr in eight weeks, repurposing the real-time interaction technology they had already created for Game Neverending. The infrastructure for sharing, chatting, and interacting around visual content was already there. They just stripped away the game and let the photos stand on their own.

Flickr launched on February 10, 2004, six days after Mark Zuckerberg launched Facebook from his Harvard dorm room. At the time, it was not obvious which product was the bigger deal.

Flickr was immediately different from anything else online. Earlier photo sites were digital albums, places to store images for friends and family. Flickr was social. You could tag photos with descriptive keywords, a concept Butterfield and Fake borrowed from a bookmarking site called del.icio.us. You could mark other people’s photos as favorites. You could comment, discuss, and build community around images. And there was the “Blog this” button, which let users link photos directly to their personal blogs.

“We were very small and very poor,” Fake explained, “so we built a lot of features that were deliberately viral.”

It worked. Flickr became one of the fastest-growing sites on the internet. The timing was extraordinary: digital cameras were getting cheap, broadband was spreading, and blogging was exploding. Flickr sat at the intersection of all three trends. By late 2004, the site was worth roughly $25 million. Even its name, with the missing vowel, spawned an entire naming convention for startups. Tumblr, Grindr, Scribd: they all owe a debt to Flickr’s missing “e” (which happened only because the domain flicker.com was taken and the owner wouldn’t sell).

In March 2005, just thirteen months after launch, Yahoo acquired Ludicorp and Flickr for a reported $25 to $35 million. Butterfield and Fake and their seven employees moved to Silicon Valley.

By 2007, Flickr was the 19th most popular website on the entire internet. By 2008, it hosted over 3 billion images. Time magazine would later write that no Web 2.0 site was more important than Flickr.

The Man Who Pivoted from a Failed Game Twice

Here is where the Flickr story becomes something more than a pivot case study. Because Stewart Butterfield did this exact thing a second time.

After leaving Yahoo in 2008, Butterfield founded a new company called Tiny Speck. Its first project: another massively multiplayer online game. This one was called Glitch, a browser-based, nonviolent world where players roamed the minds of eleven cosmic giants. Someone described it as “Monty Python crossed with Dr. Seuss on acid.”

Glitch launched in 2011. It was charming, strange, and beloved by a small audience. It was also not commercially viable. Butterfield shut it down in 2012, but not before spending a month and a chunk of the remaining venture capital helping every departing employee find a new job.

Then, for the second time in his career, he looked at the internal tools his team had built while working on the game. During Glitch’s development, the team had outgrown IRC and built their own communication platform. They had added integrations, search, file sharing, channels. It was faster and more powerful than anything on the market.

He turned it into Slack.

Slack launched publicly in February 2014 and became the fastest company in history to reach a $1 billion valuation. By 2021, Salesforce acquired it for $27.7 billion. The internal chat tool from a failed video game became one of the most important workplace communication platforms in the world.

Stewart Butterfield is, as far as anyone can tell, the only founder in history who has pivoted from a failed game to a generation-defining communication product not once, but twice. Game Neverending became Flickr. Glitch became Slack. The games died. The features survived.

What Flickr (and Slack) Teach About the Fifth Pattern

This month’s series has explored how companies discover what they’re really building. Twitch found its second product in the data it was ignoring. YouTube’s users brought the product the founders couldn’t imagine. Shopify’s infrastructure outgrew the store. Canva’s founder always knew, and used a narrow first product to prove the thesis.

Flickr adds a fifth pattern: the feature that outgrows the product it lives inside.

This is different from the others in a subtle but important way. Butterfield didn’t ignore the photo-sharing feature. He wasn’t surprised by it. He knew it was popular within Game Neverending. What made the pivot difficult wasn’t seeing the opportunity. It was killing the thing he actually wanted to build in order to pursue it. The game was his passion. Photo sharing was his survival.

That’s a tension product leaders face more often than the pivot literature acknowledges. The question isn’t always “what should we build?” Sometimes the question is “what are we willing to stop building?” The team at Ludicorp could see that photos were working. They voted against the pivot anyway, because they had joined to build a game. The rational choice and the emotional choice pointed in different directions, and it took a tie-breaking vote, a conversation about paychecks, and a company on the edge of bankruptcy to force the decision.

The Butterfield double-pivot also complicates the common advice to “follow your passion.” He followed his passion twice, into two different games, and both failed commercially. The products that succeeded were the pragmatic features he built alongside the passion projects. Flickr and Slack weren’t born from vision. They were born from infrastructure, from the tools a team builds for itself while trying to do something else entirely.

If there’s a single thread connecting all five stories this month, it might be this: your second product is rarely the thing you’re most excited about. It’s the thing your users, your team, or your market keeps pulling you toward while you’re busy building something else. The founders who find it are the ones willing to pay attention to that pull, even when it means letting go of the thing they originally set out to create.

What feature, tool, or process in your current product do people love that you’ve been treating as a side project?