Part IThe Story
A Man Called Markus
In the last week of March 2019, Mojang shipped a development snapshot of Minecraft designated 19w13a. Snapshots are weekly previews of the build in progress, read mostly by modders and obsessives who comb the changelogs for new blocks. This one contained something that wasn't a block. Two mentions of Markus Persson's name had been removed, including from the yellow splash texts that pulse beside the title on the main menu. A month later, on April 29, the press learned that the man who had written the first version of the game alone, in Java, at night, while holding a day job, would not be invited to the celebration of its tenth anniversary. A Microsoft spokesperson said his views "do not reflect those of Microsoft or Mojang." It was a corporate sentence, and it was accurate.
The deletion was incomplete, though, and the gap was not a matter of courtesy. Minecraft has one stretch of real prose in it. When a player kills the Ender Dragon, the boss at the end of the game's only ending, an exit portal opens and the credits roll, followed by roughly 1,500 words written by the Irish novelist Julian Gough. The End Poem takes about nine minutes to scroll past. It is the game's only narrative text, an unsigned dialogue between two voices discussing the player as if the player were asleep, and somewhere in it is a description of the place the player has been living: a flat, infinite world created by a man called Markus.
The name stayed because the poem stayed. Gough had declined to sign his rights to the text over to Microsoft and later gave the words away for free. So the one surviving acknowledgment of the game's author sits inside the one part of the game its owner does not fully own. Nobody planned that arrangement.
Then the poem ends, and the game does something that no prior victory screen had quite done. It drops the player back at the respawn point, in the same world, with the same inventory and the same unfinished houses, and lets play continue indefinitely. The credits are an intermission. After them the world is still there, and so is the player, but the man called Markus is gone.
For anyone trying to understand Persson, that may be the most useful image the game supplies: an infinite world, finished on paper, still running, with its maker reduced to one line in a text nobody is required to read. Over twelve years he was a programmer, a founder, a billionaire, a pariah, and a living rebuke to a lot of what venture capital believes about how valuable companies are built. He wrote the best-selling video game in history, in the most literal sense of wrote. He sold it for $2.5 billion, called himself "not an entrepreneur," bought a $70 million house in Beverly Hills, and spent the next decade trying and mostly failing to make anything else. Almost every chapter of that story has a cave in it somewhere.
By the Numbers
The Infinite World
400M+Copies of Minecraft sold as of May 2026, the most of any single video game
$2.5BMicrosoft's price for Mojang, announced September 15, 2014
71%Persson's stake in Mojang at the time of the sale
$1.8BPersson's personal proceeds from the deal
$129MMojang's 2013 profit, on $330M of revenue, with no outside investors
SEK 60MSpent by Persson's post-Mojang studio, Rubberbrain, by 2021, with no game released
1 trillionMinecraft-related views on YouTube by December 2021
The Modem in the Cabin
The father was Birger Persson. He worked for the railroad in Edsbyn, a small town in central Sweden, and his son later described him as "a really big nerd." He built his own modem. He owned a Commodore 128 and taught the boy to use it. He also, by his son's account, had depression, bipolar disorder, alcoholism, and a problem with medication, and he went to jail for robberies. He got better for a while during Markus's early childhood, then relapsed, and the relapse helped end the marriage. After the divorce he moved to a cabin in the countryside. For several years his son had no contact with him.
The mother was Ritva, a Finnish nurse. After the split, when Markus was about seven, she took him and his older sister back to Stockholm, where he had been born on June 1, 1979. The household was poor enough that Persson later said they ran short of food about once a month. The sister experimented with drugs and eventually ran away from home. Before that, she was the one who sat beside him at the Commodore and helped him type in programs printed in computer magazines, line by line, the way you copied software when the code arrived on paper.
So the machine and the instability came from the same person, and probably came at the same time. The Commodore taught the boy the most important fact about programming, which is that a world can be built entirely out of instructions you control. Around it was a family that could not be controlled at all. He started programming at seven. At eight he wrote his first game, a text adventure. The first game he bought with his own money was The Bard's Tale, a dungeon crawler about descending into the dark under a city. Decades later, that franchise ended up in Xbox Game Studios' catalog alongside the one he made, and both now belong to the same company in Redmond. The irony is real, though small.
Edsbyn should get its own sentence, because Persson spent his first seven years there exploring the woods with friends, and it is hard not to notice how many hours of the most-played game ever made are spent alone in procedurally generated forests. He was introverted but well liked as a small child. By secondary school he was a loner with reportedly one friend, spending his free time on games and code at home. In 1994, at fifteen, he knew he wanted to make video games. His teachers told him to study graphic design, and he did, from fifteen to eighteen. He never finished high school, although by most accounts he was a good student. Later he joined Mensa.
The family weather he grew up in matters here, and it isn't decoration. Persson later said that he, too, lived with depression and with highs and lows in his mood. Most of the public story of his life, the giddy tweets and abrupt exits, the parties and the long silences, the regrets announced and then retracted, follows that oscillation more closely than any business model could.
The most important review of Minecraft ever written was a single sentence, delivered by a man who lived in a cabin, about a cave.
That review comes later.
The Portable Name
The early career went the way such careers usually go. Persson worked as a web designer and then took a job at a company called Game Federation, where he met Rolf Jansson. In their spare time the two built Wurm Online, a sprawling multiplayer sandbox about digging, building, and surviving in a persistent world. They had begun it around 2003 under the working name "Mojang Specifications," from the Swedish mojäng, which means roughly "gadget." When the game started turning a profit, they incorporated Mojang Specifications AB on May 14, 2007.
Persson left the project later that year. He didn't take the code or the game. He took the name, or at least asked to, and Jansson agreed to rename the company. It became Onetoofree AB. That name was hard to say aloud without spelling it out, so in October 2011 Jansson renamed it again, to Code Club AB. Code Club still exists, in Motala, about three hours southwest of Stockholm, and Jansson still runs it. On July 8, 2019, a Stockholm company called Game Chest bought 82.9 percent of it for 3.37 million kronor, about $335,000, part in cash and part in newly issued shares.
That makes two companies called Mojang, founded two years apart by the same man with two different partners, both built around persistent sandbox worlds of digging and building. One sold for $335,000. The other sold for $2.5 billion. The gap is too large to be a judgment on the idea, which was more or less the same. It reflects timing, form factor, and the kind of luck that only feels inevitable afterward. What Persson carried from one company to the next was a word, a gadget-word, and he was so sure he would need it that he made his partner rename the company they had built together.
His day job in those years was at Midasplayer, the Stockholm company later known as King, which became famous for Candy Crush. Between 2004 and 2009 he built browser games there in Flash, somewhere between twenty-five and thirty of them, and learned several programming languages along the way. He left when King barred him from making games in his free time. It was a reasonable policy for a company that sold games and an intolerable one for a man whose only real hobby was the job. In spring 2009 he moved to jAlbum, a photo-album software company, and kept prototyping at night.
Two people from those day jobs would turn out to matter more than anything he built there. Jakob Porsér had worked with him at King. Carl Manneh was a manager at jAlbum. Within eighteen months both would be sitting in the first offices of the second Mojang.
Cave Game
Persson spent those evenings on the forums of TIGSource, a gathering place for independent developers in the late 2000s, where he entered design competitions and talked shop with strangers. He built small, odd things in public. One of them, RubyDung, was an isometric base-building game somewhere between RollerCoaster Tycoon and Dwarf Fortress. While working on it he tried a first-person view like the one in Dungeon Keeper, then dropped it because the graphics looked too pixelated.
Then, in April 2009, a small studio called Zachtronics released Infiniminer, a block-based mining game. Persson played it and saw what his discarded camera had been missing. The blockiness he had been embarrassed by was the style. He brought back the first-person view and adopted the cubes, along with the core loop of breaking blocks and placing them. In May 2009 he started what Mojang's own corporate history describes, without apology, as a clone, reusing assets and engine code from RubyDung. On May 13 he uploaded a video to YouTube titled "Cave game tech test." The world in it was made of grass and cobblestone, and blocks could be placed and removed. That was all of it.
Four days later, on May 17, 2009, he posted a more refined build to TIGSource. The forum named it. Users suggested Minecraft: Order of the Stone, and the subtitle was soon dropped. He wanted something Infiniminer lacked, the progression and stakes of a role-playing game, and he went looking for them in public, one patch at a time, updating the game according to what the forum said. That phase was later called Classic. A branch called Survival Test added a health bar and the first hostile monsters, zombies and skeletons and creepers. Before that change, the caves had been empty.
Then he did the thing that made the rest possible. On June 13, 2009, less than a month after the first public build, he started charging money for an unfinished game. Five days later he registered a sole proprietorship under the recycled name Mojang Specifications. Its registration number at the Swedish Companies Registration Office is 19790601-0173. The first eight digits are his birthday. At that point the company and the man were, on paper, the same thing.
He sold it through his own website and split revenue with no publisher, no store, and no platform. That independence nearly broke on its first stress test. When money started arriving faster than PayPal's fraud models thought a solo developer's money should, PayPal froze his account. The money was real. A system designed to catch thieves had no category for a one-man game studio that was suddenly working.
The development log from that period reads like the diary of someone sprinting downhill. Indev went public on December 23, 2009, after players asked for it, and added crafting and smelting, the loop that still defines the game. Its last update, on February 23, 2010, added paintings by the Swedish artist Kristoffer Zetterstrand. Four days after that, Persson opened a branch called Infdev to try infinite worlds. Alpha arrived on June 30, 2010. Redstone came with it, a material that carries a signal and turned out to be Turing complete. Players later used it to build working computers inside the game. On October 30, 2010, the Halloween Update added biomes and the Nether, a hell dimension of lava and shrieking, fireball-spitting ghasts. Beta came on December 20, and the price rose from €10 to €14.75. People who paid early had paid less for less, and they knew it, and they paid anyway.
He had collaborators by then, though not employees. Daniel Rosenfeld was a German musician who used the name C418 and had been active on TIGSource since 2007, where he and Persson met. Rosenfeld was working with an engine he would later call "terrible"; it crashed if two song files played at once. Some of the most familiar sounds in modern culture came out of improvised Foley. Real grass recordings sounded wrong, so grass became lightly touched VHS tape. The spider's screech came from a fire hose recording pitched down until it sounded, as he put it, "like a weird spider was talking to you." The creeper's hiss, the sound that has kept a generation of children awake at night, started as a placeholder matchstick noise he made with Persson, and it "seemed to work hilariously well." They kept it. The game's most famous monster speaks with a temporary sound. Persson would later call the creepers themselves "a mistake."
Rosenfeld's music was ambient, sparse, and, in his words, "repetitive by design." It played in random order at random moments, and he believed the randomness was the point. Players would be doing something when a track started and would decide, he said, that "'Yeah, this moment is specifically made for me.' Which is fascinating!" That is a decent account of the whole game. It offered no story and no goals and, at first, not even a tutorial. Reviewers complained that players had to learn crafting recipes from a wiki. In place of all that it provided a world, and the world made each player feel it had been made for them.
There was no advertising. The game spread by word of mouth, by unpaid mentions in the webcomic Penny Arcade, and above all by video. By May 2012 more than four million Minecraft videos had been uploaded to YouTube, mostly screen captures with voice-over. By April 2011 Persson estimated the alpha had sold 800,000 copies and the beta more than a million, about €23 million in revenue. Before the official release, the unfinished game had 16 million registered users and 4 million purchases.
When he looked back on that year in a 2010 interview, Persson described the method with almost no grandeur.
— Markus Persson, 2010 interviewThe vision for Minecraft was very similar to where Minecraft alpha [was then], but I focused on just getting the engine written and making sure that the controls felt smooth.
He got the engine written and made the controls feel smooth, and the rest followed.
The Job in Bellevue
In September 2010, Persson flew to Bellevue, Washington, to visit Valve, the company behind Half-Life, Portal, and the Steam store, and at the time probably the most admired studio in the world among the kind of programmers who hung out on TIGSource. He took part in a programming exercise and met Gabe Newell. Valve offered him a job. The episode is recounted in Minecraft: The Unlikely Tale of Markus 'Notch' Persson and the Game that Changed Everything, and its plot is the decision. He said no.
He went home and called Jakob Porsér. Porsér quit his job almost immediately, and on September 17, 2010, the two of them incorporated Mojang AB. Its registration number, 556819-2388, does not contain anyone's birthday. The company was now something other than its founder, which is what incorporation is for, though in this case the separation seems to have meant more than usual.
Persson kept working on Minecraft. Porsér started a card game called Scrolls. Neither of them wanted to run a company, so they hired someone who would: Carl Manneh, the jAlbum manager, became CEO. Daniel Kaplan came on as business developer and Markus Toivonen as art director. Jens Bergensten was hired as lead programmer, a quiet coder who would eventually own more of the game's lived history than its creator did. The offices were on Maria Skolgata 83 in Södermalm. In November 2012 the company still had only twenty-five employees.
What happened next is the most important strategic fact in Mojang's history, and it is a list of refusals. Around the end of 2011, Manneh held talks with a long line of venture firms, Sequoia Capital and Accel Partners among them, and turned down all of them. The company didn't need money. Sean Parker, who co-founded Napster and was the early president of Facebook, offered to invest personally and was also turned away. Mojang publicly ruled out selling itself or going public. Even as Mojang turned money away, Persson gave some of his own away: in 2011 he handed $3 million in dividends back to his employees.
The refusals had a matching ideology. In 2011 Persson said he was a member of the Pirate Party of Sweden. He said "piracy is not theft" and treated unauthorized downloads as future customers. He asked large publishers a question meant to have no good answer: "if a pirated game is a lost sale, should bad reviews be illegal?" When Facebook bought the virtual-reality company Oculus in 2014, he cancelled plans for a Minecraft port to the Rift with four words: "Facebook creeps me out." The independent developer's whole catechism of that era is in those positions: own the store, refuse the money, distrust the platforms, let the users help themselves.
It also produced numbers that made the catechism look like strategy. By March 2012 Minecraft had sold 5 million copies, about $80 million in revenue. Mojang's total revenue for 2012 was $237.7 million. In 2013 it was $330 million, with $129 million in profit, from a company with no investors, no board to speak of, and a founder who had stopped working on the product.
Fourteen Days in December
The fall of 2011 had the shape of a coronation. Mojang rented the Mandalay Bay in Las Vegas for the first Minecon on November 18 and 19, and on the first day, during a presentation, Minecraft was formally released as version 1.0, more than two years after strangers on a forum started paying for it. The release added the End, a dimension of dark islands floating over a void, and the Ender Dragon. Gough's poem scrolled for the first time. Persson gave a keynote. That year he had won Best Debut Game, Best Downloadable Game, and the Innovation Award at the Game Developers Choice Awards, every category the game was nominated in. The next spring BAFTA would give him its Special Award. In 2011 he also married Elin Zetterstrand, whom he had dated for four years and who had once been a moderator on the Minecraft forums. They had a daughter.
The sequence of the next few weeks is best told by date, because the dates say more than an explanation would.
On December 1, 2011, Persson handed creative control of Minecraft to Jens Bergensten. The blog post announcing it had a Swedish title, Och med dom orden så passar jag micken: "And with those words, I pass the mic." A creator giving up the most successful game in the world twelve days after its release has almost no precedent. He said he wanted to work on something new.
On December 14, 2011, his father shot himself after drinking heavily.
From December 16 to 19, the 22nd Ludum Dare competition took place, a 48-hour game jam in which developers build a game from scratch on a theme, alone, against the clock. Persson entered. In two days he built Minicraft, a small top-down survival game written in Java. It is a miniature of the thing he had just given away.
Nothing in the record connects the second date to the third, and we shouldn't invent a connection. Still, the fact stands that two days after his father's death, the man who had just handed off the largest world anyone had ever built sat down and built a smaller one by himself, in two days, against a deadline. A Ludum Dare entry is a contained problem with its own end condition. It is, among other things, a place to be.
Later Persson talked to The New Yorker about his father. He didn't talk about the jail or the bottles. He talked about the one person who had backed him.
— Markus Persson, to The New Yorker, 2013When I decided I wanted to quit my day job and work on my own games, he was the only person who supported my decision. He was proud of me and made sure I knew. When I added the monsters to Minecraft, he told me that the dark caves became too scary for him. But I think that was the only true criticism I ever heard from him.
The only true criticism he ever heard from his father was that the dark had gotten too dark. This is the review mentioned earlier. In Minecraft, hostile mobs spawn wherever the light is low enough, in caves, under overhangs, and at night, when a full day-night cycle takes twenty real minutes. Reviewers have written about the "first night" as the moment every new player remembers: running from zombies and skeletons, or digging into the ground and waiting there for morning. The answer the game gives is torches, which take a stick and a piece of coal and which players learn to place everywhere, compulsively, so that the dark cannot fill with things. Birger Persson, who had lived through much worse than a cave, didn't want to go down there anymore.
By mid-2012 Persson was seeing little of his daughter. On August 15, 2012, he tweeted: "As of today, I am single. #mixedemotions." The divorce was finalized later that year.
Ten to the C
The new thing was supposed to be a space game. In March 2012 Persson announced a sandbox space trading and combat simulator in the tradition of Elite. On April Fools' Day, Mojang put up a satirical website for a game called Mars Effect, which mocked Mass Effect and cited the studio's ongoing legal fight with Bethesda as its inspiration. Three days later the real title appeared: 0x10c, pronounced "Ten to the C." It was set in a parallel universe in the year 281,474,976,712,644 AD, and players were going to program a simulated 16-bit computer to fly their ships. It was the most Persson-like design imaginable, a game in which the player's main tool was the thing that had saved him as a boy.
Persson halted production in August 2013. He gave two reasons: lack of interest and creative block. Rosenfeld, who had composed the soundtrack, later released the music as an album, so of the game that wasn't, only the score survives.
The Scrolls fight went better as theater than as law. In August 2011, after Mojang filed to trademark the name of Porsér's card game, ZeniMax Media, the parent of Elder Scrolls publisher Bethesda, sent a cease-and-desist letter claiming the single word scrolls infringed its franchise. Persson offered to drop the trademark and add a subtitle. Mojang ignored the letter, and ZeniMax sued. On August 17 Persson challenged Bethesda to settle it with a Quake 3 tournament. On September 27 he tweeted, "The Scrolls case is going to court! Weee! :D". In October Mojang won an interim injunction, the court finding the two names easy to tell apart. In March 2012 the parties settled. Mojang transferred its "Scrolls" trademarks to ZeniMax, which licensed the name back for that one game and nothing after it. Scrolls shipped in December 2014, stopped receiving new content in 2015, and was relaunched for free in 2018 under the name Caller's Bane.
A third project shows the instinct underneath the other two. In 2011 Persson and Kaplan agreed with the Lego Group to build a hybrid of Minecraft and Lego bricks, codenamed after a character in Napoleon Dynamite. On September 5, 2011, he tweeted: "Secret project Rex Kwon Do begins today! It will involve caffeine and giggling." Mojang hired two programmers. Persson wrote the prototype. Six months in, the project died under layers of legal requirements meant to protect Lego's family-friendly image. Kaplan said the Mojang team had felt more like consultants than designers. On July 19, 2012, Persson explained the cancellation in one sentence that works as a statement of principle: "Remember Project Rex Kwon Do? We cancelled it to focus on Minecraft, Scrolls and eventually 0x10c that we own 100% ourselves instead."
In 2025, an archival group called Omniarchive recovered a build of the game dated June 28, 2012, about a month before it was killed. The Lego Group, for its part, considered buying Mojang outright around that time and decided against it because it couldn't see how big Minecraft would become. Lego, a company that built an empire selling small blocks to children, passed on the most successful small blocks ever made.
The pattern of these years is not that Persson stopped being able to make things. In 2012 Mojang spent one week building Catacomb Snatch for a Humble Bundle charity jam called Mojam; the bundle sold 81,575 times and raised $458,248.99. That year he also gave $250,000 to the Electronic Frontier Foundation. In 2013 he made a free game called Shambles in Unity. He could still make small things quickly. What he couldn't seem to do was make a large thing a second time, deliberately, with the whole world watching.
A Tweet Worth $2.5 Billion
The end began with a license agreement. In June 2014, Mojang announced that it would start enforcing a section of the Minecraft end-user license agreement that had been on the books for three years. The section barred multiplayer servers from selling in-game advantages for real money. Servers could still charge for entry and could still sell cosmetics. What they could not do was sell power. Persson backed the change. He cited emails from parents whose children had spent hundreds of dollars on servers. To him it was plainly the right thing.
Server owners saw it as an attack on their income. They argued that the EULA's language was broader than Mojang claimed, that smaller servers would be forced to close, and that Mojang was clearing the field for its own subscription hosting service, Realms. The community that had built the game alongside him, by posting on forums, uploading videos, and paying for an alpha, turned against him in numbers he had never seen pointed his way.
Persson tweeted, reportedly half as a joke:
— Markus Persson, on Twitter, June 2014Anyone want to buy my share of Mojang so I can move on with my life? Getting hate for trying to do the right thing is not my gig.
The phone rang. According to Persson, a Microsoft executive called Manneh soon afterward to ask whether he meant it. Activision Blizzard and Electronic Arts also showed interest. Mojang hired advisers from JPMorgan Chase. On June 23, 2014, the same month as the tweet and nearly three months before any deal was announced, Persson and Porsér registered a new company called Rubberbrain AB.
The buyer's reasons were more ordinary than the seller's. Phil Spencer, the head of Xbox, pushed Satya Nadella, who had become CEO that February, to buy Mojang as "a pretty bold vision" for Microsoft's gaming business. Nadella later said the chance to use Minecraft with HoloLens, Microsoft's mixed-reality headset, had been a major factor. There was also the money. Microsoft held $2.5 billion in offshore accounts that it couldn't bring back to the United States without paying repatriation taxes. Forbes later called the deal a "tax dodge," a way to turn idle foreign cash into an asset. The purchase price matched the stranded cash almost exactly, so the most successful independent game ever made was paid for with money that otherwise had nowhere useful to go.
Microsoft announced the agreement on September 15, 2014. Persson owned 71 percent of Mojang and received $1.8 billion. Porsér got $300 million and Manneh $100 million. The deal closed on November 6. All three left.
The same day the deal was announced, Persson published a blog post headed "I'm leaving Mojang." He wrote that he had realized he "didn't have the connection to my fans I thought I had," that he had "become a symbol," and that he did not want to be responsible for a company that had grown so large. Then came the line that will be quoted as long as anyone remembers him:
— Markus Persson, "I'm leaving Mojang," September 15, 2014I'm not an entrepreneur. I'm not a CEO. I'm a nerdy computer programmer who likes to have opinions on Twitter.
Every clause in that sentence turned out to be true, including the last one.
The people who stayed behind remembered it differently. Bergensten, who by then had been running the game for almost three years, said the sale went against the studio's indie culture. Some employees cried at the office. Everyone who stayed six months received a bonus of roughly $300,000 after taxes, and it was deducted from Persson's share, not Microsoft's. It was a generous act and also a way of buying permission to leave.
Microsoft, under Matt Booty's oversight, integrated the studio lightly, leaving it independent in practice but backed by the parent company's money and engineering. Industry observers later described that light touch as the model for how Microsoft would buy studios afterward. In that sense Persson's exit taught one of the world's largest companies how to acquire creative studios without destroying them, and the person it taught that lesson by example was him leaving.
There was a pause. Minecraft 1.8, the Bountiful Update, came out on September 2, 2014, thirteen days before the announcement. The next major update didn't arrive until February 2016.
Where the Rich People Live
Around 2014, Persson lived in a multilevel penthouse in Östermalm, a Stockholm district he described, with the unsettled candor of a man who hadn't grown up there, as "where the rich people live." That year he was one of the largest taxpayers in Sweden.
In December 2014 he paid $70 million for a house in Trousdale Estates in Beverly Hills, a record price for the city. Forbes noted that it had a car lift. He reportedly outbid Beyoncé and Jay-Z. The boy who had gone short of food once a month in Stockholm, the son of a man who moved to a cabin, now owned one of the most expensive houses in America, in a city where he was mostly a stranger.
The lifestyle coverage that followed, of the parties and the money and the "hard-partying founder," as one magazine put it, is easy to report and hard to read honestly. A sharper index of his state of mind is a small game. On June 9, 2014, the first day of E3, the industry's big trade show, three months before the sale was announced, Persson released Cliffhorse. He made it in two hours in Unity using free assets. It is about a horse and some cliffs. He said the inspirations included Skyrim's physics engine, Goat Simulator, the notoriously broken Big Rigs: Over the Road Racing, and "the more embarrassing minimum-effort Greenlight games." It was distributed as "honourware": players were asked to pay in Dogecoin, a cryptocurrency that began as a joke. They sent more than 280,000 dogecoins.
That is a man playing. It is also a man making fun of the industry's mechanics (early access, crowdfunding, shovelware) on the biggest day of the industry's calendar, using a game built in less time than a meeting. Cliffhorse shipped. 0x10c did not.
Rubberbrain, the company registered on June 23, 2014, had by 2021 spent 60 million kronor and released no games, according to the Swedish newspaper Aftonbladet. In 2020 Persson talked about starting a new studio and about making virtual-reality games, and he praised the VR shooter Boneworks. He staged a series of immersive, narrative-driven events with heavy visual effects under the name
.party(), written the way a function call is written in code, and held them in several cities. On March 28, 2024, Rubberbrain was relaunched as Bitshift Entertainment, LLC. It teased a first-person roguelike called Levers and Chests.Something is visible in the economics of all this, and it doesn't flatter money. With 48 hours and no budget, Persson made Minicraft. With two hours and free assets, he made Cliffhorse. With 60 million kronor and seven years, the record shows nothing shipped. The constraint had been the engine.
The Far Lands
When Persson built infinite terrain into Minecraft in early 2010, the math held up only so far. More than 12 million blocks from the center of a world, the generator began to fail in a visible way. Terrain rose into tall walls riddled with fissures and stacked into strange, vertical, unwalkable patterns. Players called it the Far Lands. For years it was treated as the effective edge of the world, the place where infinity gave out. One long-running YouTube series is just a man walking there, on foot, in an old version of the game, a journey measured in years. Mojang eventually fixed the glitch and put an impassable barrier at 30 million blocks. The Far Lands still exist in the old builds.
Starting in 2016, Persson's public life began to look like the Far Lands: what a mind produces when it has been pushed far past the point where it was designed to make sense, and keeps generating anyway.
The record is long and documented in his own posts. In 2017 he proposed a heterosexual pride holiday and wrote that people who opposed it "deserve to be shot." After the backlash he deleted the tweets and wrote, "So yeah, it's about pride of daring to express, not about pride of being who you are. I get it now." That same year he called feminism a "social disease" and called the developer Zoë Quinn a "cunt," even though he had generally been critical of GamerGate, the harassment campaign that targeted her. He described intersectional feminism as a "framework for bigotry." On November 30, 2017, he tweeted, "It's ok to be white." He said the Pizzagate conspiracy theory had merit. In 2019 he wrote, "Q is legit. Don't trust the media." In March 2019 he responded to a pro-transgender meme by calling encouragement of "delusion" "absolutely evil," and then the next day wrote that he had "no idea what [being trans is] like of course, but it's inspiring as hell when people open up and choose to actually be who they know themselves as. Not because it's a cool choice, because it's a big step. I gues that's actually cool nvm."
That last tweet is the whole pattern in miniature: provocation, recoil, reconsideration, and a shrug spelled "nvm." The phrase "highs and lows," his own description of his moods, is useful here only because he chose it. It explains nothing and excuses nothing. It does supply the shape.
Some of the most pointed evidence in the case is an earlier post of his. In 2012, explaining that every creature in the game besides the player was genderless and showed the traits of both genders, Persson wrote on his blog that this meant "every character and animal in Minecraft is homosexual because there's only one gender to choose from. Take THAT, homophobes!" It was a joke, but a revealing one: in 2012 the joke was aimed at bigots. By 2019 Minecraft had become one of the most important places on the internet for queer kids and queer creators. Mojang marched at Stockholm Pride in 2016. The world he built had moved one way, and its builder had moved the other.
Microsoft's response came in two parts: the snapshot in March and the anniversary in April. In August 2020 Persson deleted his Twitter account after telling a game-design YouTuber to "drop the politics." He came back later.
In early January 2025 he posted a poll on X asking whether he should make a spiritual successor to Minecraft. The answer came back yes. He said he was "100% serious," that he had "basically announced Minecraft 2," and that he wouldn't "sneakily infringe" on Mojang's work. A few days later, after talking to his team, he dropped the idea and chose the other option in the poll, the roguelike, Levers and Chests. That sequence is as clean a portrait of the man as any: the impulse to rebuild the world he lost, put to a public vote, then quietly set aside by the people around him who had to do the work.
Meanwhile the world itself kept running, without him and in front of everyone. Minecraft passed 176 million copies in May 2019, the month of the anniversary he didn't attend, and became the best-selling video game in history. It reached 300 million in 2023 and more than 400 million by May 2026. The Chinese edition alone counted over 700 million registered accounts. Teachers built animal cells in it. The British Museum rebuilt itself inside it. Reporters Without Borders built an Uncensored Library inside it, a neoclassical building made over about 250 hours by twenty-four people, holding the banned journalism of writers including Jamal Khashoggi, because a children's block game turned out to be one of the few places censors weren't looking. An MIT researcher said that the man called Notch had effectively introduced forty million people to the functional logic of computer-aided design. A Minecraft Movie opened in April 2025 with a record $163 million weekend for a video-game adaptation.
None of that required him, and in a sense that was the design. A game with no goals, no story, and no tutorial, in which players supplied their own meaning, was always going to outlive its author, because its author had put so little of himself into it on purpose. He gave people a flat, infinite world and a pickaxe and got out of the way. When he finally got all the way out, the world didn't notice. Perhaps that is the highest compliment a designer can be paid, and perhaps it is also the loneliest.
Rosenfeld, who made the sounds, still owns his two albums. Licensing complications with Microsoft have blocked the release of a third soundtrack, which he says is finished and longer than the first two combined. Gough gave his poem away. Jansson kept the other Mojang until it sold for $335,000. Bergensten kept the game. Each of them held on to something. Persson sold nearly everything, and what he could not sell is a single line in someone else's poem.
The first public evidence of all of it is a video uploaded to YouTube on May 13, 2009, titled "Cave game tech test." It shows a small world of grass and cobblestone under a flat sky. A block is placed, and a block is removed. There are no monsters yet, and the caves are empty.
Part IIThe Playbook
What follows is an attempt to take apart the mechanics of an unrepeatable success, and of the failures that came after it, without pretending the two can be separated. Persson's story doesn't yield a tidy method. Some of his best decisions came from temperament, not judgment, and some of his worst came from the same temperament. Still, the decisions are documented, the outcomes are measurable, and the patterns are clear enough to be useful to anyone building something that people are supposed to live inside.
The principles
- 1.Charge for the scaffolding.
- 2.Clone the mechanic, then go looking for the missing feeling.
- 3.Carry the name, not the company.
- 4.Own the till, and expect the till to be questioned.
- 5.Refuse money that would buy your attention.
- 6.Ship the placeholder that works.
- 7.Build a world, not a story.
- 8.Hand off at the summit, but know what you're handing off to.
- 9.Name the job you are actually suited for.
- 10.Price the buyer's problem, not your product.
- 11.Treat constraint as the engine, and wealth as its enemy.
- 12.Design the product to survive its author.
⛏
The Decision Record
Key choices from the first Mojang to the last poll.
2007
Leaves Wurm Online and Mojang Specifications AB, but asks to keep the name "Mojang." Rolf Jansson renames the company Onetoofree AB.
2009
Leaves King after being barred from side projects. Posts Minecraft to TIGSource on May 17; begins charging June 13; registers Mojang Specifications June 18.
2010
Turns down a job at Valve in September. Incorporates Mojang AB with Jakob Porsér on September 17; hires Carl Manneh as CEO.
2011
Releases Minecraft 1.0 at Minecon on November 18; hands creative control to Jens Bergensten December 1. Returns $3M in dividends to staff. Mojang declines Sequoia, Accel, and Sean Parker.
2012
Cancels the Lego project to focus on games Mojang owns "100%." Announces 0x10c.
2013
Shelves 0x10c, citing lack of interest and creative block. Mojang earns $129M profit on $330M revenue.
2014
EULA enforcement backlash; June tweet offering his stake. Microsoft buys Mojang for $2.5B; Persson takes $1.8B and funds $300K retention bonuses from his share.
2019
Name removed from Minecraft splash texts; not invited to the tenth anniversary.
2025
Polls X on a "Minecraft 2"; abandons it within days in favor of Levers and Chests.
Principle 1
Charge for the scaffolding
Persson started taking money for Minecraft on June 13, 2009, less than a month after the first public build. That build was a world of grass and cobblestone. By the beta in December 2010 the price had risen from €10 to €14.75, and the logic was simple: early buyers paid less for an unfinished thing in exchange for the right to watch it grow and to shape it. By April 2011 the alpha and beta together had brought in roughly €23 million, all before an official release. When 1.0 arrived on November 18, 2011, reviewers found little to separate it from what players already had. One Italian critic called the release "a formality."
The point isn't just that early access works as financing, though it does, and Minecraft is widely credited with making the model standard in independent games. The point is that charging for the unfinished product changes what the product is. Paying players weren't customers waiting for delivery. They were investors in a direction. They had standing to complain, which is the most valuable thing a developer working alone can get, because complaints are free product management. Persson updated the game according to forum feedback, and the forum named the game.
The risk is the mirror image of the benefit. People who feel they bought a share of the direction will eventually feel betrayed by a change in direction. The EULA backlash of 2014 was the bill for the 2009 deal. A community that had been told, implicitly, that the game was partly theirs reacted to a licensing decision as if it were a constitutional crisis.
Tactic: Put a price on your earliest usable version, and raise it on a published schedule so that early buyers understand they are being rewarded for risk rather than subsidizing latecomers.
Principle 2
Clone the mechanic, then go looking for the missing feeling
Mojang's own corporate history calls the starting point of Minecraft a clone of Infiniminer, Zachtronics' April 2009 block-mining game. Persson took the camera, the cubes, and the break-and-place loop, and he took them openly. The difference was in what he felt was missing: he wanted role-playing elements, meaning stakes and progression and a reason to come back the next night. Those additions, which included survival, health, monsters, crafting, and the Nether, are what made the game a world and not a toy.
There is a quieter lesson in RubyDung. Persson had already tried a first-person camera and dropped it because the pixelation embarrassed him. Infiniminer didn't teach him a new technique. It showed him that a constraint he had treated as a flaw could be the style. The blockiness he was avoiding became the most recognizable visual identity in modern games.
Originality in this story doesn't come from the starting point. It comes from the diagnosis. Anyone could have copied Infiniminer, and plenty of people later copied Minecraft, among them Terraria, Ace of Spades, and FortressCraft. What couldn't be copied was the specific judgment about which feeling the original lacked.
Tactic: When you study a competitor's product, write one sentence describing the experience it fails to deliver, and build only toward that sentence.
Principle 3
Carry the name, not the company
In 2007, leaving Wurm Online, Persson left behind the code, the customers, and the company, and took only the word "Mojang." His partner renamed the original entity Onetoofree AB and later Code Club AB. In 2009 Persson put the word on a sole proprietorship whose registration number was his birthday. In 2010 he put it on a real company.
This looks like sentiment, and partly it was. It is also a sound reading of what lasts. Companies are bundles of contracts and code that tie a founder to particular partners and particular products. A name is portable identity. It carried continuity across ventures in Persson's mind, and later in the market's mind, at almost no cost. The rhyme with Code Club is instructive: in 2019 the company that kept the original assets sold a controlling stake for 3.37 million kronor. What the name was attached to mattered more than the name, but the name was what made it feel like a single career and not a string of unrelated jobs.
There is a caution here as well. The man who held on so tightly to the name "Mojang" eventually sold it, and then saw his own name removed from the product that name became famous for. Identity you can carry around can also be taken off you.
Tactic: When you leave a venture, negotiate explicitly for the one asset that represents your identity as a builder, whether a name, a domain, or a handle, even if you leave everything else.
Principle 4
Own the till, and expect the till to be questioned
Minecraft was sold through its own website, so Persson didn't split revenue with any store or publisher. That was unusual in 2009 and almost unthinkable for a game later. It meant every euro of early-access revenue reached the developer, and it meant the relationship with the player, including payment, account, and update, was direct.
It also meant he had no institutional cover. PayPal froze his account on suspicion of fraud, because nothing in its model accounted for a solo developer suddenly taking in that much money. Owning the till meant owning the till's failure modes too. Mojang later made the opposite choice when it was strategically right, outsourcing the Xbox 360 version to 4J Studios. That version sold a million copies within a week of its May 2012 release and set Xbox Live records with 400,000 players online at the same time. Direct sales built the base; platforms grew it.
◧
Conventional Wisdom vs. the Mojang Approach
How the first Mojang departed from the startup default, with the outcome attached.
| Conventional wisdom | Persson's approach | Outcome |
|---|---|---|
| Ship when finished. | Charged for an alpha within a month of first build. | Scale €23M before 1.0 |
| Distribute through platforms. | Sold directly from his own website. | Risk PayPal froze the account |
| Raise capital to grow. | Declined Sequoia, Accel, and Sean Parker. | Scale $129M profit in 2013 |
| Advertise. | Relied on word of mouth and YouTube. | Scale 4M+ videos by May 2012 |
| Founder stays as product lead. | Handed control to Bergensten 12 days after launch. | Try Game thrived; founder drifted |
| Partner with big brands. | Cancelled the Lego project to focus on wholly owned games. | Scan Kept independence |
Tactic: Keep the direct customer relationship for as long as your product can grow without intermediaries, and before your revenue spikes, find out how your payment processor responds to sudden success.
Principle 5
Refuse money that would buy your attention
By late 2011 Mojang had heard from Sequoia, Accel, and many other venture firms, as well as Sean Parker personally. It turned them all down because, in the company's own phrasing, it didn't need the money. It publicly ruled out an IPO or a sale. Observers at the time credited that independence with much of the game's success.
The useful distinction isn't between good money and bad money. It is between money that funds work you couldn't otherwise do and money that buys a claim on your attention: board meetings, growth targets, expectations about the next product. Mojang was printing cash. Outside capital would have bought nothing it needed and cost a lot of what it valued, mostly the freedom of a founder who wanted to make small, strange games. The same reasoning led Persson to cancel the Lego partnership in July 2012 so Mojang could concentrate on games "that we own 100% ourselves." He gave up a project with a household brand because it turned his designers into consultants.
Independence also had a cost that shows up later. A company with no investors also has no institutional machinery for succession or for a founder who wants out. When Persson wanted out, the only route was a full sale.
Tactic: Before taking capital, write down what specific capability it buys that your revenue cannot, and if you can't name one, decline.
Principle 6
Ship the placeholder that works
The creeper's hiss was a temporary matchstick sound Rosenfeld made with Persson, and it "seemed to work hilariously well," so it stayed. Grass footsteps came from VHS tape because recordings of actual grass sounded wrong. The engine was bad enough to crash when two songs played at once, so the score became sparse, ambient, and random, and that turned out to be its genius, since players read meaning into tracks that arrived at unpredictable moments.
The game's identity is built from things meant to be replaced. The general principle is easy to state and hard to follow: judge the artifact by its effect, not by how it was made. Most teams replace placeholders out of a sense of professional duty. The few that ship the placeholder that works keep a roughness that polish would have sanded off, and in Minecraft's case the roughness was the brand.
Tactic: Before replacing any temporary asset, test the replacement against the placeholder with real users, and keep whichever produces the stronger reaction.
Principle 7
Build a world, not a story
Minecraft has no required goals. Its only narrative text runs after the credits, and the game then sends players back to keep going. Reviewers complained about the missing tutorial and felt the RPG systems, such as enchanting and the dragon fight, were "undercooked" next to the plain pleasure of building. They were right about the parts and missed what the parts added up to. The game is, in Game Informer's phrasing, as much a platform as a game. Players built working computers out of redstone, rebuilt Denmark at full scale from government geodata, and turned the game into a library of banned journalism.
When the Putt-Putt minigolf chain sent Mojang a cease-and-desist letter in 2013 over maps players had built, Mojang's lawyer Alex Chapman put the design philosophy in legal terms: "Suing Mojang for what people do using Minecraft is like suing Microsoft for what people do using Word." That isn't a description of a game. It is a description of a tool, and tools spread in ways stories cannot, because every user writes their own.
There is a commercial edge to this as well. A story ends, and a world has to be maintained. Minecraft has been updated continuously for more than fifteen years, and a fixed narrative could never have supported that.
Tactic: Strip out any feature that tells users what to do until what remains is the smallest set of tools that lets them decide for themselves, then add structure only where they keep asking for it.
Principle 8
Hand off at the summit, but know what you're handing off to
On December 1, 2011, twelve days after the official release, Persson gave creative control of Minecraft to Jens Bergensten. It may be the boldest founder handoff in the history of consumer software, and by the measure that matters most it worked. The game didn't merely survive the change. It grew a hundredfold under a steward who stayed for more than a decade, through the Microsoft sale and the community fights over combat and chat reporting.
The handoff succeeded for the product because Bergensten had been building alongside Persson for more than a year and because the game's design, which was open-ended, systemic, and community-driven, didn't depend on one author's taste. For the founder it succeeded less well. Persson handed off the thing he was best at in order to do something new, and the new things (0x10c, the Lego project, the space game's programmable computer) stalled or died. He kept the title of creator and gave up the daily practice that had made him one.
The lesson isn't "never hand off." It is that leaving a product's daily direction is a decision about two careers, the product's and yours. Persson planned the first carefully and the second barely at all.
Tactic: Before handing off a product you created, write down what your next working day will look like, concretely, and if you can't, delay the handoff or start the next thing first.
Principle 9
Name the job you are actually suited for
"I'm not an entrepreneur. I'm not a CEO. I'm a nerdy computer programmer who likes to have opinions on Twitter." The sentence is self-deprecating, and it is also precise. Persson had known for years what he wasn't. He hired Carl Manneh as CEO in 2010 precisely so that he and Porsér could keep making games. He passed the design lead to Bergensten. He structured Mojang so that the jobs he hated were done by other people.
What he hadn't accounted for was the job nobody can delegate: being the symbol. His September 2014 post says it directly. He had "become a symbol," and the connection to fans he thought he had wasn't real. Role clarity protected him from management. It didn't protect him from fame, which falls on the founder whatever the org chart says. The EULA backlash wasn't about a management decision. It was about him.
His solution was the one available to someone with 71 percent of a company and no investors to answer to, which was to sell. Founders with less leverage have fewer exits, which makes it more important to see the symbol problem coming.
Tactic: List the roles your company requires of you that cannot be delegated (spokesperson, symbol, final arbiter) and decide in advance whether you are willing to hold each one at ten times your current scale.
Principle 10
Price the buyer's problem, not your product
Microsoft didn't pay $2.5 billion only for a game. It paid for a combination of problems: Phil Spencer's need for "a pretty bold vision" for Xbox, Satya Nadella's interest in content for HoloLens, and, by Forbes's account, $2.5 billion in offshore cash that couldn't come home without a heavy tax bill. The price matched the stranded cash almost exactly. The seller's leverage was that the buyer had several reasons to buy, and only one of them was the product.
Persson's opening move, a joking tweet, was the opposite of a banker's process, but it worked because it drew several bidders at once: Activision Blizzard, Electronic Arts, and Microsoft. Mojang then hired JPMorgan Chase to run the process properly. An informal signal followed by formal discipline is a good order of operations.
How Persson handled the exit is just as instructive. He funded roughly $300,000 after-tax bonuses for every employee who stayed six months, and he paid for them out of his own share. Three years earlier he had returned $3 million in dividends to staff. Generosity at exit isn't only ethics. It is also continuity insurance for the buyer, which raises the value of what's being sold.
Tactic: Before negotiating a sale, write down every problem the acquirer solves by buying you, including balance-sheet, strategic, and personal ones, and price against the most expensive of them.
Principle 11
Treat constraint as the engine, and wealth as its enemy
Persson's record as a maker divides almost perfectly along lines of scarcity. With 48 hours and no budget, he built Minicraft, and Prelude of the Chambered and Metagun before it. With two hours and free assets, he built Cliffhorse. With a full studio, money from the most profitable indie game ever made, and no deadline, he built 0x10c until it ran into a "creative block." With a billion dollars and a decade, Rubberbrain spent 60 million kronor and released nothing by 2021.
⏱
Output Under Constraint
Persson's projects sorted by the resources available to them.
| Project | Constraint | Result |
|---|---|---|
| Minecraft (2009) | Nights and weekends, day job, no funding | Shipped Best-selling game ever |
| Minicraft (Dec 2011) | 48-hour Ludum Dare | Shipped |
| Catacomb Snatch (2012) | One-week charity jam | Shipped $458,248.99 raised |
| Cliffhorse (June 2014) | Two hours, free assets | Shipped 280,000+ dogecoins |
| 0x10c (2012–2013) | Full studio, no deadline | Abandoned |
| Rubberbrain (2014–2024) | SEK 60M+, billionaire founder | No release by 2021 |
| Levers and Chests (2024–) | Relaunched as Bitshift | Announced |
The pattern isn't unique to Persson, but his is the clearest version on record, because the same person, with the same skills, generated both columns. Deadlines narrow the choices, and fewer choices make it possible to finish. Wealth widens every option at once, and an open-ended maker given open-ended resources may simply never close the loop.
Tactic: If your resources have grown faster than your output, impose an artificial constraint, such as a fixed deadline, a fixed budget, or a fixed team of one, and treat it as binding.
In March 2019, Microsoft removed Persson's name from Minecraft's menus. In April it left him off the tenth-anniversary guest list. The game didn't flinch. It passed 176 million copies that May, 300 million in 2023, and more than 400 million by 2026. That durability was not luck. It comes directly from a design that never depended on its creator's voice. There is no authored story, no signature character, and no auteur mythology in the product itself. The main character is a blank figure named Steve, or Alex, or whatever skin the player uploads.
For founders, there are two lessons here, and they pull in opposite directions. The first is defensive: a founder's public persona is part of the product's risk, and products that need the founder to keep being lovable are hostage to the founder's worst week. The second is more uncomfortable. A product designed to outlive its author will do exactly that. Persson built something that could run without him and then found out what it was like to be the person it was running without.
The one surviving mention of him is in the End Poem, written by someone else and kept partly because that someone else wouldn't hand over the rights. Authorship lasts where ownership is contested.
Tactic: Audit your product for features that depend on you personally, whether your voice, your taste, or your reputation, and either institutionalize them or remove them before you need to.
Free playbook
Get The Business Model Playbook
58 business models, one visual page each: how the money flows, the metrics that matter, and who runs it. Free when you join the Faster Than Normal email.
Free. No spam. Unsubscribe anytime.
Part IIIQuotes and Maxims
In their words
— Markus Persson, on Twitter, July 19, 2012Remember Project Rex Kwon Do? We cancelled it to focus on Minecraft, Scrolls and eventually 0x10c that we own 100% ourselves instead.
— Markus Persson, quoted by PC Gamer, March 2011If a pirated game is a lost sale, should bad reviews be illegal?
— Markus Persson, on cancelling Minecraft's Oculus Rift port, 2014Facebook creeps me out.
— Alex Chapman, Mojang's lawyer, responding to Putt-Putt, July 2013Suing Mojang for what people do using Minecraft is like suing Microsoft for what people do using Word.
— Daniel Rosenfeld (C418), on the randomness of the Minecraft soundtrackThey decide subconsciously that, 'Yeah, this moment is specifically made for me.' Which is fascinating!
Maxims
- Charge for the scaffolding. An unfinished product sold honestly turns customers into co-authors and their complaints into free product management.
- Diagnose the missing feeling. Copying a mechanic is easy; what's valuable is your judgment about what the original fails to make people feel.
- Carry the name. When you leave, take the asset that holds your identity as a builder, even if you leave everything else behind.
- Refuse money that buys your attention. Capital is only worth taking if it funds something your revenue can't.
- Ship the placeholder that works. Judge assets by effect, not origin; a matchstick hiss became the most feared sound in gaming.
- Build a world, not a story. Tools spread because every user writes their own ending.
- Plan both careers before the handoff. Leaving a product's direction is a decision about the product's future and your own, and only one of them takes care of itself.
- Know which jobs can't be delegated. You can hire a CEO, but you can't hire someone to be the symbol in your place.
- Price the buyer's problem. Acquirers pay for what your company solves for them, and the product is only one of those things.
- Guard the constraint. Deadlines and scarcity produced every game Persson finished; money and open time produced none.
