You Don't Need a CFO to Make World of Warcraft

Lex Fridman begins with the part of Jeff Kaplan's résumé that most profiles skip: before World of Warcraft and Overwatch, he was a legendary player. Kaplan says he was lucky to be born into the golden age of coin-op. He first saw Pac-Man with his Uncle Ronnie, who kept feeding him quarters, apparently too shy to play himself. At his brother's graduation in Philadelphia, an Asteroids cabinet in the hotel lobby became a high-score hunt that ended in the usual childhood tragedy — they had to go to bed, and by morning someone else had beaten the score.

Growing up in Southern California in the 1980s, born in 1972, he lived in BMX-and-skateboard range of pizza parlors, liquor stores, and arcades. Intellivision mattered more than Atari because his father, an executive recruiter, had Mattel as a client and brought home discounts and free games. The leap that still feels warm to him was the NES and Super Mario. He and his brothers played for weeks, then a friend came over and showed him secrets he did not know existed. Suddenly, he says, the world opened up, and games could be more.

His mother bought Zork, the Infocom text adventure, for the family's IBM PC. You typed "go west" and "open mailbox," and the rest happened in your head. Kaplan still ranks it among the games few others will ever touch: it is the power of imagination, and the reason the book is always better than the movie. Ultima, with Richard Garriott appearing as Lord British, taught him that a world could be rich enough to rob merchants and try, endlessly, to kill its creator. Online, Doom, Duke Nukem, and especially Quake made another person in the game feel magical, even on a 300-ping dial-up connection among the "low ping bastards."

He learned about EverQuest on Blue's News, the one gaming site he still reads, through the old `.plan` files. An id programmer named Brian Hook posted that he was leaving to work at Verant on a game called EverQuest. Kaplan could not understand anyone leaving id, "the greatest institution in all of gaming," unless the new thing was worth seeing. Without that post, he says, he never would have checked it out. The breakthroughs at id — Wolfenstein 3D's first-person intimacy, Doom's smoothness, Quake's refinement — are, in his view, why he later got to work on a shooter at all.

Before any of that became a career, Kaplan wanted to be a writer. He holds a master's in creative writing from NYU, and he names Kerouac, Hemingway, Salinger, Bukowski, and Orwell as the line he was trying to join. What drew him was not only being transported, but the way stories trigger feelings you did not know you had. Putting work in front of strangers, he says, takes ego and a streak of masochism. People who build things cannot really help it; they are wired that way.

He gave it a real try. After leaving a New York writing circle to move back to California for a relationship that ended within two months, he structured his life as an eight-hour writing job, rain or shine. He took East Coast recruiting research for his father's firm so he could start at 5 a.m., then write the rest of the day with a Jack Russell terrier named Jack, after Kerouac. In the mid-to-late 1990s, submissions meant printed manuscripts and self-addressed stamped envelopes. In one year he received more than 170 rejection letters. A handwritten "showing promise" from a magazine like Glimmer Train could carry him for a week.

Isolation, heavy drinking, and depression compounded until he decided the work would destroy him. There was no single letter that broke him, no editor who can claim the honor. A group of literary-magazine editors across the United States, he says, was telling him he was not cut out for it, and he stopped. He threw away some twenty volumes of handwritten notes, daily journals, and every manuscript, digital or typed, into a dumpster. It was like a boxer throwing in the towel: if any of it sat in a drawer, he would be tempted to try again in ten years.

The advice he extracts is aimed at adults as much as at the young. Questions like "what's your major" and "what do you want to be" are, in his view, a terrible thing to do to a human being. Society sells the costume of astronaut, firefighter, writer, game maker, and people get lost performing a fake version of the role. "Focus on what you wanna do, not what you wanna be." When you are off the clock and nobody is asking, how do you spend your time? What actually makes you happy? That, he says, is how you find out what you will become.

The empty eight-hour blocks filled with EverQuest. The game launched in 1999. A friend named Victor, one of the few people Kaplan knew who played computer games despite the stigma, had bought it. Kaplan asked if it was the game Brian Hook had left id to make, and whether it was good. Victor said he had to play. The moment he logged into Norrath as a halfling rogue, he was gone: kill things, get experience, need better loot, kill more things. He lived for the next login. When he finally quit three years later, `/played` reported about 272 days — more than 6,000 hours, most of them not healthy. It was escapism used, he says, as a way to abuse real life.

Raiding meant thirty or more people trying to kill something they could not beat alone, which in EverQuest meant joining an "Uber Guild." He leveled mostly unguilded or in a rogue role-play guild, then at the level cap of 50 was invited into Legacy of Steel, the top guild on The Nameless server. Coordinating thirty wills — some loot-motivated, some lonely, some depressed — was, he says, like herding cats. You repeated "get to North Ro" for hours and hoped 20 percent showed up. When the founder, Dread, simply vanished, the way people quit in those days, a blonde wood-elf warrior named Ariel who refused to wear a helmet made Kaplan an officer and then, as she got busier, the raid leader. It was his first education in motivation, discipline, and when to push versus when to encourage.

He met his wife in EverQuest. Almost nobody shared real-life information; you assumed everyone was male, and it was a surprise when they were not. People told him he had thrown years away on a game you cannot win. Sitting with Fridman, he is unmoved: his whole career and his family are thanks to EverQuest, so he thinks he won. Quitting for good only became easy once he was psychologically done, after an expansion that did not speak to him, and once Blizzard became the new obsession. Until then, the `/played` on television would have blown 6,000 hours out of the water. The stigma, he and Fridman agree, never attached itself to the couch.

Kaplan's EverQuest name was Tigole, and he had a reputation for edgy, specific rants at the developers — including a famous instruction to pull their heads out of their asses. Fridman notes that the posts were loved because they were not empty trash talk; they were the notes of someone who lived in the game. After Ariel stepped down, Kaplan became guild leader. Around then he was building Duke Nukem and Half-Life levels, because those studios shipped the editor on the CD. Ariel asked to see the Half-Life maps, then gave an Irvine mailing address. Sending a level over the internet would have taken about twelve hours, so he burned a CD. The reply said: my name is Rob, I am a designer at Blizzard.

Ariel was Rob Pardo, then lead designer on Warcraft III. The blonde wood elf was a man. Lunch in Irvine included Scott Mercer, the guild enchanter, and Roman Kenny, a "psychotic wizard." Kaplan calls it, with an apology for the phrase, his coming-out moment. Until then he would never tell family or friends that he loved games, that they had just killed a dragon. Sitting there talking raid tactics and talking trash on the guild, he felt like himself for the first time. He thought he had simply made friends who happened to work at a company whose games he had never bought. He was a Half-Life and Quake III person; StarCraft, Diablo, and Warcraft had not been in the budget.

At a later lunch they brought a troll warrior named Barfa. Barfa was an inside invite; Kaplan had once saved him in a new dungeon called The Hole, when the whole guild escaped except the oversized troll who could not jump out. Kaplan, a rogue who could stealth, handed over an expensive teleport potion. Rob introduced him: this is Allen Adham, he plays Barfa. Adham was a Blizzard founder, alongside Mike Morhaime and Frank Pearce. Only in hindsight did Kaplan realize the lunches had been a six-month interview, handled so decently that they really were friends. Pardo told him to check the job site. World of Warcraft had been announced. The posting was associate quest designer, and it asked for a creative writing degree.

Blizzard almost never hired designers off the street; they came from QA, tech support, or other disciplines. The grilling lasted six months. He met Chris Metzen, who stopped to buy cigarettes on the way to lunch and swore through the interview, which felt like being around his people after years of corporate recruiting norms. The last interview was with Adham, Pardo, and programmer Bob Fitch, at a Jack in the Box inside an ARCO station. Kaplan thought: these guys just brought me to a gas-station burger place; I need to work here. Underneath the origin story was a worse stretch he had not told anyone: alcohol into his early thirties, therapy, antidepressants he hated, and finally electroconvulsive therapy that, with the job, pulled him out. On April 18, 2002, he posted "one of us" to his community. He still finds the post embarrassing in its ego. His boss was Allen, who knew what they were getting.

Kaplan's first day in May 2002 was at Blizzard's offices in the UC Irvine research park. Next door looked like Cisco; outside Blizzard it was black T-shirts, shorts, frisbees, hacky sack, scooters. Inside felt like a dorm, posters on the walls, futons for people who slept there. The company was fewer than 200 people, including Blizzard North in San Mateo. Team One, the RTS veterans of StarCraft and Warcraft, was working on Warcraft III and was revered. Team Two, assigned to World of Warcraft after a scrapped project called Nomad, was the red-headed stepchild. His login came from John Cash, the id programmer who had a Quake III skin. Cash was a huge EverQuest player. They fanboyed each other on sight: you run Legacy of Steel; you are John Cash.

He walks Fridman through a game team: engineering, art, design, production, and the often-ignored audio group. Two things, he says, nobody notices until they are missing are audio and lighting. On a small team in incubation, ten people sit in the room for every decision. He is not a networking engineer or an illustrator, but he is there. Growth brings compartments, and with distance comes a human habit of vilifying the other discipline. "The artists just don't get it" is, on inspection, an asshole sentence: untrue, demeaning, and forgetful that the game is their life's work too. When the Overwatch team formed later from the wreckage of Titan, the premise he tried to install was simple. Blizzard could recruit whoever it wanted. The person hired to make crates is the best prop artist in the industry. When that person says do X instead of Y, take a breath and ask why you would not listen.

Pardo taught him the same lesson the hard way. Newly made a lead on WoW, Kaplan sat in a meeting with non-designers, systematically explained why their ideas were wrong, and laid out the plan instead. He was not cruel, he says, just an insecure, ego-driven new lead who needed to be right. Afterward Pardo pulled him aside: you are a very smart designer, but you should always listen and try to make their ideas work. He started forcing himself to stay with the suggestion he wanted to reject. Some of the best ideas in WoW and Overwatch, he says, came from developing someone else's thought until it worked — and then they got the credit.

The hours were another kind of education. Gold-mastering Warcraft III, a game he had not worked on, he sat through the same cinematic for 30 hours so a programmer could catch a rare crash. Family people trickled out; the next morning they returned in fresh clothes while he was still in the chair with beet-red eyes. Mark Kern told the WoW team they would crunch early so they would not crunch later. Kaplan believes Kern meant it. Dates slipped anyway, and early crunch became a death march. Say "Newport Rib" to that team today, he says, and they still twitch. He came in on Memorial Day with beach sand on his feet, worked Christmas, and never blamed anyone but himself. Forced, unnecessary overtime from executives who work forty-hour weeks is immoral. His own ten-hour days and weekend work are, he insists, who he is — not a demand he places on others.

"The lead character of World of Warcraft is the world," Chris Metzen said, and Kaplan believed it. You were trying to build a place that was exciting and dangerous, comfortable and uncomfortable, gorgeous and massive — half an hour from one end to the other — that people would live in. Allen Adham championed the Horde and Alliance split, a controversial call. In EverQuest, races could group together; Pardo and Kaplan thought forced factions were a terrible idea and fought Adham, Pardo, Bob Fitch, and Kaplan over lunch every day. Adham came from Dark Age of Camelot, where three factions meant you were instantly on a team. Before beta, Adham retired to a hedge fund and poker, nine months to a year before ship. Pardo took over as lead designer and, to his credit, reversed himself: Allen is smart, he fought this hard, we have to do it. Players still walk up to Kaplan with Horde tattoos. Invite him to play Alliance on Tichondrius, and the friendship is in trouble.

The early WoW team was what he calls the Bad News Bears: veterans like Mark Kern, Shane Dabiri, art director Bill Petras, and Metzen as the cool older brother, mixed with people who had never made an MMO. Dungeon builders were hired out of the Quake community and blocked spaces in QERadiant. Kaplan describes himself as some idiot who had played a lot of EverQuest and was now going to design World of Warcraft. In hindsight, not knowing what they were doing was a huge part of early success. Titan later assembled experts who knew exactly how to make a groundbreaking MMO, and failed. At the time the threat was EverQuest II and Star Wars Galaxies, with Raph Koster attached. Kaplan remembers thinking they were fucked.

EverQuest had quests, but they were not in front of the player; you looked them up on a website. Adham wanted quests at the center, so he hired Kaplan and, on the same day, Pat Nagle, whose previous title had been HR and facilities — applications and overflowing toilets. They shared an office and designed the system, with Eric Dodds on the interface. On a whiteboard they guessed how many quests EverQuest had after several expansions, and decided matching that number would be enough. Nagle took Elwynn Forest, Kaplan took Westfall, they met Metzen for story, and they playtested. Shooter and StarCraft people on the team ran out of quests in an hour or two and were furious. The assumption had been that quests would point you at creature camps, then you would grind for hours. Players expected quests the whole way.

The philosophy they adopted, one Kaplan has since heard across the industry, was to design along the path of least resistance. If the fastest way to max level is standing still killing easy spawns, that is what people will do. Load the experience into the quests, and the easiest path moves you through the world, shows you everything, and carries the story. EverQuest reached hundreds of thousands of players; WoW reached tens of millions. The difference, he argues, is that WoW let you play it like a single-player game. Other people make the world feel alive, but forcing interaction to do anything is off-putting to a huge audience. Directed gameplay that *felt* optional, and really was not, is the quiet revolution.

Kaplan calls himself a quack psychologist whose job is to motivate people into having fun. Players shift between extrinsic and intrinsic motives. Overwatch loot boxes, he notes, were designed by a game designer, not a commercial person, and the team obsessed over the sound, the spill of items, the animation — lizard brain and spreadsheet brain at once. He lists unofficial elements of fun: a sense of progression that the game recognizes your investment; raw mastery; creativity; customization, whether an earring or the first whirlwind barbarian build. He is not loot-driven; he wants to see what the world offers. That often forces a loot detour, or a creativity detour when a boss's curse demands a counter. First-person shooters add intimacy and a pure skill test. Nothing, for him, is as pure as looking at your own hands.

He bristles at being called the anti-story guy in games. He loves story; he is the anti-shitty-story guy. The most magical stories he knows from games are player stories — the time he gave Barfa a potion and later met him in real life. Ask why someone loves the cow level, he says, and you will hear a loot goblin, another player, and something goofy. Naughty Dog, Valve's Marc Laidlaw and Erik Wolpaw, and Rockstar sit on another level. Red Dead Redemption 2 is one of his favorite games: it holds a heartbreaking Arthur Morgan arc and still leaves room to accidentally blow a stranger's head off with a sawed-off. When Daniel Lanois's "That's the Way It Is" arrives after an instrumental Woody Jackson score, with Dutch and Arthur's lines coming back, Kaplan felt the top of the pyramid. He knows it is entertainment and a business. That moment was art.

The Green Hills of Stranglethorn quest is his public mistake. Players hated it. He still loves it. He wrote a short story into the game as homage to Hemingway; the quest giver is Hemet Nesingwary, the letters of Hemingway rearranged, with a Kerouac name mixed nearby. Over winter break, alone in the office and happy, he split the story into pages scattered across the zone, lootable from any creature — a McDonald's Monopoly of literature. Sid Meier's line sits over the whole confession: there are three types of fun, fun for the player, fun for the designer, and fun for the computer. This was ant-farm design, the designer as God watching the ants. Pages did not stack, there was no dedicated container, bags were tiny, and Stranglethorn chat filled with people trading page sixes. Most players quit the quest or bought the pages from the auction house. Because he said out loud that he had made one of the shittiest quests in the game, it became safer at Blizzard to criticize their own work.

Blizzard polish, Kaplan says, is studio-wide culture: nobody satisfied with a bug, and a QA department of passionate players, many of whom want to become developers, sitting as close to the team as space allows. On Overwatch they had specialists who could feel a frame of input delay from a hundred meters. WoW built a full raid team in QA and asked them not only whether the dragon sank through the world, but whether they were having fun. The other half of live quality is engineering that makes the game hotfixable — a server change with no client patch — so a crashing service or a hero that must be disabled can come back in half an hour. Players feel whether someone loved the product enough to clean the yard, not only whether there is a cow level.

His offer letter was $35,000 a year. Shortly after ship, with Adham already gone and Pardo leaving for StarCraft II, he and Tom Chilton — a designer from Ultima Online — were running WoW design. The world was on fire, servers barely holding, and his title was still senior game designer. He thought all of it was normal because it was his first game. Game masters called his landline at 3 a.m. about an exploit in Stranglethorn; a GM spawned giant infernal "Guardians of Blizzard" across the zone while Kaplan whispered so his wife could sleep. Shipping the box was chapter one of twelve. They argued about the launcher until Morhaime insisted it was the best way to talk to players. Kaplan had wanted to hide it. Morhaime was right; it became Battle.net.

After launch, morale on Team Two was in the gutter. The game had done far better than expected, databases could not keep up, and two groups walked out to found Carbine and Red 5, taking Mark Kern, Bill Petras, animators, and programmers. Kaplan sat facing Morhaime's office and watched people quit. A game is never as good as the pitch; every day toward ship, the ideal gets farther from the thing you can actually finish. Then came the first BlizzCon in 2005. Forums were all lag and login failures. Tickets on the website were crickets until Morhaime put them on the launcher and they sold out instantly. Showing up at that hall, Kaplan says, was one of the most emotional moments of his life: nothing but love, people who only wanted to talk about the game. Online, passion performs as hatred because hatred gets attention.

He feels complicit in that culture. A few of his edgiest EverQuest posts survived thirty years; the chill ones did not. Social media, as he reads it, rewards maximum hyperbole, usually negative. Say a mug is pretty nice and nobody cares; call it a crime against humanity and the machine feeds you. Fridman watches the same reflex in students: the safe pose is that everything is stupid, and saying "this is awesome" is an act of courage. Kaplan's darker worry is social Darwinism among makers. Would van Gogh have existed on Reddit? How many people could speak to Beethoven in a week, versus insulting Brad Pitt from ten devices? Jay Wilson, whom he calls one of the great design minds and the game director of Diablo III, took so much heat that he left games for novels for ten or fifteen years. Millions loved that game. The community, Kaplan says, essentially removed him. It feels criminal.

Around 2005 and 2006, with WoW succeeding, the studio assumed the game might last five years and that Blizzard would be in trouble without another MMO waiting. Real development on Titan started around 2007, including a new engine. The first programmer Kaplan names is John LaFleur; that engine failed for Titan and later became Overwatch's. The fiction was future Earth: secret agents with over-the-top shooter abilities by night, and by day a life borrowed from Animal Crossing, Harvest Moon, and The Sims, with Matt Brown, creative director on The Sims, coming over. Unlike sharded WoW, everyone would share one server and one world: Bay City as San Francisco, Hollywood, the rest of California, Cairo, London, and GTA-style driving between them. They wrestled over whether the IP had aliens. Kaplan left the WoW team the day Wrath of the Lich King shipped — Election Day, his birthday, the night Obama won — after the champagne toast, and joined Titan.

They shut it down in 2013. By 2009 he knew the current form could never ship; in 2010 he told Morhaime to shut them down because they were going to burn money. Fridman notes the project cost $83 million. Failure, Kaplan says, lies with leadership, himself included, across art, engineering, and design. He is careful about art: some of the best Blizzard art ever was made for Titan. The problem was no cohesion. It looked like ten different games. He distinguishes ideas from vision. Ideas are cheap — secret agent by day, flower shop by night. Vision is shepherding an idea into existence with inspiration, a tech plan, a design plan, an art style, and a production reality. Titan was Blizzard hubris at its height: we made WoW, we can do no wrong.

His hiring rule is the opposite of what Titan did. He will not hire until the team is already working overtime or might not ship. Titan practiced anticipatory hiring: the world would be big, so hire environment artists, even though the story, layout, art style, and technical constraints were unknown. In one year they hired about seventy environment artists from around the world, visas and all. A new hire from Belgium asked on day one if they were making a StarCraft MMO, then, after the tour, "What is this game?" The right incubation, Kaplan says, is the smallest group proving the idea with whatever engine and leftover art you can grab, then expanding. Idle brilliance becomes busywork: Chinatown props or the Hollywood set? The editor, TED, was so broken that technical artist Dylan Jones logged twenty usable hours in a forty-hour week. Fighter pilots, he says, who are not allowed to fly. The demoralization was a disaster.

After cancellation, most of Titan's 140-odd people went to Heroes of the Storm, a Diablo III expansion, WoW, and Hearthstone. Morhaime gave a small creative group six weeks and two criteria: ship in two years, and match WoW-like revenue. Kaplan threw out the revenue line as Monopoly money. He took the schedule seriously. They spent two weeks on each pitch, all in, then threw it away. First was StarCraft Frontiers, inspired by an old Metzen drawing of a space prospector — Indiana Jones in space, not a marine in an army. Arnold Tsang and Peter Lee painted the cigar, the hydralisk skull, the medivac. That image was later redone as McCree. Even focused, it was five years and 150 to 200 people on Blizzard's best day. Forty people were not making it in two years. Space flight alone, he says, adds three years. It was an idea, not a vision, because it lacked a path to reality.

Second was CrossWorlds, Metzen's Mos Eisley planet at the edge of the universe, a neutral den of smugglers and diplomats. You would not have to design every homeworld, and a later expansion could simply visit the planet of the green people. In a class meeting, Geoff Goodman — original WoW encounter designer, Onyxia and the raid bosses people still name — said offhand that he wished they could make fifty classes with one or two interesting things each, instead of six classes with a hundred abilities. The meeting ended. Kaplan sat down, opened the Titan art directory, and distilled the Jumper, a dual-pistol teleporter based on his Modern Warfare 2 G18s, into blink, recall, and time bomb. He asked Tsang who she was as a person, not a class. Tsang said: what if she is British, and her name is Tracer? Gunjack became Reaper. Ranger split into Soldier: 76 and Bastion. The pragmatic question was what Goodman and Tsang could actually ship in two years.

While the team wrote CrossWorlds docs, Kaplan built a seven-page deck titled Monetized Shooter, League of Legends plus Team Fortress 2, six sloppy heroes. Ray Gresko looked over his shoulder and told him to show Metzen. Metzen said yes. Producer Matt Hawley stopped him in the hall: you cannot walk in with a first slide that says Monetized Shooter. Name it. Kaplan said Overwatch on the spot. During Titan, a fake democratic naming vote had already been steered toward a name the organizer wanted; Overwatch, a police group in that fiction, had gotten the most votes anyway. He picked it up as a high-five and a middle finger: do not pretend it is a democracy. Jeremy Craig turned the infamous gray "Jeff deck" into a real pitch. They sold crawl, walk, run: Titan failed because it tried to run, to be the next WoW, without Warcraft I, II, and III underneath. The first game only had to make people care about a universe. Dates on the slide were decorative self-defense. Mobile strategy was Arnold art Photoshopped onto a tablet.

The world was a bright, hopeful near future, "a future worth fighting for." Locations were a tour of places you would actually want to stand in for hundreds of hours, not cargo-container mazes. Oasis asked what a hopeful Iraq could look like. Tracer is Kaplan's OG, simple until the first recall. McCree is the six-shooter feel he did not design, plus Mike Heiberg's High Noon and a tumbleweed on the ultimate. Reinhardt, his main, is Goodman's shield you can shoot through, a rocket hammer, and a charge stolen in spirit from Left 4 Dead 2's Charger: a commit, a runaway train. Matchmaking, built to hold about a 50 percent win rate, is thankless. Players say they want fair; they want close and then a win. Memory discards the eight-game streak and posts the six-game loss. If he made a hero shooter from scratch today, he would downplay the team and emphasize individual contribution. People are selfish in how they feel joy, even in a six-stack. Asked for great shooters beyond Overwatch, he names Quake as GOAT, then Call of Duty's gunfeel, then Rust — "the most PvP thing in all of PvP," monthly wipes, bases you can explode, highs that require real lows. He has about 5,000 hours in it. He would never recommend it. It will come calling if you are up for it.

They started talking about Overwatch 2 in 2015, before Overwatch 1 shipped in 2016: Kaplan, Metzen, and Michael Chu sketching a cooperative PvE campaign, pitching it to put runway in front of the team. He calls it one of his biggest mistakes as a creative leader. Some developers loved the universe but not competitive PvP; once the shooter shipped, they asked when the other game would start. Executives, especially at Activision, pointed at the slide that said 2019. "You never want to put a PowerPoint deck in front of a corporate executive. You might as well etch it in stone." Meanwhile the live game caught a wave. For the Rio Olympics they could not say Olympics, so Mike Heiberg and Dave Adams built Lúcioball and the Summer Games; Halloween followed. Fans loved the events. A coalition on the team called them over-scoped distractions from Overwatch 2. Kaplan's instinct was the opposite: most games never catch the wave. If you catch it, ride it until the end.

The major derail was Overwatch League. He believed in it and helped pitch regional teams, minimum salaries, player protections. Then the road show sold something like the Brooklyn Bridge: more popular than the NFL, billionaire owners. The day he got back in 2018, a huge Twitch media-rights deal meant in-game camera control, team skins, uniforms, and ugly technical constraints. Plans for world events and for Overwatch 2 went out the window; the team was treading water. When the deal disappointed, the answer was not to give money back but to demand make-goods from the developers. Owners who had expected ticket sales and merch discovered geography, then asked what a live game that had made something like $500 million could sell them instead. It was, he says, a house of cards. The Overwatch 2 that later shipped is not the PvE game they planned.

What broke him was a meeting in the CFO's office. He was given a date, 2020, already likely to slip to 2021, and a recurring revenue number Overwatch had to hit every year after. If it did not, they would lay off a thousand people, and that would be on him. It felt like the biggest "fuck you" of his career, complete with the fantasy that hiring Fortnite's 1,400 people and going free-to-play would print the money. He had believed he would never work anywhere but Blizzard, that he would retire there. "I never thought the day would come, and that was it." The CFO, he notes, is no longer there. Saying goodbye "broke me." He did not realize how broken until the mourning caught up: how could he give that up? How could he not be there?

The sermon he wants other makers to hear is that craft-obsessed people are not cutthroat, and a whole ambitious world shows up the moment they succeed. They keep giving themselves away. When they made WoW, Blizzard had no CFO. You need artists, engineers, designers, producers, and audio. There are finance people who get it and get out of the way. He wishes developers understood their value and stopped handing the golden goose to people who do not deserve it. What made the company legendary, looking back, was three gamer-programmers — Mike, Allen, and Frank — who had made the games before they ran the place, plus a COO, Paul Sams, who protected the floor. He joined a company that was about 95 percent developers and 5 percent operations. He left a 4,500-person company that was 50/50. Working with Metzen was translation into gameplay: an hour of maps, old Warcraft manuals, and Sam Didier drawings, then zone order and dungeons. When Metzen demanded the tallest tower in Azeroth for Silvermoon, level designers started measuring Blackrock, Karazhan, and Stormwind Cathedral. Kaplan told them Chris did not know those heights. He meant make it really tall. He would take the heat. It was a vibe.

Instead of a corporate goodbye, Kaplan played the team a David Bowie clip: never play to the gallery; if you feel safe, you are not in the right water. His wife told him to take at least a year. He had gone nineteen years barely vacationing and had let Blizzard consume him. He pulled weeds — she would not let him garden — and ground out Dark Matter Ultra in Call of Duty: Black Ops Cold War. Then, late spring, he sat in the backyard with Notepad and started writing the game he wanted to make. After nineteen years next to people he thought were the best in the industry, being alone was terrifying and, it turned out, joyful. He had forgotten the craft. He watched YouTube to learn Unreal, Illustrator, Blender. Two conclusions: he never wants to work for someone else again, and he is not interested in the games industry, the business, or entertainment. Just game-jamming things they will play together.

Tim Ford, a programmer Kaplan calls his development soulmate and then associate tech director on Overwatch, quit in the same season. Kaplan suggested they take a stab. Ford's last day was Friday: exit interview at 1 p.m., at Kaplan's house at 2. Kaplan told him to take a month and pull weeds. Ford said he is a programmer; a month off is a month of programming, so they might as well start their game. From outside, Kaplan had disappeared. About sixty people reached out in April 2021, VC and strategic money, especially from Chinese firms. He and Ford calculated the headcount, the time, the budget, raised that number, and stopped. They optimized for control, not for the auction. The company is Kintsugiyama: kintsugi, the Japanese repair of broken pottery with gold lacquer so the scars become beautiful, plus yama, mountain. They are grateful for Blizzard and did not leave unscarred. Nothing is perfect; the pursuit of perfection is a mistake. Games, he says, are a constant pursuit of imperfection. Ask the players.

The game is The Legend of California, an open-world action game on a mythical island of California in the Gold Rush — not historically accurate, uninhabited at discovery, deliberately inauthentic except in feeling: prospectors, cowboys, mines you can actually dig because the terrain is voxels. California's silhouette is hand-authored, Yosemite where you expect it, but dozens of points of interest shuffle with the map seed, and four difficulty tiers reshuffle too. Mojave might be the newbie zone on your server and endgame on his. Dread Rock, inspired by Alcatraz, sometimes sits in San Francisco and sometimes in the desert. Lighting artist Mike Marra is chasing Albert Bierstadt landscapes Kaplan found while researching the state: he wanted to walk around the painting and see around the corner. Metzen called Blizzard a hero factory. This tone is lonelier, more mysterious, larger than you. You feel small until you earn the right to feel big. If the sun is setting, get to shelter. They will put it on Steam quietly, run a public-ish alpha, then early access, rough edges and all. The team is 34 people.

Asked for the greatest game ever made, he does not hesitate: The Legend of Zelda: Breath of the Wild, every system a toy, art matching design matching Switch, a better Zelda after Ocarina of Time already existed. His son, who also makes games, said he was sad because he knew he would never make anything like it. Red Dead Redemption 2 sits on a shrine as craft: only Rockstar, after years, can do that, and almost no games actually have great story even when they have writing. EverQuest and Rust defined his life; he would tell you to play neither. From Rust he is taking the reset, probably monthly, and the trick is to make players excited that the world starts over, two people naked on the beach, instead of angry. He is not taking the part where everything can be stolen. On AI, integrating it into development is "mostly a hot mess," overconfident, right about one time in ten on a simple Unreal UMG question. Batch-resizing 2,000 mis-sized images is fair use of a tool he would not hire an intern for. Training on artists' and voice actors' work without permission is stealing. Arnold Tsang's drawing and Metzen's story are irreplaceable. Small studios, he says, are the future. The advice is the same one he gave about the goose: own the craft. Stop giving it to corporate jackals. You are the golden goose. Keep your eggs.