David L Craddock has almost completed the second part of his trilogy documentaing the formation of Blizzard Entertainment and the studio's biggest games. Stay Awhile and Listen: Book II covers the creation of StarCraft and Diablo II, detailing the crunch and challenge of starting a new series in a saturated genre, and building a sequel worthy of the original game's fame.
Craddock has allowed us to publish a chapter of his book here on PCGamesN. Chapter 7: Hubris or Fear focusses on the 14 months of crunch the StarCraft team put themselves through in creating their classic RTS game. Members of the StarCraft team tell him how the work brought them closer together, while putting significant strain on their health and their relationships outside of work.
StarCraft isn't on our list of the best strategy games on PC, but its sequel certainly is.
Craddock has almost completed the book and is funding the final stages through a Kickstarter campaign. To read more about Stay Awhile and Listen: Book II and support the project head over the book's Kickstarter Page.
This chapter was prepared for us to publish but may change between now and publication. Also, we've emphasised the next with pull quotes and added in imagery.
Chapter 7: Hubris or Fear
'The way we were always able to get so many cinematics done, and the [quality] of how they looked was due to two things. One: Games were always delayed, which gave us more time. Two: We worked ourselves to death.'
Harley Huggins, artist, Blizzard Entertainment
'It's easy to hear these kinds of stories and think there was a lot of ego at Blizzard. I really don't think there was. There wasn't ego at Blizzard. There was just extreme passion with a lack of social graces.'
Gage Galinger, programmer, Blizzard Entertainment
'When Diablo shipped, everybody went back to working on StarCraft. Everybody` said, "All right, we can ship this thing, we're only three months away." Then, "Okay, now we're only two months away." StarCraft was "only two months away" for fourteen months.'
Pat Wyatt, programmer/VP of R&D, Blizzard Entertainment
Despite the best efforts of Blizzard Entertainment's managers to incorporate structures and protocols such as strike teams and producers, those initiatives had been grafted onto the company. They were wet cement poured over a foundation paved from passion and perfectionism. Those qualities had guided StarCraft's developers to jettison the game's derided "Orcs in Space" design like so much space wreckage.
Some leftovers from the game's original direction still had worth. Some basic components of WarCraft II's engine made the jump, but Bob Fitch, its chief architect, expanded on it. Over two months in early 1997, he wrote code allowing for advanced capabilities unheard of in WarCraft II, such as cloaked units, the ability to burrow underground, and units able to be nested within other units such as the Protoss Carrier's fleet of Interceptors.
Fitch's retooling brought about sweeping technical and creative changes. In WarCraft II, terrain had been made up of square-shaped tiles individually painted by artists and stitched together by programmers. Diablo had used diamond-shaped tiles, a decision both stylistic and functional, that Fitch applied to StarCraft. Terrain made of diamond-shaped tiles could be viewed at an isometric angle, and artists could show three sides to units instead of two. Functional benefits included the ability for units to take cover behind foliage to mitigate damage, and inflict more damage by attacking from high ground.
Meanwhile, designers and artists revised the game's mechanics and audiovisuals. "After we had a CES screening of a version of StarCraft that looked like WarCraft—very cartoony, primary colors—Allen came to me and said, 'Hey, can you make this look like that WarCraft II test that you did?'" recalled Duane Stinnett, co-lead artist alongside Sam Didier in addition to heading up the burgeoning cinematics team alongside Matt Samia. "So, StarCraft's look became what WarCraft II's look was supposed to be: More realistic and more gritty."
Changes ranged from superficial to subtle. Decreeing "Nightmarish Invaders" too generic a name, the designers rebranded StarCraft's insectoid aliens the Zurg, until they realized "Zurg" was also the name of an evil alien race in Disney's and Pixar's upcoming Toy Storm animated movie. Attached to the name, they modified it slightly to Zerg. The Phoenix, a Terran airship that fired missiles against airborne units and lasers at forces on the ground, was renamed the Wraith. Goliaths, a bipedal mech also on the Terran side, had been equipped with a flamethrower, machine gun, and missiles. Designers James Phinney and Chris Metzen dialed back its firepower, gave its flamethrower to the Terran Firebat, and tweaked the Goliath so it fired missiles at flying attackers. WarCraft II's units had inflicted a range of damage, randomly chosen by rolls of invisible dice. StarCraft's units always dealt a prescribed amount that players could increase by researching upgrades or exploiting tactical advantages such as attacking from higher ground. Fixed damage values were easier for players to understand, and easier for programmers to balance.
"StarCraft's look became what WarCraft II's look was
supposed to be: More realistic and more gritty."
One of the driving goals behind taking StarCraft back to the drawing board was creating a real-time strategy title whose three playable races were as radically different from one another as they were superbly balanced—the ultimate game of rock-paper-scissors, but one where scissors could cut through rock if they were sharp enough. The bug-like Zerg no longer constructed buildings, like the Orcs and Humans from WarCraft. Instead their worker units, Drones, transformed into buildings, pulsing like beating hearts as they evolved into their new shape. Damaged Zerg buildings spurted blood, reflecting their biological nature. Furthermore, Zerg structures could only be erected on creep, a slimy purple carpet that spread outward from buildings. Each structure expanded the creep, giving players more real estate on which to build.
In a holdover from WarCraft, Terran players raised buildings by assigning a worker unit, the SCV, to the task. SCVs could be ordered to halt construction in a pinch and sent to perform other functions. Once those were finished, any SCV could be reassigned to complete the task. The Protoss straddled ground between the biological Zerg and flexible Terran, with Probe workers summoning structures by warping them from their home planet. Warp portals shimmered with blue light until the summoned building materialized, but Probes could be put sent to harvest resources or beam in other structures immediately after initiating a warp. All three races had differing methods of repairing damaged structures: One or several Terran SCVs could be assigned to a structure to repair it faster, the Zerg's organic buildings healed slowly over time, and Protoss shields recharged.
Besides passion and perfectionism, the bedrock of Blizzard Entertainment consisted of one other mineral: Pressure. The pressure for StarCraft to succeed—and by extension, the studio—came from two sources. The first was external. During an interview with Johnny Wilson, then editor-in-chief of defunct magazine Computer Gaming World, Pat Wyatt found himself in the hot seat. "He said, 'You know, in about the timeframe that StarCraft is coming out, there's eighty-plus other RTS games coming out. What makes StarCraft any better than any of the rest of them?'" Wyatt remembered the editor asking. "And it was hard to come up with an answer because it wasn't like we had some cool technological feature that was above and beyond WarCraft II."
Competition had grown fierce in the scant three years since Blizzard had improved upon Westwood Studios' Dune II with the original WarCraft. Westwood had rebounded by putting out Command & Conquer, a futuristic real-time strategy (RTS) title that resonated with players tired of fantasy settings, and a prequel, Command & Conquer: Red Alert, set players in an alternate-history-inspired take on the Allied Forces and Soviet Union. Cavedog Entertainment's Total Annihilation pushed RTS into the third dimension by featuring 3D units and terrain, as well as an ambitious plan to roll out a new unit every week after release. Dark Reign: The Future of War by Auran had two races and would add two more in an expansion pack. Then there was Dominion Storm, Ion Storm's graphically stunning entry that turned heads, including those of Blizzard developers, at trade shows.
Wilson's pointed question, and the reality of the answer, left Wyatt shaken. He put on a brave face. "There were lots of games, and the only thing we could really say was, 'You know, we've done this before, and the balance in the game is going to be great. All I can tell you is that we just know it's going to be excellent.' But that was about it. It was very easy to imagine someone else coming along and eating us for lunch if we didn't do a great job."
Blizzard's second source of pressure was homemade. "One of the definite pressures already at that point—I'm sure it's even worse now—is that, okay, StarCraft has to be really good, otherwise everyone's going to really point to it because we had high expectations and success from all these other titles," said Rob Huebner, a programmer on the game. "I think that was the main impact of having a string of hits. We can't cut corners here because if we do, everyone's going to jump on us. No one wants to be responsible for the first non-hit."
The 'Piranha Effect'
Rob Huebner slid his mouse cursor to the submit button on his screen. His finger lifted to click, hesitated. He scanned his code again. Then again. He wanted to be absolutely certain it worked as intended before he checked it in. It may cause a few logic bugs, but those wouldn't be difficult to iron out. The source of his reluctance was a programmer—name withheld—who worked on the first floor of Blizzard's main building. Taciturn and gruff, he reminded Rob of a fairytale troll, dwelling in his programmer's cave and only emerging to hunt down programmers feckless enough to submit code that broke StarCraft's latest build. The build was his bridge, and woe to anyone who compromised it.
"He was famous for being a threatening presence," Huebner recalled. "If you break the build, he will get you, like the boogeyman. He was a big street hockey player, too, so he was a big guy to begin with. Then his gruff demeanor made him a feared presence. Nobody wanted to break the build. A lot of companies now will have sandboxes and automated tests, but it really wasn't quite that [structured], so you could easily check in something that would take the whole team down for a day or two if you weren't careful."
On his first day of work, Gage Galinger settled into his workspace and pulled up a few sample maps to get a feel for StarCraft's look and flow. One map consisted of a series of space platforms set against a star field. Players would build up their bases on those platforms and traverse them using ramps and bridges to expand their army and hunt down opposition. Pat Wyatt came up behind Gage and pointed at the starfield. At present, Pat explained, the star field scrolled by at the same speed as foreground objects such as platforms when players moved the mouse cursor to any edge of the screen. The effect was unrealistic. Pat mandated that Gage implement parallax scrolling, a visual technique that caused backgrounds to slide by at slower rates than foreground objects when users scrolled the screen, such as the way a mountain far in the background crawls by while mailboxes on the side of the road pass in a blur.
Having given marching orders, Pat marched off to see to other tasks. Gage took a breath. He was a junior programmer, yet Blizzard Entertainment's vice president of research and development and a senior engineer besides expected him to sink or swim just like anyone else. There was a term for the intimidation caused by the infamous build engineer and other programmers at Blizzard, a manifestation of the internal pressure to thrive, and Gage had just experienced it firsthand.
"Pat called it the piranha effect," Gage said. "It basically was, if you did anything wrong you can expect the programming team to jump all over it. You were so careful with the code. The code was sacred. If you're going to check something into that code base, it better be fucking spotless."
"It's something that Blizzard shares with Microsoft," Pat added. "Microsoft has a really harsh culture where there are a lot of battles. At Blizzard, it wasn't really political battles, it was more like everybody was hyper-competitive playing Street Fighter, writing code, doing artwork—whatever it was."
"The code was sacred. If you're going to check something into
that code base, it better be fucking spotless."
The piranha effect was not exclusive to programmers. Anyone who believed in an idea had to make their case to implement it in front of the strike team. Artists who proposed absurd ideas or who turned in subpar work were heckled by peers. Developers likened the piranha effect to hazing at a college fraternity, though it never went so far as extreme or meanspirited pranks. "I don't really think I would characterize it as dog-eat-dog because it wasn't really that people were attacking each other," Pat clarified. "It was more [that people could be] incredibly harsh judges of what was going on, so that people who didn't deliver really high-quality results would just get kind of... it would be really painful."
"They mentored you at the same time," added David Pursley, an artist on StarCraft. "If you hadn't seen a movie that they thought you should see, they'd have movie nights for you to catch you up on Aliens, all those types of movies. It definitely made us a closer-knit group."
At times, Blizzard's core values seemed contradictory. Spontaneous design and iteration that led to perfectionism warred with initiatives such as the strike team, whose members decided on and disseminated everything from a game's creative direction and visual style to its technical underpinnings in order to maintain cohesion and scheduling. Developers were determined to make the best games, but hiring sprees to fill out the ranks often resulted in bringing on untested engineers who should have been mentored by experienced peers so that they could contribute without unintentionally bogging down the project.5 Gage Galinger juggled expectations by sticking to a routine. He slept in, got to work by noon, coded straight through until six a.m. the next morning. On the way home he swung through McDonald's for breakfast. After catching four or five hours of sleep, he dragged himself back to the office, perpetually battling exhaustion. Stubborn refusal to cave to expectations motivated him to work harder.
Senior developers such as Pat Wyatt, Bob Fitch, and Jeff Strain would have made excellent mentors, but they were too busy coding, the area where they were most needed. That led to yet another drawback: Team leads who failed to learn the fundamentals of good leadership because they were too busy coding, or painting, or rendering, or sculpting, or designing, or testing. "The staff got bigger, and there were more and more people on the core team, people that were influencers and business-makers, but it was the same style throughout—very 'go, go, go, hair on fire, we can figure out how to do everything if we just work harder,'" Pat admitted.
Time became meaningless. Ten-hour workdays stretched to twelve hours, fourteen, sixteen. Weeks bled into weekends. Another valiant effort to splinter Blizzard's growing staff into multiple projects was scoured when StarCraft's team devoured a smaller group like a stronger twin devouring the other in utero. "As tends to happen at Blizzard, whatever project gets ahead starts to pull people from the other team, and other teams die from attrition and things like that," said Rob Huebner. He had been hired for the enigmatically named Team 2, then working on a game codenamed Nomad, only to be consumed by StarCraft. "That's not a bad way to do things. It definitely hurts the project that gets resources stolen away from it, but it's for a good cause, obviously."
"My first day at Blizzard, they took me to lunch, and the first thing they asked was if I had a girlfriend. I said no," remembered David Pursley. "They said, 'Good, because you wouldn't for very long.' Everybody loses their girlfriends within the first week or two of working there because of the hours."
The practice of excessive workdays and nonexistent weekends is known within the game industry as crunch, or a death march. Some crunches are scheduled. StarCraft's was not. "What would happen was sort of a combination of ambition and delusion," Pat Wyatt said. "The team would go, 'We're two months away. Well, since it's going to be two months now, let's get in this one new feature.' Then that would snowball and become a couple of features. The scope of the project kept getting larger and larger."
"Everybody loses their girlfriends within the first week
or two of working there because of the hours."
"I remember when I was interviewing with Duane Stinnett, and I said, 'So when is this game going to be out?'" recalled Harley Huggins, an artist hired explicitly for Blizzard's then-nascent cinematics team. "He said, 'Oh, like, three months.' That was April of '97, and that was the mantra. Anytime someone asked when it will be done, someone would say, 'It will be done in three months.' No one really knew."
Crunching is often self-defeating. Bob Fitch, the game's programming lead, felt obligated to arrive first and leave last. After a while he rarely went home, pulling two- and three-day coding marathons to get as much work done as possible. Near the end of one such sprint, Pat Wyatt pulling up code that Bob Fitch had checked in and finding a twisted mass of bugs and flawed logic. Seemingly, his friend's overworked brain had mangled it. Pat, equally drained, shooed Bob out the door to get some sleep and directed his attention to weeding out his friend's errors.
Blizzard's culture of excellence led to games of chicken. Nobody wanted to be the first to blink and go home. On the other hand, StarCraft's crunch often led to breakthroughs that may have otherwise been unattainable. "Cinematics crunched nonstop, and it was self-imposed, because we wanted to do more stuff," said Harley. "We wanted to perfect our cinematics. That's why we ended up separating stuff into layers later on. You'd render the whole [cinematic] and go, 'There's something that's flashing' or 'Maybe you could go in and change the frames.' Instead of just saying it was good enough, we'd go in and redo the whole thing. Sometimes that meant not going home. Everybody just did it because they wanted to do it."
"It was the same as the LucasArts Jedi Knight team," Rob Huebner agreed. "It was a self-demand more than an external demand. Maybe at a higher level it was from management and they stealthily made it seem like it came from grassroots, but if so, they succeeded at that. There were still long hours, but to me, a sweatshop is like a producer coming in and mandating these hours, or you hear horror stories about L.A. Noire developer Team Bondi. That definitely wasn't at all like what Blizzard was."
Pat Wyatt remembered battling over code, even though giving in to the other would have meant one less headache for both of them. "I wanted units to be smarter like they were in WarCraft, and he wanted them to be dumber because he wanted the player to be making the decisions, which provided more tactical advantage to players who were better at manipulating units," said Wyatt. Pat argued that giving units more autonomy led to more efficient A.I. such as that in WarCraft II, whose units automatically directed attacks toward an opponent's physically strongest units who had the least amount of life. Back and forth they went, each creeping into the code database and undoing the other's preference, creating more work on both sides.
In another instance, StarCraft's path-finding—a unit's ability to navigate from point A to point B—broke down when worker units got bunched up at clusters of minerals. Workers were programmed to follow a simple cycle: harvest crystals or gas, walk back to the player's headquarters, deposit its load, and return to harvest more. The problem was that workers tended to get bunched up between resource patches and headquarters, causing them to weave and shimmy around one another and delay getting back to the building. That, in turn, slowed the rate at which players accumulated the crystals and gas needed to construct buildings, train combat troops, and research upgrades. Stressed out and short on time, Pat hacked out a solution by instructing workers to ignore collision with any other units in their path.
Jeff Strain was in charge of StarEdit, a free map editor that would ship on every StarCraft disc so that every player could create and share maps over Battle.net. Behind schedule, Jeff carried his laptop everywhere, a decision that didn't go over well with his wife. "I was writing code while she was in labor," he said. "Here it is, twenty years later, and I'm still paying for that. That's her trump card, right? Maybe in another ten years I'll have made it up to her."
Many of StarCraft's most urgent problems boiled down to minor details. Technical artists such as Jeffrey Vaughn were saddled with delicate tasks such as making sure raw artwork gelled with how a gameplay feature or A.I. implementation was supposed to work in code, and how that action manifested on-screen. "If you have fifty weapons in your game, you need to set up hard points for each weapon so your characters know where to hold them," Vaughn said. "If you have twenty buildings that can be set on fire, you have to tell the buildings where to catch on fire. Glue screens—the front-end UI that 'glues' the game structure together—need to be mocked up before they're handed off to an artist to make pretty, and so on."
"I was writing code while she was in labor. Here it is, twenty
years later, and I'm still paying for that."
For many, the ends seemed destined to justify the means. "It was this special period right at the end, the hardest of the hardcore," Gage said. "We didn't want to give up. We wanted to keep working on it until the very last minute. Even after we launched, there was definitely a [break period], but several of the guys, myself included, were still coming in the same hours because we wanted to see how the game was doing. We'd get on Battle.net and want to know what people were doing."
Despite shared enthusiasm, no developer was immune to the toll crunching was taking on their bodies, their minds, and their relationships, not even Allen Adham. Even though he was part of most strike teams, Allen tired of having his directives questioned and picked apart. In the past, when Blizzard had been smaller, he had made decisions on the fly. Now, he had to wade through the processes and structures that he had helped construct. "Allen would get up in front of everybody, and by the time he finished talking, you were ready to walk out of there and do whatever it took," said Harley Huggins. "He was really good at that. That happened over and over. They'd say, 'If you don't have to go home for Thanksgiving, don't. If you have all you need to celebrate Christmas, don't leave.' Allen was good, but I think even his Jedi powers were taxed by the end."
Allen was not one to expect his team to do anything that would make him balk. Although no developers felt forced to choose work over family vacations, birthday parties, and holiday celebrations, they chose to remain at the office of their own volition. So did Allen. Crunching eroded the health and personal relationships of developers and their families even as it tightened bonds within the studio. Most days at three a.m., packs of developers headed down the street to grab cheap food at Del Taco. Employees who played on the company's roller hockey team skated around the parking lot to shoot around. Others crowded around Street Fighter II cabinets and pinball machines. Some nights, Duane Stinnett parked a chair outside Pat Wyatt's office and strummed tunes on his bass, providing the perfect backdrop for the Magic: The Gathering players sitting cross-legged on the floor amid spreads of cards. After an hour or two of play, developers drifted back to their workstations.
At trade shows, Pat Wyatt gave interviews with an air of confidence that sometimes felt as hacked together as his pathfinding fix. "It's either hubris or fear," he said, describing what developers experience when they walk show floors and scope out competing products. "It all depends on what you're bringing to the show or what you think you have that other people don't have yet. Usually it was fear because there's just lots of other products out there with gigantic marketing budgets. Blizzard was just an up-start company. Even though we'd had success with WarCraft I, WarCraft II, and Diablo, there wasn't anything that was necessarily going to make StarCraft rise to the top compared to other games except what we ourselves put into it. That's a pretty daunting thing to do."
By late March of 1998, fourteen months after Blizzard Entertainment began crunching, nothing could stop StarCraft from invading the hard drives of PC players around the world. Nothing except StarCraft itself.
On March 24th, the gold master disc was shipped off to Blizzard's production factory in Utah. Back in Irvine, Bill Roper decided to run one last test cycle. Just to make sure. Just in case. He discovered a game-breaking bug that was present on the gold disc and, if left unchecked, would inevitably damage StarCraft's reputation with critics and consumers. Programmers pounced on the bug. A new test cycle was initiated. For three days Bill and others alternated between testing the game and sleeping under their desks until, on March 27, his birthday, the game went gold again.
StarCraft landed on store shelves on March 31st. The fatigue that had settled over Blizzard like a fog burned away, temporarily replaced by exultation. Gage Galinger loaded a space station map and scrolled up, down, left, right. His parallax effect, one of his first trials by fire at Blizzard, had survived the piranhas. He gleaned satisfaction from that, and from the fervor with which players snatched Terran-, Zerg-, and Protoss-themed game boxes off shelves, going on to sell over one and a half million units by the end of the year. Even so, he decided that Blizzard's culture was not the right fit for him.
"That's probably the happiest thing about that time that I recall," Gage said. "Almost everybody is extremely successful today. Many of them went on to start game studios outside of Blizzard; many of them came back to Blizzard. It's just amazing. I just feel really happy to have a shared DNA with that group of people, because I really think that's one of the few hotspots that spawned so much gaming goodness. I'm really happy to have been a part of it. Pat [Wyatt] and I didn't get along when we worked together—he was my boss—but to be honest, I'm really thankful he hired me; I'm really thankful that he gave me that opportunity, because I would be a different guy and I wouldn't be nearly as successful as I am today if I didn't have that experience. I hope that makes it into your book."
In what would become a tradition for every product release, company leaders fêted StarCraft's launch by ordering champagne in bulk and passing around bottles. Everyone who had one popped the cork and doused themselves and neighbors with streams of bubbly. A representative from Blizzard's parent company at the time was incensed when he saw two developers, drunk from the bacchanalia, get into a fist fight. The brawl ensued after a misunderstanding, when one guy had stepped on the head of the other, who had been lying on the floor. The representative fired them on the spot, only for Allen Adham to swoop in. No one had authority to fire anyone on his team, especially not some outsider.
"Blizzard guys can party like rock stars: we drove to Vegas in
a giant bus overflowing with beer."
A few months later, following regular support for StarCraft to address bugs and game-breaking glitches, Blizzard treated the team to a wrap party in Las Vegas, another tradition. "Blizzard guys can party like rock stars," said David Pursley. "We drove to Vegas in a giant bus overflowing with beer. I was just barely twenty, so I didn't drink. I didn't know if I was allowed. Everyone released all their tension in Vegas in every way possible."
As the celebration wound down, developers trickled out of the office. "The company would typically provide some extra time off depending on the amount of crunch," said Frank Trevor Gilson, an engineer. "People would string that together with vacation time and maybe be gone for four weeks."
Years later, Bob Fitch claimed that StarCraft's development, in particular work centered on its engine, was no more draining than any other project. Many of his colleagues, who dragged themselves back into work after taking vacations, disagreed. "This is going to sound kind of weird to people who work eight-hour days, but everybody worked a solid sixteen-hour day, then went home and got eight hours of sleep, and did that over and over and over again until we made a hit," said Wyatt. "It was much more sustainable than working forty-seven-hour days, then getting twelve hours of sleep, then another forty-seven-hour day. But we just did that schedule for a really long period of time. Relationships were destroyed. People got sick, so their health was very fragile. But eventually we managed to get it out the door."
"There's a point where you can get a certain amount of people in a group, and everybody's going to share the feeling that they'll do anything, put as much as they can into the cinematics," explained Harley Huggins. "But at a certain point, you can't hire that many people who show that mindset. You're going to get people who just want to work nine to five."
Jeff Strain, who had caught hell from his wife for coding a map editor for his video game during the birth of their child, was ready for change. "A lot of it, in terms of my observations about StarCraft, are coming from the perspective of an employee of the company. Thinking to myself, I don't think I can do this again. I can't put my family through this again. Literally I had a blanket in my office, I'd collapse around three a.m., sleep for four or five hours, get back up, get a cup of coffee, and keep going. It was very clear for me that I as an employee couldn't go through another cycle like that and would not work through another cycle like that. As much as I loved the games I was making, I knew that wasn't sustainable. It was very clear for me that I as an employee couldn't go through another cycle like that. As much as I loved the games I was making, I knew that wasn't sustainable."
You've been reading a chapter from Stay Awhile and Listen: Book II, you can read more about the book and support it on its Kickstarter page.