If you find this article interesting, consider backing The WoW Diary. The campaign will be ending on Feb 24, 2023 at 9:00am PST —
Below is an excerpt from The WoW Diary on the pitfalls of success, written by John Staats —
The Pitfalls of Success
One of the most surprising takeaways from my book, The World of Warcraft Diary, was its melancholy ending. Readers and fans wanted to know how WoW’s launch affected the team’s morale. Were the new employees similar to the original devs? What were the growing pains of shipping a mega-hit like World of Warcraft? Did development become harder or easier after Vanilla WoW? Does money solve everything?
Success Wasteland
Early in the dev cycle, surprises and discoveries from new technologies and features buoyed the team’s spirit—controversies about how things might pan out kept our imaginations fresh. By the end of the project, WoW was a known quantity. By the time we were playing it on live servers, we had had too much of it.
The dev team’s low morale came from a general burnout from fixing bugs, iteration, and long hours. Instead of coming up with new ideas, we were putting out fires. Some devs wanted to work on a new game—or do anything different. We were sick of eating dinner with one another. No one had anything new to talk about since no one had a life outside of work.
The dev team knew our post-launch fate would involve more of the same—fixing bugs, iteration, and long hours. And we were right. It took longer than a year to ship significant amounts of new content. Nearly half of the employees moved on to other companies after the game launched. Many left before receiving bonuses. Some didn’t care about bonus money, needing only a change of scenery. In the dynamic business of video games, it’s healthy for creatives to come and go.
After Blizzard North’s leads left Diablo to form Flagship studios, Blizzard’s top executives interviewed the remaining staff to assess their willingness to continue working on Diablo or another Blizzard project. They uncovered stories of a toxic culture that “shocked and appalled them.”
One problem was Blizzard North’s founders ran their studio like a mom-and-pop shop. Even though the Diablo team’s size was comparable to WoW and much older, they’d never fired a single employee. That was a big red flag because people don’t always work out. Even Vanilla WoW had to let people go.
After WoW shipped, Blizzard executives enacted measures to avoid repeating Blizzard North’s predicament. The first measure required everyone to take annual sexual harassment courses. The second one was more controversial—they forced the WoW team to cut down its workforce by ten percent.
In theory, layoffs help organizations cut down to their most productive employees, a practice many large companies employ to prevent ossification. The flaw with this decision was team 2’s careful hiring. We didn’t have unproductive staff members. HQ’s decision was costing us self-motivated coworkers whom we’d bonded to throughout years of late hours and crunching. We were losing rare and valuable staff members for no apparent reason.
Mark Kern, the team’s lead, quit in protest over the layoffs. He left with several employees, including WoW’s art director, beginning a company called Red Five. More walkouts followed, including over a dozen team 2 members who formed Carbine Studios. Others fled to join Ready At Dawn, a studio previously founded by ex-Blizzard devs.
I based The WoW Diary on notes about the team’s progress. Under these distressing circumstances, I couldn’t bear the thought of keeping track anymore. In my mind, team 2 was gone. Aside from bug fixing, no one knew what we were working on because absences from walkouts and vacation time made meetings pointless.
New Blood
With the money pouring in from World of Warcraft, Blizzard began hiring en masse. It acquired Swinging Ape Studios, who’d worked on a second iteration of the ill-fated Starcraft: Ghost. After canceling Ghost, team 2 absorbed its employees. After WoW had proved itself, plenty of viable candidates submitted their resumes to replace the people we’d lost. Devs from other companies filled our empty offices.
As an old-guard veteran of Vanilla WoW, it would be easy for me to claim there was no replacing the original devs, but nothing could be further from the truth. The new devs were overwhelmingly positive, talented, and committed workers. They were just as sharp and enthusiastic as we had been when we started the project.
Not all the new hires worked out so well. WoW’s new team leader announced he planned to outsource artwork to save money and lay off Blizzard’s in-house art staff. Money was the least of Blizzard’s problems, and keeping employees was its most significant challenge. The idea was so incendiary that Blizzard ushered him out the door before he’d completed his first day. You can’t just drop a bomb like that and expect to lead the team.
For the most part, employees weren’t Blizzard’s growing problem—it was the team’s new structure.
Much of WoW’s appeal came from team 2’s ability to crank out tons of MMO content. We did this by fostering a relatively hands-off environment. WoW’s old art director often mused, “You guys know what you’re doing.” Some departments, like the quest designers, had zero supervision. Unless decisions affected the schedule, producers entrusted the devs to make a good game. For the longest time, only four directors for tech, art, animation, and design served as its big-picture decision-makers—and everyone working under them had direct access.
The minimalist structure kept everyone on the team at the same level of responsibility. Our salaries, tasks, and job titles differed, but we shared the same amount of ownership. Without supervision, we occasionally veered off and wasted time. Some experiments yielded useless byproducts, but the freedom empowered us to create a surplus of content. Without red tape, our output astonished Blizzard’s other dev teams, who’d witnessed the project’s growth from behind the scenes.
Blizzard decided it was time to organize team 2. WoW was a success and was a live product—meeting customer expectations was paramount. Who could argue with the wisdom of keeping team 2 on schedule and dialing back on experimentation?
Team leads for every department popped up. Our new time clocks prevented devs from crunching, but everyone hated the hassle of checking in and out. But adding people and governing our productivity didn’t improve our ability to create content. WoW’s first expansion took over two years to make—far from the nine-month dev cycle that management had hoped. Team 2 wasn’t wasting energy, but it took forever to get anything done.
As the roster ballooned beyond a hundred employees, department leads had little time to administer the growing staff. They promoted sub-leads to deal with coordination, task oversight, and general management. Yet content creation remained ridiculously slow.
Management discussed ways to increase productivity. One such suggestion involved splitting the project into two dev teams—one for live updates and one for expansions. I doubt the idea would have ever worked. Both groups use the same assets and tech. Everyone would have preferred to work on expansions filled with exciting new features and tech. Relegating devs to patches would have felt punitive. Another idea involved splitting team 2 into separate expansions, leap-frogging over each other. This measure would have caused more problems since each generation of WoW piggybacked on its predecessor.
Blizzard kept hiring more people to speed things up. Why was development taking so long? WoW was a known quantity. Rinse-and-repeat seemed like it should be easier and faster.
Hierarchy Versus Decentralization
Kevin Kelly’s 1994 book Out of Control investigated types of organizations. Kelly’s book covers chemistry, biology, society, and business to illustrate organizational patterns. He observed that highly controlled organizations produced predictable results at low yields. They were very inefficient. Decentralized and loosely structured systems were far more productive, yet waste and erratic results were common byproducts.
The classic example is an orchestra versus a garage band. A conductor controls the orchestra and plays a predetermined music score over long hours of practice. Orchestra members earn well into six figures. As with anything involving high salaries, there’s politics about who plays the first chair or joins the orchestra. There are no surprises. This system works because the goal is to replicate a known result—a piece of music.
A garage band works under modest conditions. Instead of dedicating themselves to known compositions, bands create new music. Everyone has a voice, and the group functions as a meritocracy, following whoever has discovered the coolest vibe. Instead of top-down management, direction comes from the bottom. Members jam off of each other’s riffs. The group’s sound changes as members come and go. Anyone who’s read The WoW Diary recognizes I used these phrases to describe team 2’s development process.
Vanilla devs produced waste, but the bits we lopped off left far more content than the highly orchestrated expansions and updates. People work faster and are more creative when following their instincts. People had ownership of their work.
Making devs responsible for one another introduced unnecessary paranoia. No department lead wanted the blame for putting the team behind schedule, so every diversion from the master blueprint was viewed with nervousness and suspicion. And who could blame them? Who among us would risk their shot at Blizzard management over a subordinate employee’s dalliance?
On Vanilla WoW, devs could abandon tasks that weren’t fruitful and explore fertile ground without half a dozen leads signing off. It was faster to ask for forgiveness than to get permission. WoW’s expansions enjoyed no such flexibility.
Perhaps an anecdote might illustrate the shift in team 2’s new workflow. I’m a level designer, not an artist—at least not up to Blizzard’s standards. Dungeons began as a part of the art department, then moved under design after a couple of years. When Blizzard moved dungeons back into the art department after it acquired Swinging Ape Studios, it eliminated my chances of getting promoted to a lead. That was fine by me. I didn’t know the Swinging Ape artists, nor did I want to manage them.
At first, it was a comfortable fit. We had a good lead and a strong staff of talented artists. Instead of hogging all the juicy raid dungeons, I focused on technical projects, like rebuilding old work, working on new tech, and polishing anything that fell through the cracks. I was the only person on staff aside from the engine programmer who could do them.
Problems came up whenever I saw room for improvement. Instead of just polishing things, I had to convince various department leads that things needed polishing. But department leads weren’t comfortable redoing things that the art director had given the green light. I had to go around people’s backs to get conversations started. My efforts made the leads nervous, and I found their resistance frustrating. I wasted more time pitching ideas than it would take to execute them.
Uldum is a perfect example of why I left the WoW team. One day, I received a task to combine blocks of temple wall sections in Uldum, an Egyptian temple in Tanaris. One of the team’s landscape designers built the walls using the only tools available—an in-house editor called WoWedit that combined art and design assets into playable zones. It offered basic placement features, far less flexible than our 3D modeling software dungeon designers used to make architecture. To my eyes, Uldum looked like rigid copy-and-paste geometry. Combining its repetitive features was meticulous and boring work, but it wouldn’t have bothered me if the wall sections weren’t so ugly.
The landscape designer who built Uldum had a strong artistic eye, but WoWedit’s orthogonal tools produced LEGO-like architecture. Uldum’s walls all looked the same. Each of its stair sections was identical, and it came across as mediocre. If you’ve seen one part of Uldum, you’ve seen it all. What’s worse, cobbling together all these pieces was time-consuming, as was my clean-up work of joining them together. The designer who worked on it didn’t understand what was wrong—he’d taken Uldum as far as anyone could with his tools.
I knew of a better way of doing it. I built smooth walls behind Stormwind and modeled Utgarde Keep, two areas where the architecture reflected the surrounding landscape. In each instance, I’d constructed them in just a few days—although, in Utgarde’s case, I worked from an excellent concept sketch by Eric Browning, a dungeon sub-lead.
For Vanilla WoW, I rebuilt the Slag Pit, Loch Modan Dam, and Thandol Span. The previous versions were clunky, and the art director approved the upgrade. Rebuilding the landscape around the new versions took less than a day.
But in a deeply structured team, too many people were answerable to every aspect of the game. No one wanted to be responsible in case my idea didn’t pan out. After over a week of discussing it, enough people agreed to hold a meeting about it, leaving no one out of the loop. The meeting needed leads from three departments, two sub-leads, two producers, our art director, and the landscape designer who’d originally built Uldum.
Department leads were often away from their desk, and the art director was always in meetings. If one was on vacation, taking a personal day, or had booked another appointment, I needed to find another time slot. After many days of scheduling conversations, I found a time two weeks into the future. Until that time, I occupied myself with other tasks.
When the meeting came, only two other people showed up. One producer was out sick, and the art director and the leads got pulled into an emergency matter. Six weeks had passed since I’d first pitched the idea. I tried scheduling another meeting but abandoned my efforts after no opportunities arose within the coming weeks.
I’m a persistent person, but I threw in the towel. Uldum remains as it was when I first saw it. I spent the next several days welding geometry, letting the matter rest.
The most frustrating part was that I couldn’t fault any of the leads. None were unreasonable. Expecting anyone to sign off on things they didn’t understand was antithetical to their job description. The dev team’s hierarchy made so many people responsible for the game’s assets and features that communication became impossible. Instead of fostering polish and iteration, team 2’s structure suppressed it.
Uldum wasn’t a disaster. It was fine for giving Tanaris some personality. I’m illustrating what happens when devs get ideas. Instead of doing what they’re good at, devs must become pushy salespeople to be heard.
Artists and designers aren’t typically tenacious. They won’t advocate for their ideas, nor are there many opportunities to do so. I’ve often wondered how many cool ideas succumbed to this process. How many were never pitched because of the hassles involved?
On Vanilla WoW, devs with an idea usually executed it themselves. If not, they at least had regular contact with the department director, who made decisions on the spot. Another weakness in team 2’s deep org chart manifested when devs lost contact with directors. On The Wrath of the Lich King expansion, it occurred to me that team 2’s art director had never set foot in the dungeon department until the end of the dev cycle. He regularly attended dungeon team meetings, but with twenty people involved, it gave little time for one-on-one conversations.
Without direct access, pitching ideas or asking for tools involves a telephone game of convincing the sub-lead to talk to the department lead to talk to the art director to talk to the producer. Intermediaries were uninspired advocates for such suggestions. Given the scheduling constraints and bottleneck of issues to discuss, the conversation ever reaching the ears of a decision-maker was improbable. Communication was so sporadic managers usually forgot the idea before pitching it.
When devs don’t control their work, problems roll uphill into the lap of the beleaguered department director and producers—who often don’t fully appreciate the minutiae involved. Directors don’t seem to enjoy the situation either. Instead of proactively leading the team, they’re putting out fires that should be far beneath their job description.
Control Versus Efficiency
Devs hitting roadblocks often wait hours, days, or weeks before getting permission to go around it or find another solution. The lack of creative freedom increases the need for communication under conditions that suppress it.
Highly controlled systems are efficient only in static conditions. It is efficient only when a factory has all its raw materials, personnel, and working machines. Broken machines and missing resources can bring the entire process to a halt. Game development is dynamic. The nature of software development is unpredictable—conditions change.
Throwing more people at the game only exacerbates the problem in controlled organizations. Hiring more devs required more middle management. More people have to sign off on every change, making things harder for the worker bees, not easier.
But what choice does a prosperous company have? After WoW became the goose that laid golden eggs, having lots of eyes on development seemed a prudent way to maintain success. Empowering employees to be responsible for their work would summon a hundred-headed hydra. Who wants to herd that many cats? Team 2 had about 150 people when I left it so many years ago, and I wouldn’t doubt if today its number is double.
Mike Morhaime cites self-publishing as Blizzard’s chief reason for success. It keeps the company agile enough to improvise—an essential ingredient because development is an iterative, exploratory exercise. Yet creatives without the same freedom are ineffectual.
Working on a deeply structured team often feels like riding on a bus. It’s fine if everyone knows the destination, but game development is never so cut and dry. If passengers spot something worth investigating, the driver is usually too preoccupied to change course in time to exploit rising opportunities.
Creative companies must hire carefully and trust their employees. The flatter the organization, the more it empowers devs to become more accountable, productive, and pleasantly unpredictable.
Kevin Kelly’s Out of Control is available on Amazon. If you enjoyed this article, there are many more like it in my book, The World of Warcraft Diary, written from the lessons learned while making WoW. I am crowdfunding the printing of its second edition.
