It was early May, 1998. nearly all of Ensemble Studio’s employees were in San Jose, aboard the Queen Mary. We had just been awarded a pair of Spotlight Awards from the Computer Game Developers’ Conference, including “Best of Show.” The statuettes looked like miniature Klieg lights, and someone had plugged them in on one of our tables. Sheets and sheets of stickers advertising an entertainment network called Berzerk littered the bar. We had taken to using them to replace the labels on our beer bottles and, as they became more numerous, to toasting “berzerk” and then just randomly yelling “berzerk.”
Eventually, the people who actually owned the service approached us, cautiously, to ask if we worked for them. By this point, we’d received a number of awards for Age of Empires (the game shipped in October of the year before), but this was the first award we’d been given formally. Officially. In a ballroom. Surrounded by hundreds of our peers. It was unbelievable. We’d also been informed of our updated sales numbers, well north of a million now, which didn’t hurt the mood any. In a few weeks, the entire company would be in Atlanta for E3, attending an award ceremony where we’d take home the Interactive Achievement Award for Best Strategy Game of the Year.
It was an amazing time to be part of Ensemble Studios. But even at this point, while we were still celebrating Age of Empires, we were already months into a new struggle—trying to figure out how to do it again.
The man most obsessed with this, our president Tony Goodman, wasn’t at the table that night, drinking “Berzerk beer.” It would be romantic to claim that he had skipped the celebration to stay back and work, but Ensemble wasn’t that kind of shop—Tony wasn’t at the table because he had sneaked into the off- limits area of the ship and climbed up one of the Queen Mary’s smokestacks.
In The Beginning
Ensemble Studios had its roots in a consulting firm: Ensemble Corporation. The story of its genesis goes something like this:
One night in 1993, Tony Goodman was talking to Angelo Laudon, one of the programmers at Corp. While discussing the business, one of them (neither remembers which) asked, “Wouldn’t it be more fun to make games?”
Boom. That’s it. No multimedia presentations, no spreadsheets, no millions in projected profits for investors, no conversations about “market opportunities.” Just two pals who had been thinking about how much fun it would be to make games for a long time.
And that’s Ensemble Studios in a nutshell, really. Everything that was good about it, everything that was bad, every success and failure. Right there. “Wouldn’t it be more fun to make games?” That was the unspoken vision for the company—take the awesome, manic energy you get when you have two pals working on something they love and scale it up to 25 or 50 or 100 people.
Judging from the end results—no games that weren’t in the million-seller club, a franchise that has sold more than 20 million games, a developer that stayed in business for over a decade, a profitable acquisition by Microsoft, exceptionally low turnover—that vision served us incredibly well over the years. But, effective as they may have been in the final analysis, a lot of the principles and practices that defined us also became things that we struggled with.
The problem is that nothing stays still. Once you have worked and worked and worked, and your labor has been rewarded with success, it’s tempting to imagine that you have things figured out. That all you
really need to do is follow the recipe you perfected the last time and you’ll bake the same cake.
This is fiction.
When I think back to the making of Age of Kings today, I can’t help but do so with that in mind. All of my recollections have a kind of duality to them—these are the things that made us strong… but also made us weak. It was during the development of Kings that this duality started to surface—when we first saw that some of the ingredients responsible for making Age of Empires a delicious cake would produce very different baked goods in the future.
What follows are six of these “ingredients”—the founding principles of Ensemble Studios that were vital to our initial success and that morphed as we went from start-up Age of Empires to follow-up Age of Kings.
Principle 1: Full-contact Hiring
If your goal is to put together a company that feels like pals working on something, a good first step is to bring in people who actually are friends. Among the earliest members of Ensemble were Rick Goodman (Tony’s brother), Brian Sullivan (Tony’s boyhood pal), Bruce Shelley (who met Tony at a University of Virginia game club when Bruce was in grad school), Bob Wallace (friend of Tony’s father), and John Evanson (son of another of Tony’s father’s friends).
To achieve and preserve this culture of friends, Ensemble required the entire team to be involved in the decision to bring any new member aboard. Every candidate met with everyone in the company, usually several times. Typically, this meant a lunch or dinner with everyone, a few interview sessions, and a night of playing board games. When these were completed, there would be a meeting to discuss the candidate and then vote. Unanimous assent was required for anyone to get an offer.
The intent was that everyone would own the decision to add someone to the family. And it worked. Nobody ever showed up on Monday to find the new guy “management” had hired, mysteriously in the office. Anyone there was there because all of the others had agreed to it.
New hires knew this too (the full team involvement and time dedicated to hiring was almost universally mentioned as one of the things that impressed candidates about the studio). On their first day, they went in knowing that they had a vote of confidence from all of the people they were now working alongside.
During Age of Empires, which required growing the team over a relatively long period of time, this approach to hiring worked well and produced a close, unified crew. Age of Kings had a different profile. We needed to hire more people within a shorter period of time. A hiring process that took at least two full days, involved most of the team, and included group dinners after work caused problems as the number of candidates piled up. Some weeks, the average employee would have to block out six or eight hours just for hiring.
Clearly, that was a substantial amount of time, but we weren’t worried about any lost productivity—we felt it was worth it to ensure that we hired well. The problem was fatigue. Because of the pace at which we needed to add to the team, the hiring process had gone from being an amazing opportunity to a boring chore for a lot of people. This was but a taste of what was to come—we grew from 20 people to more than 30 between Empires and Kings, but when we moved on to Age of Mythology and beyond, we would cruise past 70 and end in the triple digits. But even during the hiring for Kings people were starting to tire of the process.
During Kings we also started to see that, although it produced excellent results overall, entrusting everyone with this responsibility could lead to abuse and missed opportunities. For example, someone could think we shouldn’t grow past a certain size and then decide to vote “no” on any proposed new hire.
Principle 2: Consensus-based Everything
As the “friends working on something they love” bit might lead you to believe, Ensemble Studios was not a “here are your orders” kind of place. Nothing illustrates this better than the manner in which we decided to actually do Age of Kings: the entire company piled into the old conference room, arguing the merits of the various alternatives, then voted on what game we were going to do next. “Knights and Castles” beat (if memory serves) “Elves and Castles.”
There is a fantastic strength in this approach. The productivity and potential of a group of people working on something they all “see” and want to achieve are magnitudes greater than those of a team that needs specific marching orders for every step. We experienced that strength during the development of Age of Empires, when everyone involved seemingly just rowed feverishly in the same direction. We wanted to preserve that feeling for Age of Kings.
There are two ways you can achieve this strength. The first is to have a small team. The second is to spend a great deal of time trying to get people to agree. When my business card changed from “Assistant to Lead Designer” to “Lead Designer,” I got a crash course in exactly what was involved in the latter. If someone on the team was unhappy about anything (too many hitpoints on the towers, silly looking capes on the swordsmen, an ugly font on the UI), it was a problem—and my job largely became dealing with those problems.
Strong as it might be to have everyone in agreement and marching in the same direction, consensus- ased operations are difficult to achieve and to scale effectively. As we grew during Kings, the process became more and more difficult to manage. We learned that when we hired new people, they likely had no experience with it and certainly could not call upon the countless hours of discussion that had taken place before their arrival. A person who didn’t like or play realtime strategy games, for example, would often complain about fundamental elements of real-time strategy games in an attempt to steer things toward the sort of game-play they preferred. (Using logic to explain why you shouldn’t attempt to insert a first-person shooter into your already-late real-time strategy game is surprisingly ineffective, by the way.)
Overall, we managed to stick with a consensus-based approach for most of Age of Kings. The majority of the systems in the game, from the sheep to the scenarios in the campaign, had their roots in long design meetings that involved the entire company. When we had to make calls without full agreement, we learned another important lesson about consensus-based operations: Once people get used to operating this way, it is very difficult for many of them to accept decisions that aren’t made by consensus.
Principle 3: Hardcore Play-testing
During the late ‘90s, when we talked about our process at conventions and the like, nothing caused more disbelief than the claim that the entire team play-tested. Today it doesn’t seem uncommon (most devs I speak with now say their teams do this), but back then it was a pretty foreign concept to have all of your artists stop making art and all of your programmers stop writing code so that they could play their game for a few hours and then give some feedback.
Play-test was the crucible for Ensemble—the arduous gestation, labor, and delivery process through which the game was born. The idea was to get something up and running as soon as possible and then start hammering on it. On the average, at least half of every day I spent running a play-test, taking feedback from the people there (about a quarter of the team), making changes, and sending out summaries and tasks. Additional discussion on anything in the summaries was sometimes taken into a design session, but more often issues were hashed out in giant, endless, multicolored threads that everyone complained about reading.
This play-testing process always worked for us—in fact, the Ensemble alumni at Robot Entertainment and Bonfire Studios still use it today. Looking back, I don’t recall any significant signs of problems with this approach to play-testing during the development of either Empires or Kings. However, where it has indirectly caused some issues—issues that have grown more pronounced over the years—is in the minds of those who were actually there at the beginning.
You see, when you’re 22 and you’re hell-bent on getting your first game done and you’re going to set the world on fire and make the cover of PC Gamer, and then you actually do get your game done and set the world on fire and make the cover of PC Gamer, it burns a set of warm and fuzzy memories into your little sleep-deprived brain—awesome, awesome memories of how fantastic everything once was.
That stuff is great.
Great, that is, until you’re working on your next game and the AI has been broken for a week and multiplayer is out-of-synch crashing and you turn to the guy beside you and ask something like, “Remember when we used to all play the game all night every night, and everyone loved everything, and it was perfect, and fans lined up to tell us how awesome and cool we were, and it was sunny all the time, and kittens never died?”
Then it sucks.
Principle 4: Crunching Like You Mean It
Age of Empires had crunch. Not crunch like “we’re working until 10”; crunch like “if not asleep, then in office.” For months. This was not specifically mandated. There was a general call to put in more time—something like: “OK, we need to start crunch next week to get done on time.” But to my knowledge, nobody said, “Be here between these hours.” People on the team were motivated to put in silly amounts of time.
Kings had crunch as well. It was expected. Some of us looked forward to it. But, unlike crunch during Age of Empires, people started to complain about it. Not gripes about the extra hours (which, of course, we had during Empires), but real complaints. Complaints of the “I’ll have to leave if this doesn’t change” variety.
Now, I am a mutant, and I have taken many lumps over the years for this, but: I like crunch.
Yes, yes—it isn’t healthy. It isn’t good. It’s damaging and it should be avoided. I do whatever I can today to ensure that none of the people I work with have to endure it. But my fondest memories, not just of Ensemble, but of life, are of finishing something as the sun comes up, weeks of beard on my face, awake only thanks to gallons of coffee, standing beside others who loved whatever enough to do the same. There’s some magic there, and I love that with all my heart. I was not alone in this feeling at Ensemble so, for some of us, the degree of serious opposition to crunch that surfaced during Kings was disappointing—not because people didn’t want to spend dumb amounts of time at work, but because it was a sign that things were changing. With Empires, our first time up to bat, there was a universal feeling that we would do “whatever it takes.” With Kings it felt like we were taking a step back.
There’s only so much you can do to change this problem (aside from eliminating crunch). Ensemble already had a general disdain for Mickey Mouse bullshit, which does a lot more than you’d expect to lessen the impact of things like extra hours. Nobody really kept track of vacation or sick days or when someone rolled into or out of the office. Drinks and snacks were free. Lunch and dinner were provided any time crunch was in effect (and you could expense dinner for impromptu late hours any time you wanted). Some of this, like the drinks being free, might seem trivial, but if you don’t understand why someone working in your office at 2:00 a.m. is going to be insulted by having to pay a quarter or two for a drink, you probably shouldn’t be running a team.
What did change was that crunch went from being unplanned and perpetual to being scheduled and limited. Instead of just announcing a need to start putting in extra time and keeping that in effect until we were done, we started setting hours and picking periods of time—12:00 to 12:00 during the last week of the month, for example. This at least allowed people to plan around crunch and have downtime.
Principle 5: A Cult-like Culture
A lot of people are friends with the people they work with—some more than others. Their closest friends are usually the people they visit around noon, to see about getting lunch together. At Ensemble during Age of Empires and the early stages of Age of Kings, someone would start marching around the office at noon, bellowing “Lunch Train!” and the company would turn out to go get something to eat.
The same thing happened around 7:00 p.m., when someone started walking around recruiting followers for dinner or drinks. On any given Saturday night, a good chunk of the team was at someone’s apartment. Friday after work there’d be a dozen people in the play-test area playing Quake (then Quake 2, then Half-Life) until 3:00 a.m. That was the way of things. If you didn’t actually share an apartment with the guy you sat next to, you were probably going to be having a beer at his place later that night.
A great deal of this cult-like culture was a function of circumstance. A startup game studio, in Dallas, with no published titles, was not an attractive proposition to most people with experience in the industry. Many of the early members of the team were young, single guys, recently out of college or the Art Institute. Similar age, similar interests, similar situations.
Ensemble’s general approach to teambuilding only reinforced this culture, and not just because of things like the hiring process and the shared hardship of crunch. When GDC and E3 rolled around, we sent the entire team. The entire team, as in everyone. If we were showing that year, everyone went on to the floor and demoed the game. When the show was done for the day, 20 or 30 geeks clad in shorts and gaming t-shirts loaded onto a bus that, after a short ride, unleashed them upon an unfortunate fivestar restaurant. Finished there, the same crew would make its way to a nearby bar to drink until someone did something that got everyone tossed out.
I know that sounds like dot-com-era shenanigans. It wasn’t. Keep in mind that everyone went to E3 in 1995, long before Age of Empires shipped or Microsoft acquired the studio, when the company was certainly not possessed of vast sums of disposable income. It was part of the philosophy of Ensemble. We sent everyone to demo the game because we couldn’t imagine allowing anyone else to show our baby to the world. The company organized and paid for bacchanals at upscale restaurants because we wanted our people to have shared experiences—and the stories to tell that come with those shared experiences.
Of course, in time the culture changed. Of particular note: Somewhere between Empires and Kings some of us grew up. The guys who could hang out every night in ‘96 had wives or kids in ‘98. Through Kings and beyond, more and more of our crew fell into the “adult” category, and the balance shifted.
The other thing that changed the cult was hiring. Pre-Age of Kings, for better or worse, there was really nobody at the studio for whom Ensemble was “just a job.” After Empires, we had enough credibility to attract experienced talent, and for some of these people Ensemble was their third or fourth stop. The more we hired, the more we introduced people who had other things going on outside of our circle—people who, at the end of a day of demoing a game, wanted to go back to their hotel rooms.
As with complaints about crunch, this was difficult for some of us to understand or deal with. It’s not hard on an intellectual level to comprehend why any person who just endured six hours of flashing lights and looping sirens would want to enjoy some quiet for a bit. Still, we just did a fantastic job of selling a game for which we all worked our asses off, and now we’re all going out to celebrate together… and someone doesn’t want to go?
That change is not the kind of thing you can “fix.” You cannot say, “Everyone must go out and have fun together.” But when it doesn’t happen, there’s a sense that something is amiss.
Principle 6: Free-range Development
I have a form of narcolepsy that sometimes visits me with acute insomnia. One night, my eyes pop open at 3:00 a.m. and I know I’m not going to fall back asleep. I get dressed and head into the office. I’m off the elevator and headed for my desk when I hear something a few offices down, so I detour to check it out. Angelo Laudon is there. Writing code for Age of Kings. He couldn’t sleep either.
There’s a chapter in Steven Pressfield’s The War of Art that, paraphrased, states that you’re doing the right thing if you’d do it even if you were the only person on the planet. Arnold Schwarzenegger, he suggests, would still go to the gym even if there was nobody around to impress with his physique. Likewise, I’m certain Angelo would still code games were he alone on Earth. In those days, Ensemble was powered by people like that—people doing, in essence, “this thing they were born to do.”
That isn’t to suggest that there wasn’t a massive amount of planning and scheduling. Guys like Rick Goodman (lead designer on Age of Empires), David Pottinger (lead programmer), Brad Crow (art lead), Chris Rippy (producer), and Harter Ryan (executive producer) certainly did a great deal of work to organize and direct development. Still, everyone had a great deal of freedom and might head down one path or another.
Frequently, this creative freedom had fantastic results. Numerous pieces of Age of Empires were completed because someone thought a feature should be in there and felt like working on it. It was not unusual to roll into the office on Monday and find an “over the weekend” email informing everyone that some horrible bug was fixed simply because someone had grown annoyed with it. The artists learned tricks on Age of Kings that allowed them to do things with a 256-color palette that nobody thought they should be able to do, not because it was planned, but because they decided to start playing with various techniques.
Empires, despite being Ensemble’s first game, did have an advantage in that it hadtime to find its way a bit more than Age of Kings, which Microsoft initially wanted delivered after only one year of development. The limitations of a 12-month schedule did not pair well with established freedom to pursue objectives you found interesting.
Nowhere was this more obvious than in the initial design of the game. While envisioned at the start as simply moving Age of Empires into the Middle Ages, the initial design direction for the game consisted of mixing in every idea that hadn’t made it into Age of Empires. The poster child for this was a full-screen diplomacy matrix which allowed players to negotiate all manner of convoluted relationships. However, there are dozens of other things we worked on that did not pan out, including unit facings in combat (micro each of your 200 little guys to stab enemies in the back), off-map trade (sometimes you send a trade ship out and it just doesn’t come back—you don’t know why), and pillaging (if you killed a villager who was gathering, he’d drop a lump of whatever he had on him). We also worked on persistent fire, mercenaries, outlaws, roads, heraldry design and display, raider civilizations, renewable resources—all interesting ideas that didn’t make it into the final game.
The freedom to pursue things like this was a double-edged sword, and it forced us to reset the design of the game and spend an additional year on development. It would also bite us later in development when people decided to rewrite systems like the AI and pather instead of improving what had shipped in Empires. Consequently, we had to pull people off our second game and engine team to finish Kings on time.
Here We Go Again…
It’s late summer, 1999. We have just learned that the last release candidate we put up has been accepted and Age of Kings is headed for manufacturing. Thirty-some people are popping bottles of Dom Perignon in the cafe of our new offices.
This is just one more thing that has changed. Two years prior, a handful of guys crawled out of an un-air- conditioned ex-dentist’s office to celebrate finishing Age of Empires with Moons Over My Hammys. Now we’re drinking $200 bottles of Dom on the second floor of an office we had built out to look like the Star Trek ride in Las Vegas.
In a few hours, we’ll descend upon The Bavarian, the traditional final stop for any Ensemble project—a restaurant we’ve terrorized so thoroughly that we must pretend to be with Texas Instruments when making the reservation. Which is to say, even in times of change, some things stay the same.
(This is a cut and paste of an article I wrote for the Spring 2010 issue of Gamesauce. Check them out here.)