Endless Space: When Triple-A Developers Go IndieBy Jeff Spock
From its start in a fourth floor apartment (no elevator) in Paris in early 2011 to its final release on Steam on July 4th 2012, Endless Space was both a work of passion and an unexpected adventure for our small team.
The entity now known as Amplitude Studios was the brainchild of Mathieu Girard and Romain de Waubert. Unhappy at the existing crop of space-based 4X games and desiring more creative freedom than they had as executive producers at Ubisoft, the two of them decided to strike out on their own. As Romain later said, he wanted to "...fill bookshelves at home with all the games I wanted to play that were never developed." The only way to do that, they agreed, was to go it alone.
When we first got together, we used "Star Empire" as the working title. Star Empire would be a space-based 4X game, using some of the classic mechanics from games like Civilization and Master of Orion. The 4X genre might be a strange niche to start with, but we felt that it would play to our strengths and minimize our weaknesses.
4X games are challenging to design, but we had an experienced team and believed that this would not be a problem. In addition, the genre also has lower requirements for production of expensive elements like animation and art, and budget was definitely a concern.
We also had to think about what we could do to make the game stand out in an increasingly crowded market:
In order to maximize the reach and potential customer base, the game needed to be PC and Mac-based, with ideally the possibility of being able to port it elsewhere. We looked at a number of game engines (Gamebryo, Unreal, open source tools, etc.) and ended up choosing Unity for a few reasons:
- Multi-platform (PC, Mac, iOS, browser)
- Free trial copy, permitting a quick prototype of the game, upgradeable to a Pro version, which included high-end game engine features
- Level tools adapted to an open sandbox game (unlike e.g. Unreal)
- Ease of use of the C# coding language
- Instantaneous compiling for test and workflow
We felt that most space-based games had mediocre graphics. Stars, spaceships, aliens, and galaxies should be way cooler and more modern than the simplified, comic strip look of many previous games.
Following on to that, the influence of movies and series like Star Wars and Battlestar Galactica radically changed the idea of what was "cool" in a space battle. There was a strong desire to give that sort of powerful, cinematic feel to the battles.
Finally, the backgrounds and universes of most science fiction-based games often seem tacked-on and poorly developed. For Star Empire, we wanted a universe that felt deep, real, and consistent; a universe that had a past and a future in which Star Empire portrayed just one particular era.
The name Star Empire lasted almost through the end of the project; even though it was a working title it was devilishly difficult to agree on something else. The naming process consisted of Max and me in a room, each of us coming up with names that the other one didn't like. Then we'd take our best ones to the team, and they wouldn't like them either. The process at least had the advantage of being predictable. In fact, it was only when the first article went to press (an interview we did in March 2012) that we finally managed to agree on "Endless Space."
But that was far in the future. To make the game, get the team together, and find the money, the first thing we had to do was found a studio.
"Hive Mind" was the original name for the studio, based on the feeling that it indicated a lot of smart brains working toward a common goal. We decided to change it, and after an insanely long and complicated and time-consuming series of arguments and discussions we narrowed it down to a shortlist. This list was problematic because most of the names had both strong supporters and strong detractors -- whatever was chosen, somebody would be unhappy. In the end the team went with Amplitude, because it seemed to suggest most of the key values:
...and, besides, no one hated it.
The next step was to figure out who, and how many, and for how long, and all those other questions that involve talent, cash flow and financing. Mathieu and Romain shanghaied a friend who would become the unofficial CFO, and together they developed the first business plans.
The initial plan was for €1.5 million in development costs. After a few cold showers in front of banks and business angels, the number was brought down below €1 million. This required cutting a few things that hurt -- intro and endgame cinematics, the number of ships per faction, and the number of types of ships. But we knew that we would have to let some things go if we wanted a realistic -- and profitable -- project, and we figured that we could still maintain the core gameplay mechanics and the game's depth. While the project and goal were ambitious, recent changes in the game industry meant that this kind of project was very realistic. In particular, the fact we could avoid selling through the traditional retail channels.
In retail, a game studio would be lucky to receive 20 percent of the retail price of a game after costs (publisher, manufacturer, distributor etc.) were taken out. With online storefronts, that number is closer to 50 percent. In other words, it had become 2 1/2 times easier to hit break-even. Even so, given the world economic situation it was a worrying time to be going around asking for money. What came through in the end was a large round of financing from friends and family. While it is not much fun, administratively speaking, to manage more than forty shareholders under French business law, we all figured that the extra hassle was worth the price of trying to fulfill a dream.
Amplitude was also fortunate to be contacted by a number of distributors and publishers who were interested in the team and the project. Late in the dev cycle we ended up partnering with Iceberg Interactive, a Dutch company that was a good fit internationally as well as being familiar with strategy games and that market niche. Iceberg was happy to handle retail and non-Steam online sales, freeing our team to focus on development and the Steam relationship.
But of course, it takes more than money to build great games; you also need a team.
Before we get to that, though, here's how our financing broke down:
Financing model pros and cons:
- Find a publisher
PRO: Someone else carries financial risk.
CON: Loss of IP control, loss of creative control, loss of schedule control
PRO: Trendy, fast, wide-reaching
CON: 'One-off' with the community/fans, project stops if you don't hit the target, no one on the team was a rock star
PRO: Control, independence, long-term relationships with fans
CON: Someone has to find the money...
Final Decision: Self-financed
- Final Funding Sources:
- Team: 16 percent
- Friends and Family: 49 percent
- Private/Gov′t support: 26 percent
- Loans: 9 percent
Building the Core Team
In parallel with the funding, the most critical element of any creative enterprise is to get great people. We put together a core team of leads (who would also be investors) for each discipline. Breaking down by role, they were:
- Their finance whiz, to act as CFO and help plan budgets
- Romain de Waubert, COO and creative director
- Corinne Billon, first as a consultant, then as an employee, to be the art director
- Arnaud Barbier, lead technical artist, who left and was replaced by Mathias Grégoire
- Me, to be the writer/narrative director
- Mathieu Girard, CEO and producer
- Eric Audinet, an experienced game coder, as lead programmer
In addition to the core team -- with an average of eight years of experience in the industry -- the decision was made to bring in relatively inexperienced but very high quality "juniors" to fill the other roles.
Programmers, artists, designers, and an associate producer who all had from zero to two years of experience in the industry were hired. The theory -- validated by the final product quality and aggregate reviews -- is that the leads had the knowledge and abilities to bring the beginners along and get the best out of them.
From the start -- and even before -- we were convinced that game development in the 21st century had to be done in close cooperation and coordination with the fans and the community. While engaging the community may seem like a "no-brainer", there is an enormous difference between setting up moderated public forums and actually tracking, responding to, and prioritizing the community's suggestions in a way that is open and transparent.
For that, Amplitude made another key acquisition: Max von Knorring. Max was an up-and-coming marketing whiz at Ubisoft who we seduced into trading a brilliant career path for poverty, uncertainty, and stress. In compensation, we gave him the title of "director." Max ended up heading marketing and communications and developed the entire community interface concept. His brainchild was that interested members should be able to watch, track, and support the development process to the point of voting on features and commenting on game design. Known as "G2G" (GAMES2GETHER), the community interface had two guiding principles:
- Publish the game design documents so the community could immediately see and comment on the ideas, plans, and directions of the studio. This would help provide direct access to the dev team and start a communication loop between the developers and the players.
- For those less interested in detail or lacking in time, there would be a series of regular votes on the forums so that they could still be heard and get engaged in the game development.
And did it work? Well, let's put it this way: Max had to create a special award, Space Ninja, for the three hardcores who figured out the web address and signed into the forums before they were open and announced (Alderbranch, Sharidann and Znork, are you reading this?).
While we were still in alpha, we had such astronomical growth on the forum that we had to bring someone in full time to help manage the community. During the beta phase of the development cycle the forums had over 20,000 members, with hundreds on-line at any time, and we broke the 30,000 barrier just before launch. Even better, a number of particularly engaged fans and modders, excited by the game and G2G, volunteered to help moderate the forums. Enormous thanks to:
One of the first decisions that the core team had to take was to select a development methodology. In that fourth-floor apartment in Paris, the team had a number of discussions on waterfall versus RAD with bits of Scrum and maybe iterative/Agile or whatever, and there was a lot of back and forth about what the opportunities and the challenges of the different methods were.
In the end, the team decided to go with Scrum, "sprints," and "user stories" primarily because it was iterative; testing and building and changing the design mid-stream seemed like something that would be critical to keeping the project manageable. That was also a good match with Unity, as the engine permits updating and prototyping on the fly.
We covered one wall of the studio with a white board that we used to show progress by discipline. Our Associate Producer, Laurent Lemoine, had the dubious pleasure of leading the daily 10 AM Scrum meetings where everyone explained what they did yesterday, presented what they planned to do that day, and exchanged ideas and offers of help. Mathieu and Romain acted as the "customers" for the Scrum, creating an enormous spreadsheet of the user stories that were slowly whittled down, week-by-week, over the course of a year.
Without making things sound too simple, after all of the preparation work had been done it was more or less a matter of turning the crank and getting the job done. There was of course a great deal of give and take during the Scrum sessions and the technical planning meetings; it was complicated at first to get to grips with the project management style and the way estimations were made and tasks were tracked. However, the small physical size of the studio (900 sq. ft. for 12 people) and the readiness of the leads to dive into the details helped enormously in creating a driven but supportive studio culture.
Of course, free Coke and espresso might have helped that.
From the first round of discussions with the core team in early 2011 to the final launch date, roughly 18 months passed. In those 18 months, Amplitude generated:
- 200,000 lines of C# code
- 273 MB of design documents
- 4.4 GB of textures and images
- 1.6 million polygons
- 2,600 sodas
- 12,000 espressos
- 4,500 cookies
But the first public, critical milestone occurred on May 5, 2012, when the alpha went on sale. This was in and of itself a coup for the studio; it's unusual that a new independent studio is allowed to roll out an alpha version under the glowering eyes of Valve.
However, as several team members had worked with Valve and on Steam -- and in fact on Dark Messiah, the first non-Valve game ever to use Steamworks for multiplayer -- the good folks in Seattle decided to give us a chance. However programmers never get a break; as Unity is in C# and Steamworks is in C++, Eric's team had a lot of work to do to get the back ends of the two packages talking to each other.
The alpha launch went as well as we might have hoped. It helped, of course, that using the Scrum process meant that we had been testing and iterating the game, playing on stable (if limited) versions for weeks before the alpha date. Had the alpha version been full of bugs, the story would have been entirely different. As it was, the alpha and beta of the game garnered glowing comments for their stability and reliability.
We saw a number of development hypotheses validated:
- The market was waiting for a good 4X space game
- The ships were good looking, and the lore of the universe was interesting
- The gameplay mechanics were solid; reassuringly familiar and cleanly executed
- The game was surprisingly stable and bug-free
And some surprise comments:
- The GUI got lots of praise, with some saying it was the best ever in a strategy game
- The battle system was truly love-hate; some wanted tactical battles rather than ES's cinematic ones.
For the release, we built a standard "Admiral" edition ($29.99) that had a few extras, plus the "Emperor" special edition ($34.99), which added more extras and included a new faction with re-skinned ships (The color of the new skins, incidentally, was chosen in a G2G vote). The alpha only had four of the eight factions and was released at a 25 percent discount to the list price. The beta version was content-complete, and sold at only 10 percent off the list price.
Some interesting stats are:
- In Alpha, players purchased more than twice as many Emperor versions as Admiral versions.
- In Beta, players purchased them in roughly equal quantities.
- In the Final version, players purchased about 20 percent more Admiral than Emperor.
During May and June studio activity hit a peak; in May it was making sure the beta version would be content complete and in June it was balancing the beta's AI and fixing bugs. That was the only real crunch period that the studio had to handle. Otherwise, the pace was intense but rarely did the developers have to work outside the unofficially official hours of 10-ish to 6 or 7-ish.
On the down side, we had some bad luck with release dates. The alpha came out ten days before Diablo III, and the final version just one week before Steam's Summer Sale. We had some worried days and sleepless nights, wondering if the announcement of our baby's birth would get lost amid the noise of the elephants trumpeting.
In both cases, however, we didn't really have other options. A self-financed studio has a development plan and a bank account; if the game does not launch and start bringing in cash the studio will go out of business. Early July ended up being the date that made the most financial sense, regardless of the risks due to Steam's annual sale.
Community reaction to the launch was generally positive, but some members felt that the game needed more time in beta to smooth the rough edges and, in particular, improve the balance of the AI. We didn't disagree with them; the launch was version 1.0 and as of publication, we're on 1.0.30 and are still tweaking the AI. But as explained previously, we didn't have a whole lot of options. We needed to generate cash, and even though we had very encouraging sales during the alpha and beta releases, we weren't earning enough to keep the lights burning for several more months of dev work.
Overall, we are ecstatic with how great the community has been and are very happy with our sales. The sales arc has been fairly predictable; three weeks after launch we had sold a total of 110,000 copies (all versions and all editions), and a month later 150,000. We are now working on free add-ons in 2012 and probably a paying expansion in 2013. The first add-on, Rise of the Automatons, was launched October 1st and included a new community-created and G2G-voted faction. We have started asking the community what they would like to see in the expasion.
Lessons and Next Steps
While we won't say that the game is a resounding success, it has at least sold over 150,000 copies and is breaking even. A lot happened in a relatively short time, and after polling the various members of the team the following elements are what stood out as Key Success Factors (or, depending on where the sales end up, Key Not-Total-Failure Factors):
- Experienced core team combined with high quality hires
- Collegial, friendly atmosphere leading to efficient workflow and teaming
- Scrum method -- transparency, communication, iteration
- Close, structured, and serious integration with the community
- Quality and stability of releases
What we might have done differently:
- Leave more time for coding and debugging multiplayer
- Leave more time for working out gameplay balance issues
- More clearly define some Scrum elements: sprint length, task approval, time for preparation / definition of a new sprint, better integration with the bug-tracking system
- Be clearer, more responsive, and more demanding with external suppliers
- Start the press and publicity efforts sooner, and anticipate predictable questions (indie or triple-A? Why no tactical battles?)
So what will we do next? Primarily, there is a serious desire to continue developing and evolving Endless Space. As Mathieu put it on the forums, "...this release is just a step in the life cycle of Endless Space, this is not end. We are committed to constantly improving and expanding the game for many months after release." So a dedicated team will stay 100 percent on Endless Space, and G2G votes will continue as the team and community decide what new features and gameplay needs to be added. We have the extension to work on, and will probably eventually start thinking about a second IP.
We just hope that close contact with the community, a passion for great games, and the fun of working with a great team will ensure a long and entertaining future for all of us at Amplitude.
Return to the full version of this article
Copyright © UBM Tech, All rights reserved