This postmortem by the lead designer and technical art director of Call of Duty 4: Modern Warfare first appeared in the March 2008 issue of Game Developer magazine.
Call of Duty 4 was Infinity Ward's third Call of Duty game, and as such we approached it knowing we needed to do something fresh.
We don’t want to pigeonhole ourselves any more than we have to, and many members of the team came off Call of Duty 2 promising never to work on another WWII game.
We tried several different directions, many of which were failures, but the ultimate result was the best game any of us have ever worked on. As a game development experience, it seemed to go so smoothly that it was difficult to come up with five things that went wrong.
Coming off Call of Duty 2, we knew we wanted to do something different for our next game. We don’t agree with some critics who say that WWII as a genre is dead, but we couldn’t muster the same passion for the subject that we had in our first three WWII games (Call of Duty 1 and 2 and Medal of Honor: Allied Assault). We had a few ideas that we wanted to do and eventually settled on two. One was Modern Warfare, and the other was a new project.
Modern-day warfare is very emotional for people, which is both good and bad. We really wanted to avoid referencing any current, real wars, and one aspect of the gameplay that we really didn’t want to change from previous titles was the idea of two large opposing forces with similar numbers and technology. To facilitate that, we invented a war with several fronts, primarily involving a group splintered from the Russian army, with a secondary front in the Middle East.
"Many members of the team came off Call of Duty 2 promising never to work on another WWII game."
The modern setting inspires an enormous amount of gameplay variety. Modern warfare is very different from more traditional warfare in that direct confrontations between huge armies are relatively rare. Instead, you have a huge variety of different types of low-intensity conflicts and special forces missions. Because we already had a very sophisticated scripting language in our engine, we were able to implement and iterate on that variety quickly, and take advantage of the modern setting to shake up the gameplay, but still deliver a polished result. Modern weapons and tech are something that people like to see and play with. Kids the world over grow up fantasizing about being a soldier, and we aimed to let adults live out their childhood fantasies (Call of Duty 4 is rated M). But we also knew we wanted to keep that signature Call of Duty grittiness and avoid making the game feel too techy. One thing that helped us there was focusing the U.S. part of the game on Marines, who get a lot of their equipment second-hand from the Army.
By moving away from history and into the current day, we were able to do much more useful reference gathering. For example, the effect that happens when you are near an Abrams tank when it fires was inspired by our designers, artists, and sound designers experiences at a live-fire exercise at 29 Palms, which is a Marine training facility in the California desert. We were able to talk to real marines only weeks out of combat to get a feel for the background, emotions and attitude of soldiers in combat, and we had vets supervising our mocap and AI design to make sure our tactics were sound.
At the beginning of Call of Duty 4, looking at what we had done with Call of Duty 2, we saw two main areas we needed to focus on improving. First, by dedicating more development time to multiplayer, we felt we could make some really big improvements. Second, we knew we needed to tell a story.
Call of Duty 4 is our first game where we had a team working on multiplayer for the entire project. The quality bar for single-player first person shooters is really high right now, but there were and still are a lot of things that no one has really tried to do with multiplayer, and based on the success of the multiplayer in our previous games, we thought we could really impress.
With a seasoned lead and some dedicated designers and programmers, Modern Warfare multiplayer was much more ambitious, much more polished, and generally much better than ever before.
Story is something we’ve always put a little effort into, but by and large we’ve prioritized it below other aspects of our games. Moving away from WWII and into a fictional war removed that option. We spent hours brainstorming with military advisors, trying to come up with a credible scenario that would involve a large-scale war, and then weeks interviewing writers trying to find someone who could help us craft a narrative that would draw the player in. The result, while not Shakespearean, has drawn almost universal praise. We feel like we have a new skill, and we intend to build on it in our future projects.
Low turnover is Infinity Ward’s secret weapon. You can throw all the money, top talent, outsourcing, mocap, and high-end middleware you want at a project, but without a team that knows how to work together, you’ll only end up with delays and a fragmented product. We still have 20 of the 27 or so developers who worked on Allied Assault six years ago, so our team has a remarkably stable base.
Before we started on Call of Duty 4, we spent two years developing and using our new engine on PC and Xbox 360, and two years before those using parts of it (notably the scripting system and the level editor) on Call of Duty 1 on PC. By retaining almost all of our people, we retained almost all of that experience. We were able to leap right into development on Modern Warfare, improving the engine, creating art assets and building levels immediately. Our leads were able to work directly on game content rather than spend all their time wrangling a team of new people.
We were very lucky (or smart, depending on who you ask) to have a team that had built a game on this generation of hardware before. Our engine already worked on the 360 and PC, and we already had 360 dev kits and tools that worked on them. We knew what performance to expect and had a good idea of how to optimize our assets for the hardware. The PlayStation 3 and Xbox 360 have similar graphical capabilities, which meant that almost all the experience our artists had on the 360 on the last project was directly applicable to both consoles this project. While PCs are always changing, they didn’t change so much in the two years since Call of Duty 2 that we had to relearn anything—we just made improvements to our technique.
Relating to the previous point, we didn’t need to hire many new people. We had staffed up dramatically for our first “next gen” game, but starting Call of Duty 4 we already had the team in place, and they already knew how to use our engine and tools.
By building on what we already had, we were able to reach higher in two years than if we’d been forced to start from scratch. This goes not only for engine features, but also for tools and content too. By the end of the project we had replaced almost all the content carried over from Call of Duty 2, but having it available during development removed bottlenecks and allowed us to work faster.
From the point of view of our artists and designers, the PlayStation 3 and Xbox 360 have very similar abilities, so we could share assets between them. PC is a broad target, and at this stage in the console lifecycle, the consoles still have similar performance to mid-high end PCs, so we could share assets there, too. Of the thousands of assets in the game, only a few dozen are platform specific.
Working with an engine that already runs on all the target platforms—and keeping it running on all those platforms during development—is far easier than trying to port it near the end. While the end result on the two consoles is almost identical, the innards are dramatically different in many cases. On the PC, the differences are obvious. Depending on the specs of the PC or the user’s choice, the game can run at one of dozens of different resolutions, with different texture and model detail settings, different control schemes, and different graphics hardware and drivers. Adding the functionality to allow the user to choose between those settings at the end of the project, while simultaneously trying to finalize the game on other platforms, would have been impossible.
"As a game development experience, it seemed to go so smoothly that it was difficult to come up with five things that went wrong."
Naturally enough, early in the development of Call of Duty 4 we were trying to take what worked for World War II and move it into the modern era. So we were trying to adapt popular movies and well-known battles from the modern time period and turn those into missions. We didn’t have any Russian missions for the first year of our two year development. Our enemies were, like you see on TV, poorly trained and equipped Middle Eastern soldiers and militias.
In fairness, looking to real life for direct inspiration worked at first and produced some great missions. The missions that survived this period of the game design include the AC-130 mission, Aftermath, Charlie Don’t Surf, and The Bog. Almost all the desert environment multiplayer levels also came from this period. The problem with the adaptation approach was that modern battles tend to be very lopsided and everything we saw was in desert environments. We needed battles where the opposing forces were well-trained and equipped, and we needed more settings.
Eventually we decided to go back to the drawing board and change the high concept for the game. It’s worth mentioning that at this point in time our full team had been moved back onto Call of Duty 4 (see “Distracted by Second Project” below) and we had also found a writer who we felt we could collaborate with successfully. We finally had the focus and the skills to build a fictional scenario that would enable us to take the game anywhere we wanted. We decided to add the British SAS characters and a second plot line about a Russian civil war. We stopped all level building and scripting, cut a bunch of levels, and started designing missions again.
This design “reboot” is where we came up with the ghillie suit missions, the stealth missions, and all the rest of the missions set in Russia. Ultimately, this reboot was a good thing for the game, but being so late into development it did slow us down for a while.
Early in alpha we learned that we would have to finalize the console versions early because it takes longer than we expected to make Blu-ray disks. This meant that our time for beta testing was shorter than planned. This also meant that some missions didn’t get as much balance testing as they needed and ended up being more difficult than they were supposed to be, which is one of the most common complaints we’ve received about the game.
Our pre-release buzz was stronger than it had ever been for any of our previous games; we were getting tons of press despite it being a very crowded holiday season for games; our trailers and other videos on the internet were getting amazing numbers of viewers. Despite all this, we did as we had done with all our previous games—about a month before release we put out a single player demo on PC consisting of one of our missions.
The reaction to the demo completely blindsided us. Our fans were disappointed. The demo was “more of the same,” or even worse, just “meh”—not even worth talking about. After a couple of days we realized what went wrong. Anticipation was so high that we couldn’t possibly live up to expectations.
Also a huge part of the appeal of our single player game is the gameplay variety. Playing Call of Duty 4, you almost never do the same thing twice. That makes it impossible to select just one mission to represent the entire game. Instead we had to choose what part of the game to represent with the demo. If we had chosen one of our radically different missions we would have alienated fans of the previous games, so we chose a level that we felt represented our “core gameplay,” which is fairly similar to the core gameplay of Call of Duty 2.
Lastly we had to worry about story spoilers, as most of our favorite missions also advance the story. Giving away one of those as a demo mission was out of the question, as we didn’t want to wreck the game for players.
Given all these constraints, looking at the examples of other games which managed to build tons of pre-release buzz like Gears of War and Halo 2 and 3 without doing pre-release demos, we should have realized that a pre-release demo would be likely to hurt us rather than help us.
In hindsight, the PC demo was a distinctly different case from the Xbox 360 multiplayer beta. The beta was released earlier and was responsible for much of our buzz, it was much more novel on the Xbox, where gamers are not as used to free content, and it showcased a large amount of what was new in Call of Duty 4. The beta also played a vital role in helping us ship a polished game.
At the start of development on Call of Duty 4 we tried to branch into two teams. We started a second project with a small prototype team, intending on shipping it a year after Call of Duty 4. Our intentions were to create a new risky IP, which would allow us to stretch our creative muscles. We are determined not to stagnate creatively and just make clones of our previous games indefinitely. Growing a second team was one idea for how to do achieve this.
Almost immediately, the two projects began to compete with each other for ideas and people. We hired extra people, including some seasoned leads, so that neither project would be understaffed. As time went by, we were aware of the difficulties, but we initially focused on how hard it was for the team on the new game, failing to notice the damage that the second project was doing to Call of Duty 4.
The area hit hardest was the game design. Our design leadership was distracted by the second project and put a lot of their creative energy into it. This meant that problems like “too much desert” were allowed to linger for longer than they should have.
Our second project was abandoned at the end of May 2006, allowing the entire team to focus on Call of Duty 4. Naturally there were a lot of different reasons, but ultimately it came down to the realization that what made our previous games so strong was the chemistry of our team. Splitting the team into two parts broke that chemistry and both projects suffered because of it.
Our design workflow is very iterative. When members of the press visit us to see early versions of our games, they are always surprised to find that our games are fully playable, with what appear to be levels that are ready to ship, up to a year before the game is scheduled to be released. This is necessary for us to be able to iterate and throw out as much work as we do and still ship our games on time. This level of iteration applies not just to the gameplay and look of our levels, but also to the story and dialog. In this game, that led to some big things not getting done until the last minute.
We knew we wanted high-tech looking movies between our levels to cover our load times and help tell our story. We didn’t know who to get to create them or if we should try to make them ourselves. We didn’t finish writing the game’s story until very late, so it wasn’t until well into alpha before we knew with confidence what we wanted these movies to be.
At the last minute, we contacted the producers of Discovery Channel’s FutureWeapons, and signed a contract with Spov, which did the title animation for that TV show. Spov was able to deliver the movies in a very short time—something like two months from agreement on terms to final delivery. The movies turned out really good and everyone on the team is really happy with them, but the late schedule meant that these movies were some of the absolutely last things to go into the game. Had they not been good, there wouldn't have been any time to fix them. Another scary schedule issue was with Captain Price’s dialog. (Captain Price is easily the most important character in the game, with more mission-critical dialog than anyone else.) We always record dialog very late to prevent having to redo it all when we change the story and missions. After finishing most of the dialog recording for the game, we decided that we needed a different voice actor for Price. So, like the movies, Captain Price’s dialog was some of the absolutely last assets to go into the game. His dialog turned out great, but had it not, there would have been no time to fix it. Unfortunately, Price’s facial animation suffered as a result of his dialog being completed so late.
Because of our process, outsourcing is hard for us. We did outsource a small amount of art for Call of Duty 4, but because of our design iteration, a large amount of it was not useful by the time we received it. We delayed work on outsourcing in order to be more confident in the assets we were requesting, which meant that we received final versions of many assets right around alpha. The last assets we received were too late to use because we had locked the game tree. Even some of the earlier assets were too late to actually use because most of the missions had already maxed out their memory budgets. Other assets were requested for parts of the game that we cut while the assets were being built.
Call of Duty 4 feels like a watershed moment in the history of Infinity Ward. We are all tremendously passionate about our games—all the choices we make in our design, even the ones that people complain about, we make because we feel they make the games more fun. Up until now we’ve always felt like we were underdogs, with each game fighting for recognition as one of the best in the genre. This may sound conceited, or stupidly modest, depending on your outlook, but that was the vibe around our office. With Call of Duty 4, we suddenly feel like we’ve done it—we’ve produced a game that everyone loves. Coming out in a year with so many other great games, and still being counted among the best, is an amazing experience—one we will have to work very hard to surpass next time around.