When Jim Brown, the Lead Level Designer at Epic Games, went to the Gears of War 3 beta forums to figure out what was wrong with the multiplayer map Trenches, it was in his words “one of the scariest things a developer can ever do.”
As a former professional gamer himself, Brown was intimately familiar with the levels of devotion and criticism that hardcore gamers embody, and so his trepidation at going into the beta forums to investigate the weaknesses of Trenches was entirely understandable. But what mattered to Brown was getting the map correct, and his act of humility paid off.
As part of the Level Design In A Day workshop at the Game Developers Conference this year, Brown gave a presentation called “The Legacy of Fail” which told the story of how the Gears of War 3 Trenches map was developed. His presentation was a breakdown of the iterative design process that fans don’t usually get to see. We asked Brown and Epic Games if they would allow us to share the presentation and the tale with our audience. The following images have been taken from Brown’s GDC 2012 presentation with his and the studio’s permission.
Where some level designers like to sketch things out on paper, Jim Brown prefers to jump into the game space as soon as possible. “You never really know for sure how something will turn out until you actually play it. It’s that in-game testing that allows for true iteration,” said Brown. And in his opinion, sometimes the translation from a paper design to an actual game build is one of the first places that level designer’s fail.
Flow is a key concept in map design. This is the way players move throughout the map during a match, and level designers try to encourage flow in certain directions and on certain paths in order to keep a match moving and to prevent play from getting stagnant. A staple flow pattern from first person shooters is an “O” design, which could also be turned into a figure-8 design if level designers wanted to make things a little more complicated.
When Epic Games began designing multiplayer maps for the Gears of War series they realized this classic flow pattern didn’t work. The classic pattern was about keeping players in constant, frenetic motion, but Gears was about cover, flanking, and lines of engagement.
Likewise, the random spawn points, which were common to the maps that used a classic “O” shape, also didn’t work for Gears, because flanking the enemy was such a key component of the gameplay. Players had to earn those flanking positions through teamwork and coordination. Random spawn points could mean that enemies would appear out of nowhere behind players and make the flanking positions irrelevant, so Epic realized that Gears of War multiplayer maps also required fixed spawn positions.
Both of these considerations evolved into the basic form of many Gears multiplayer maps, which are based on more of an “I” shape. The goal was to have positions where players could establish a line of engagement, but could also see obvious flanking routes and then fight for control over those routes. This became the basic design methodology behind Gears of War multiplayer map design.
When Epic set about designing the Gears 3 multiplayer map called Trenches, they started with a derivation of that basic design, an “H” shape with two facing lines of trenches and an open area in the middle. The theory for the map, which could be likened to a paper design, was that players would fight for control of powerful weapons, and then meet in the middle.
In practice, however, the center of the map was too open and the trenches were too shallow, so players would avoid the middle altogether and fight along the sides of the map, and this created a circular flow pattern. While this early map design would eventually inspire the Dreadlands portion of the Gears of War 3 campaign, as a multiplayer map it wasn’t working. Brown and his team had to scrap the design and try something completely different. Sometimes level designers don’t just iterate. Sometimes they have to admit failure altogether and start over again.
Trenches Version 2 made the trenches deeper and the middle front narrower, but it was an over-correction. The trenches were so deep and the middle so narrow that players never saw each other from far enough away to set up lines of engagement. That meant no taking cover or flanking maneuvers. People were still running in circles, and there was no teamwork.
The next iteration of the map, Trenches Version 3, provided more cover right out of the spawn points and good lines of sight so that people could get set up and work together, but there were no flanking routes. Players would spawn, dig in, and just sit there, or leave cover and start circle-strafing each other. Trenches Version 4 should look very familiar to Gears of War 3 multiplayer fans, because it was very close to what Epic shipped with the game, but it still had an “O” shaped flow pattern.
So the first order of business was to get rid of that and turn it into a “U” shape that played more into the Gears map design methodology of drawing players into a line of engagement. By making a series of changes Brown and his team changed the map to funnel players into the desired flow pattern. Instead of moving between large failures, the team was making small tweaks to the map to get it into shape.
One of the tools multiplayer level designers uses to gather data while they test levels is called a heatmap. Heatmaps can display where players move, where they die, and when they commit actions. Level designers can then look at those heatmaps to determine whether the flow is operating the way they want, to make sure players are using the portions of the map the designers want to funnel them into, and other measurements for whether the level is functioning properly.
Whenever the player commits the action that the developer is tracking, the map gets colored in on the spot where the player is standing. Blue and green areas mean less data, and red areas mean a lot of data. For example, Brown noticed that in playtests of Trenches there was slightly more movement on the right side of the map, which you can see on this heatmap which charted player movement.
There was a lot of movement near the player spawns – which are in the upper-left and upper-right-hand corners of these images – but people were fighting and dying in the center of the map, at the lines of engagement that Brown and his team wanted to establish. The heatmap below shows death locations.
Things still looked okay to Brown. He figured that people were spending more time moving on the right side of the map because it’s easier to track targets and aim looking over your right shoulder in Gears than over your left shoulder, as the following image demonstrates.
As playtests continued the data became erratic. Brown saw that movement was sometimes all over the map, which meant that lines of engagement weren’t forming. It also meant that deaths were all over the map. And in some playtests players would get trapped in their spawn points.
The heatmap below shows the death locations of a playtest in which players were trapped in the spawn on the left. The fact that it took place on the left side of the map debunked Brown’s theory that players were favoring movement on the right due to the camera positioning. It wasn’t something wrong with the camera. It was something wrong with the map.
Epic Games uses two kinds of playtesting. General playtesting is open to everyone at the company and therefore pulls in a good mix of skill levels and play styles. Then there’s a form of playtesting that attempts to mimic the behavior of players who will master the maps. Epic’s pet name for this hardcore form of playtesting is HITS, or Hot In The Street testing.
HITS testing looks at groups of players who know how to operate as a team and get to know the maps really well. Epic wants to know how the hardcore players will be using their maps six months after the game has been shipped, when it’s “hot in the streets.” And the HITS playtesting for Trenches showed that while players could get trapped in the spawns on either side of the map, coordinated teams of players could break out of that trap.
At this point the public multiplayer beta for Gears of War 3 was looming, so Brown and his team added some additional, defensible positions in the spawn points to give the defenders a height advantage over the attackers, and also added some power weapon spawns to give teams the tools they needed to break combat back out into the middle of the map.
These were just theories. Epic didn’t have the time for a full iterative pass, and they were still getting used to some of their data-gathering tools and didn’t see the problem right in front of their eyes. But there was another, more important error that Jim Brown admitted to unequivocally at GDC.
“We failed at our jobs. We let our egos get the better of us, and thought we knew better than the players,” Brown said. “We said, ‘Sure, people will get spawn trapped occasionally, but it’s not that big of a deal, is it? A good team can get out, they just need to know what they’re doing.”
Trenches was the least played map in the Gears 3 multiplayer beta. Players were regularly trapped in their spawn points. Yes, it was possible for players to break out of the spawn trap, but only if they had enough experience to know how to use the right weapons for the job - And only if they were organized… And only if they were communicating properly… And only if they had enough respawns left.
Brown and his team hadn’t accounted enough for the average player. What they had thought was a minor problem, the spawn trapping was actually a level-breaking problem. It wasn’t a problem that Brown could expect the players to overcome by using the right tools and being organized and communicating. It was a problem the level designer was responsible to fix.
“Even if it was easy for them to overcome, would anyone really want to play a match where they were pinned into their spawn and repeatedly killed?” Brown said. “That’s what we call a shelf moment – when the gamer puts the game back on their shelf and never picks it up again.”
Brown couldn’t surrender the challenge of fixing Trenches. It had become a personal mission, and so he decided to read the Gears of War 3 beta forums and the feedback from the fans. When Brown was a competitive gamer he would analyze weapons, maps, systems, and details to get every competitive advantage possible. The rabid fans on the Gears 3 beta forums did what Brown had used to do. They had broken the map down to figure it out. Brown played games online with these hardcore fans and learned from them, and determined how he could fix Trenches.
He moved the spawn points out of their dead-end positions, which made it easier for players to find defensible locations right out of their spawns. He flattened out walls to remove ambush locations that players could camp at and use to kill opponents coming out of the spawns. He even created a secondary path leading out the back of the left spawn to give players another way out.
Making changes this big to a map so close to the shipping of a title was unheard of at Epic Games. Whenever a developer makes a change, they introduce potential instabilities, and Brown didn’t have much time to test these changes. But the attempt to design a map that was meant to inspire Gears of War 3 multiplayer the way it was “meant to be played” resulted in a broken map, and that wasn’t acceptable.
While there were still minor changes and imbalances on Trenches that were addressed via updates once Gears 3 had been released, Brown and his team were able to turn the map around through all of these tweaks. Trenches spent time bouncing around the top three positions in terms of popularity out of all the maps that were shipped with the game. The spawn camping issue disappeared, and the heatmaps showed player movement and kills spread across the map fairly evenly.
And the irony was that in the end, if you look at the final design of Trenches, the play falls back into the “O” shape that Brown had been trying to avoid in the first place. “I learned that no matter how we wanted people to play our game, we had to design toward how the players wanted to play our game.”
Dennis Scimeca is a freelance writer from Boston, MA. His First Person column runs on The Escapist, and you can enjoy his excitations on Twitter: