Gameplay Director Patrick Fortier discusses the game design behind Deus Ex: Mankind Divided, including how the team created the characters, weapons, and AI that filled a plausible, but not necessarily realistic world.
For more on the Making of Deus Ex: Mankind Divided, enjoy the following articles:
- How Eidos-Montréal brought the Deus Ex DNA to 2016
- Player’s Choice in a Deus Ex-Game
- The Story Behind Eidos-Montréal’s Dawn Engine
One of the most interesting aspects of working on a game like Deus Ex is the opportunity to use the setting as a »window to the future«. This is the result of the game being set roughly fifteen years in the future, and being built around the core theme of transhumanism. As developers, we get to spend years exploring and discussing technological concepts which are slowly (or quickly in many cases) becoming a reality effectively reshaping the world around us.
Contemporary Reality, a double-edged Sword
One of the main reasons for choosing to frame the game this way came down to the emotional experience we were aiming to create for our audience. Pushing further ahead in time might provide greater creative freedom, but pushing our universe into true »sci-fi« territory would also mean giving something up. By providing players the ability to go about things the way they want, with real consequential freedom in a world that is very akin to their own, our goal is to create something which will resonate with them, which will impact them emotionally, and which may even help them to discover a little something about themselves.
Having players operate in a world that is a stone’s throw away from ours means, they are particularly sensitive to it. They can draw parallels more easily with their own reality. While this can make it easier for them to project themselves into our world, it can also lead to some headaches for us as developers, since it makes our audience much more demanding in terms of the overall »realism« of all the different aspects of the game.
People understand how the media manipulates information, they know how a city would react the day after a terrorist attack, they understand the complexity and corruptibility of important governing systems, they have an idea about what is humanly possible, and they understand what remains constant in a society, even as technological progress invariably pushes us forward. People still eat, they still drink coffee, they still gossip about the latest happenings, they still need downtime, etc. Regardless of its setting, any narrative-driven game needs to spend time establishing these things, framing its own particular »reality«. No matter the time period, the exercise remains roughly the same. You need to understand the realities of the world you are creating (its inhabitants, technology, major historical events, etc.) and deliberately expose it to your players.
The first element players will discover is the world itself. How things look, how people are dressed, how they behave, how they interact with the environment, and with one another. While maintaining the illusion of »realism« can be challenging, being able to draw from »reality« can be advantageous as well. You only need to look outside to find inspiration for common occupations. Drinking coffee, using vending machines, talking on the phone, or taking pictures are activities that are easy to imagine will still be part of our lives in the near future. However, before long, you have to make some important gameplay decisions that are ultimately not »realistic«, but which are essential to the overall flow and enjoyment of the game. For example, we chose to have a game where NPCs react to you having your weapon drawn out (they refuse to talk to you and comment on their discomfort), but we decided against having them panic and run away at the mere sight of it. Obviously, in real life, you would immediately dial 911 if you saw someone like Jensen walking around your neighborhood with exposed firearms. Actually firing a weapon, whether it hit a target or not, would simply cause complete pandemonium. Everyday life would stop immediately, police would shut down the area and begin an investigation. People would be talking about it days, or even weeks, later with their neighbors, and some would stop going to work for a few days.
Obviously, that level of realism would be majorly detrimental to the gameplay experience. Players would never be able to experiment – never-mind fail – at exploiting the different gameplay abilities. However, players do have expectations that there will be some level of consequence for such behavior. That is why we do trigger a reaction: people yell and hide, and try to alert the local authorities. You do feel that your presence and your actions are generating panic, and the police force reacts immediately by coming after you. After a while though, there is a cool-down period and life returns to normal. This approach allows players to remain immersed in the game world, since NPCs do react dramatically to their actions, but also carry on and recuperate. Players don’t need to reload a save to have NPCs go back to a normal state. In fact, it’s often something users will play around with intentionally to see how the game deals with the situation and its consequences. Since they are so punitive, players will generally choose to avoid gratuitous altercations like that afterwards.
While maintaining the illusion of »realism« can be challenging, being able to draw from »reality« can be advantageous as well.
Another element which is really important for the credibility of our world is to witness how the population is affected by the events and the overall themes of the game. Transhumanism does not simply translate into cool powerful characters like Jensen. It affects everyday lives in many ways, from the pressure to compete, to the desire to reshape yourself or heal a loved one. It’s important for players to see the far-reaching impact of the theme as they travel around the game world, along with hearing people talk about current events (which players may or may not have an influence on). To that end, numerous overheard conversations are placed all around the game world, newspaper articles are created to be adaptable to the players choices, and radio and TV shows are made to provide more background information, framing things from a perspective that goes beyond what players readily have access to. We also take the time to create content on computers and mobile devices, showcasing exchanges NPCs are having with one another about things affecting their lives and their futures. All of these elements come together to create a rich tapestry on which players can project their own thoughts and assumptions. It bridges the gap between the »game design« objectives of the game, and the »virtual world« portrayal of the setting.
Jensen, the Augmented Man 2.0
In Mankind Divided, it was of paramount importance to show the evolution of Adam Jensen as a character. Some of the naiveté he showed in Human Revolution had to give way to a strong personal motivation to get involved, a hunger to uncover the truth, and greater self-acceptance. Transhumanism being the core theme of our game, we wanted it to come across differently this time around. Adam is no longer just coming to grips with the idea of being augmented. He has had time to get used to the many augmentations inside his body, and he is no longer hesitant to take advantage of them for his own personal benefit. Gameplay-wise, we chose to translate this feeling of comfort first through the control scheme. With more augmentations to choose from and our main character being more comfortable using them, we wanted to create a more direct access to augmentations for players. We started by designing the quick selection wheels and mapping them to the control sticks. This allowed for instant, symmetrical access to both the inventory and the aug-menu. We also echoed this in the UI, with augmentations on the left-hand side of the screen, and weapons on the right-hand side.
Having to hold either the right or left thumbstick to make a selection meant our control scheme would be a bit more demanding than a typical FPS, but it also meant augmentations and equipment were always only one click away at any time during the game. Seeing as the game is less »action-frantic« than most, we felt comfortable enough to experiment with this system. Sure enough, it immediately became clear that it was putting augmentations front and center, while also feeding nicely into the overall fluidity and rhythm we associated with Jensen’s new »self-comfort«. We also decided to dedicate one of the shoulder buttons to the augmentations. Not only would doing this make a strong statement to players about the importance of augmentations, but it also allowed them to use their favorite augmentation smoothly and seamlessly – even while being in movement. Again, this made a tremendous difference once we started experimenting with it, and it became clear we had to embrace this direction.
We made a deliberate choice to show, through his interactions with Koller, how complicated it can be to have maintenance over these augmentations
In regards to the augmentations themselves, it’s also important to explain where they come from: these are not »magical powers«. The experience point system, while a useful game design tool, also has to be justified in relation to the themes of the game. Although we needed to »re-spec« Jensen in order to provide players with the freedom to design him in a way that matched their playstyle, we still felt the need to explain all this through the »logistical« issues related to having augmentations. Jensen is unique in that he does not require the neuropozyne drug that prevents aug-rejection from the body, but he is still subject to brain adaptation in order to fully control his mechanical augmentations. This is how we justify the experience points system. They are a visual feedback built into Jensen’s HUD to provide information about how well his brainwaves and his body are adapting to the presence of augmentations. We also made a deliberate choice to show, through his interactions with Koller, how complicated it can be to have maintenance over these augmentations. It’s not a matter of simply adding or removing augmentations, and there is clearly an element of danger tied to it all.
When actually designing the augmentations, we first looked at function over form. Like much of what we do, it started with a brainstorming session and many post-it notes. We were just throwing out ideas for abilities that would be cool to have. We then took that initial set and pruned its content down based on usefulness. For instance, seeing through walls is not just something cool to do, it can prove invaluable when attempting to play stealthily. Similarly, in a world with a lot of verticality, the ability to fall from any height without taking damage is also very useful. The question we were asking ourselves at that point was: »What would players want to do and why?« We were looking at how much value we could bring in terms of interesting decisions to make, while also ensuring that all four gameplay pillars (combat, stealth, hacking, social) were represented in some way.
Once the functions had been determined, we spent a lot of effort justifying how each augmentation could be explained from a technological standpoint. We could have left it vague, but we felt that grounding our augmentations in real existing concepts would help tremendously in fleshing out the world along with making it feel more credible. After all, since the beginning of the project, we’ve always said that the augmentations were the »stars of the show«. It was only natural that we took the time to really build them up and put some care into detailing them properly. We helped maintain plausibility by doing lots of research on where technology might be heading in the next 20 years. We also employed an external consultant to ensure that any science stuff would be as close to the mark as we could make it.
The main weapons found in the game are also contemporary, and need to reflect players’ expectations about relative strengths and weaknesses. At the same time, it was very important for us to give each weapon a »job«, which is not always the case if you base yourself purely on real-life weapon values. We wanted players to switch between weapons based on what they needed to do, each weapon being best suited for specific situations. Ultimately, we did this by taking each weapon’s attributes and elevating one or two of them, making them better than any other weapon’s. For example, there’s a weapon with better damage, one with a big clip-size, one with a high rate-of-fire, etc.
Once each weapon had been made unique in its own way, we balanced all other attributes accordingly. The goal was to counter-balance a weapon’s strengths with some weaknesses. For example, the weapon with the best damage is given a small clip-size, or a lower rate-of-fire. Looking at DPS (damage per second) is a way for us to compare weapons, done by aggregating several attributes into a single numerical value. This comparison is not perfect, since some attributes cannot be factored in, but when coupled with other values (such as accuracy), we get a good idea of where changes are warranted to achieve the desired balance. We validated our changes by having people play the game averaging out how much usage each weapon gets over the course of the game. If we see that one weapon is favored or ignored disproportionately, then we go in to make changes and test again.
NPCs and AI behaviors
Having a »realistic« setting means we primarily use »human beings« as our main NPCs. We need to make them feel like they have a real purpose in the universe, other than patrolling patiently in wait of the player’s arrival. They have to feel »alive« in their environment if we want to covey the narrative properly. Yet, they need to be predictable enough for players to readily identify patterns allowing them to exploit them to their advantage. This is always a bit of a quandary in game design,
and one that’s especially true for this game, since we essentially have two AI systems for players to interact with: one for combat and one for stealth. The challenges of designing for each are very different.
Stealth is all about detection. That was our main focus during the design process. AI can see, hear, and touch; that’s how it can detect the player’s presence. But from how far away? At which angle? Are attenuating factors, such as lighting, taken into account? Those are questions that can have far-reaching impacts on how players experience the game. When you think about it, the answers to these questions are invisible to players. It’s not like they see an NPC’s cone-of-vision and understand why they’ve been detected. On the contrary, vital information, such as the width and length of an NPC’s vision cone, is often obfuscated, which means players can only interpret what’s going on, but they never really know. Herein lies the challenge of designing for stealth. You can’t just go about setting your rules based solely on difficulty or realism. You also have to keep in mind how players will interpret what they’ll see, and whether they’ll think it’s fair or not. In fact, the more you succeed at effective and believable world building, the less players will actually see the NPCs as »detection machines«. They will start expecting regular »human« behavior, and will be disappointed if the NPCs are too mechanical.
Early on, we told ourselves that we wanted stealth to feel like a game of cat and mouse, where the player is the cat and NPCs are the mice. But for that to happen, players needed to feel in control – to understand the rules of stealth so they could act accordingly. However, since stealth can sometimes feel »opaque« in how it works, we knew we had to be careful in how we built our detection systems. A good example would be each NPC’s cone-of-vision. At first, we did what reality dictated: we paired it with the NPC’s head, so that the cone would move along with the head and »see« in the same direction as well. When the feature was tested with external players, the results were abysmal. Players kept getting detected without knowing why, which made stealth a very frustrating experience for them. After some analysis, we realized that players often misread where NPCs were actually looking, because the head was too small to see accurately at a distance. Following this, we attached the cone of vision to the torso instead. This made it easier to see from afar in which direction an NPC was pointed, and greatly improved how players fared.
The AI design for combat centered on cover. How do we get NPCs to use cover, and make them feel credible while doing it, all the while making sure they aren’t too lethal? Ultimately, it came down to a series of parameters, such as distance to player, accuracy, and cover availability. We wanted NPCs to move around, but not too much. By tweaking these parameters so they were in balance, we were able to have NPCs that moved from cover to cover, either to adjust their distance from with the goal of reaching the optimal range from the player for the gun they were currently using, or staying put if the conditions were right for them. The rest came down to hit points, damage, and how many NPCs could fire at the same time.
»True« Reality is Perception, not Facts
Ultimately, the real goal is to create a »plausible« world more than a »realistic« one. By being true to our own themes and the rules of our setting, by being consistent with our characters and their relationships with the world, and by being mindful not to add elements gratuitously, we manage to create a universe which feels tangible, but which is also interesting enough for players to want to lose themselves in. We continually strive for credibility, but rarely for realism. We give that leeway to ourselves when justifying our features in terms of how they would work in real-life. We accept that we can’t be faithful to reality one hundred percent. The game does take place in the future, and no one truly knows what technology will look like twenty years from now. However, by looking at what is possible now, and where prominent experts think we are headed, we can extrapolate the inner workings of an augmentation, and deliver a believable explanation of how it would work. The benchmark is: could this thing be real, soon? If we reach a point where the answer is »yes«, then we know we’ve done our job. If not, then we keep on digging until we get there.
About the author:
is Gameplay Director at Eidos-Montréal
As a child, Patrick already wanted to do Game Design before even knowing it could be an actual career! Over the past 19 years, he’s had a chance to work with fantastic people on a variety of projects such as Speed Devils, Myst, Splinter Cell, 1666 and currently with the Deus Ex: Mankind Divided team where he acts as Gameplay Director.