Would the Holodeck be any fun? 5. Agency and Affordances

To make a game immersive, we don’t need total realism, but would be better off aiming for the centre of the realism-gamism spectrum. Doing so will be easier with stylised mechanics and process intensity. An additional benefit of process intensity is its impact on agency.

We perceive not just images, but possibilities for action, and in the spirit of play we feel the impulse to try acting on those possibilities whenever we find ourselves in a novel environment (Goodrow, 2013; Tornqvist, 2014; White, 1959). The reason why the interaction component is just as important as the visual component of immersion, perhaps even more important, is because when we perceive an object we see the capacity to act upon, use, and interact with it. In the field of ecological perception in psychology, these possibilities for action are called perceived affordances (see eg Dreyfus, 1996; Gibson, 1979; Jenkins, 2008; McGrenere & Ho, 2000; Norman, 1998, 1999; Rambusch & Susi, 2008, Toups et al., 2014).

Giving function to fiction empowers players to act upon their playful impulses. Process intensive designs that include emergent gameplay will allow players to combine interactions of multiple objects, creating arbitrarily complex sequences of action. For example when a player sees a cannon and remembers that he could pick up that rabbit he saw before, he thinks it would probably fit down the barrel, and discovers to his delight that it does. This is how combining process intensity with giving function to fiction overcomes the problem of the world feeling like fake set dressing due to a lack of responsiveness (Sweetser & Wiles, 2005). It minimises the chance of falsely perceived affordances that disappoint the player when their attempts to interact fail.

Figure 3 Perceived and actual affordances may or may not align. Credit McGrenere & Ho (2000).

Perceived and actual affordances may or may not align. Credit McGrenere & Ho (2000).

Warren (2006) describes how the intuitive idea of ecological perception has been formalised into precise predictive models of behaviour based on the agent’s relationship to its environment. Using dynamical systems theory, he shows how different dynamic environments can create different phase spaces that have different attractors and repellors. These combinations of attractors and repellors result in different equilibria, oscillations, or chaotic behaviour in response to that dynamic environment. This also gives a very precise meaning to the notion of the possibility space in games (see eg Juul, 2005; Nutt, 2012; Pearce, 2002; Salen & Zimmerman, 2003; Schell, 2008).

Having a lot of agency is often assumed to help immersion (eg “[T]he player will be able to go anywhere and do anything,” Frank Lantz in Wonderland 2006. See also Rose, 2008). Agency is often mentioned as though it were a simple spectrum, where some games give the player lots of agency, and others only a little. But the amount of agency in a game can vary depending on the situation, and is determined not only by constraints, but also more subtle influences on player choices. Describing agency in terms of a landscape of possibilities (or a phase space with different attractors and repellors) provides much more detail and subtlety than a one-dimensional spectrum. The cliffs bordering the landscape form the constraints on the player, and the undulating terrain corresponds to the various influences.

Agency Basics


A very linear game might just be a winding tunnel for the player to slip down. A choose-your-own-adventure game would look more like a branching tree, where the player can choose different forks in the tunnel. Many games of progression instead have a “rivers and lakes” design (Schell, 2008), where the player is given pockets of free play separated by gates or tunnels, ultimately resulting in a linear experience. Other more open-world games might promote a general direction and ultimate end goal for the player, while allowing the player to explore, experiment, and choose their own way to play on their journey to the end. Sandbox games and software toys don’t even have this general direction, they just have a wide open space for the player to meander around in.

Agency Landscapes


There are many different ways to create different kinds of possibility spaces in games, and consequently agency is not adequately described using only a spectrum of 0 to 100%. The infinite different landscapes that could be created will give players qualitatively different feelings of agency. For example, the rivers and lakes design may create very compelling gameplay and narrative experience, but all too often it is abundantly clear to the player that the experience is linear, particularly when they frequently bump against the boundaries of the simulation when it comes to the rivers (see eg Shapiro, 2012).

Jesse Schell (2008) talks about using indirect control to gently guide people through a level so that they don’t even bump into its edges but choose of their own volition to move where the level is most interesting and the plot is advanced. Similarly, we could try to design the mechanics and interface such that the perceived affordances encourage interacting in ways that are possible and consistent in the game, and so players are less likely to bump up against the boundaries of the functionality of the fiction. Another option is to design calculations instead of incomparables (Portnow & Floyd, 2012d). It is possible, even common, for games to provide the illusion of choice to improve the feeling of freedom without actually implementing additional choices (Cardona-Rivera et al., 2014, Schell, 2008; Beram, 2010a, 2010b, 2010c, 2010d, 2010e, 2010f; Portnow & Floyd, 2013a, 2013b, 2013c). This might seem like a cheap trick, but people are quite partial to think their actions have effects even when they don’t, which is known as the illusion of control (Langer, 1975).

But there is also the problem of directionlessness: With too much agency the landscape is flat, and the player isn’t pulled or pushed towards anything, giving them no reason to actually do anything. This is a relatively common criticism of flawed sandbox games (Breslin, 2009). It is a myth that it is a bad thing to have any reduction in agency from subtle influences on choice. In fact, the concept of perceived affordances in ecological perception suggests that we can’t avoid influencing choice even if we try to. Rather we should make sure that perceived affordances are everywhere, and things to do constantly suggest themselves, playful activities positively crying out to be tried. Also that there are repellors to push the player, such as resource scarcity and predatory threats. This would be manipulating the player and reducing the freedom of their choices, but that is okay. It is better to have a possibility space that is traversed and explored, than one that is left untouched. We don’t need to funnel the player towards a particular spot on the landscape, we just need to keep them rolling around without losing too much steam that they become disinterested and stop playing.

Sandbox Agency


That addresses the influences as undulations, but what about the cliffs which form the boundaries? There are two separate questions concerning the boundaries: Size and shape. All the writings so far suggest that when it comes to size, the bigger the better (although see Rose, 2008). However there are likely diminishing returns. For example, having ten toggles that affect different on-or-off parameters for the design of a spaceship would be quite interesting when you press “fight” and see how your choices affect the combat effectiveness of your ship. And if it instead had 20 on-or-off toggles, it would probably be a more interesting game to explore (note here that I am also assuming that all choices actually affect a simulated system, and that all simulated systems are affected by choices). But if the game already had one million toggles, would increasing that to two million really make the game feel more expansive? Probably not as much as the increase from 10 to 20. So I suspect increasing the size of the play landscape will have diminishing returns.

Although bigger might be better, it is not currently technologically possible for the size to be infinite. Luckily, it doesn’t need to be. Does all fiction need fun functionality, or just a subset? Do we need to just give function to only the immediately available fiction at the borders of our game, or do we need to go out of our way and give function to parts of the world that are irrelevant to the kind of play we had in mind? For example, it is sufficient to give simplified or abstracted functionality to cleaning your gun in a shooter, or do we also have to give full functionality to the chemistry of baking in case the player wants to discard his weapon and become a pastry chef instead? What is the best shape for the boundaries of the play space? Here there seems to be much less literature or discussion (although see Juul 2007a).

I hypothesise that there is no universal best shape for the play landscape, but I suspect there might be better shapes for the micro structures of sections of cliff. For example, very sharp, thin cliffs that create an acute angle for the play landscape are probably not conducive to immersion because they are very tunnel-like, and the player doesn’t have much room for movement before bumping into the boundaries. So broader, flatter boundaries are probably better than thin, sharp, acute angles such as in star shapes.

From this, one could infer that perhaps the best shape for the play landscape would be a circle, since that minimises the presence of sharp angles in the boundaries. But I suspect that another variable is much more powerful in affecting the ideal shape of a play landscape: The undulations. As previously discussed, there will invariably be attractors and repellors in the play space, creating hills and valleys, and that is the way it should be to create an engaging and interesting world. So it is quite likely that if we just go with a circle as our shape, there will be an attractor near to a boundary. This is something you want to avoid, because it increases the chances of the player bumping up against the constraints of the simulation. Ideally there will be repellors near the boundaries and attractors that serve to pull players away from the edges to reduce the chances of breaking immersion. So if you find an attractor is near a boundary, a more effective strategy than trying to modify the undulations might be to push the cliff back to give the attractor some more room to play. Therefore, I suspect the ideal shape for the boundaries of the play landscape is not a circle, but a unique shape for every game determined by the idiosyncrasies of the undulations created by the mechanics. This allows indirect control by perceived affordances to successfully divert players away from the boundaries.

It is a myth that immersive games should have total agency, with every boundary pushed out as far as possible, and as few influences or manipulations of player choice as possible. That would be an impossible game to make, it would not be very engaging, and there is nothing to minimise the chances of players running into the boundaries of the simulation. To make immersive games of worlds, it would be more effective to very deliberately include many influences that make for an interesting landscape to negotiate. One should also very deliberately sculpt the boundaries to match well these influences so as to minimise the chances of players hitting the edges of the simulation. The boundaries of a simulation have to accommodate the competing variables of maximising the size of the play landscape, avoiding sharp spikes and tunnels that make it easier to bump into the boundaries while playing, and ensuring that the boundaries align with repellors rather than attractors.

However indirect control and perceived affordances are stochastic processes – At best all we can do is achieve a strong tendency for certain behaviour in the majority of players. There will still be cases at the ends of the bell curve where players get a rare situation in the game or the game is played by rare individuals and as a result, despite our best efforts, immersion is broken. For example, transgressive play and exploratory play are activities that are specifically interested in pushing the parameters to extremes and in identifying the rules and constraints in the game system (see eg, Aarseth, 2007; Goodrow, 2013; Gopnik & Schulz, 2007; Tornqvist, 2014). Some people even prefer these playstyles, and would be dissatisfied with a game that did not allow or support them. Other players suspend their disbelief with great enthusiasm. These players are more likely to role-play and will actively avoid situations that might break immersion (Chimelarz, 2013; Edwards, 2003, 2004). We don’t control the personalities of the infinite number of possible players of our games, but we do control our games, and at least there we can try to maximise immersion by designing games of worlds: How to Make Games of Worlds



Would the Holodeck be any fun?

1. Myths of Immersion

2. Realism vs Gamism

3. Stylised Mechanics

4. Process Intensity: Game as Referee

5. Agency and Affordances

6. How to Make Games of Worlds


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s