Difference between revisions of "Agents"

Latest revision as of 08:18, 26 September 2016

Diegetic game elements that can be interpreted as having goals in game worlds, and working towards those goals.

Games can be viewed as simulations, i.e. an imitation of some aspect of reality. While these may overlook other aspects and include fantastical elements, games often include representations of Agents that actively work towards goals through manipulating the game environment. These Agents may be the points through which player can interact with the game or be the conduits for game facilitators or separated sets of rules to enact other inhabitants in the game world.

Examples

While the ghosts of Pac-Man and the alien of Space Invaders can kill the player they do not actively react to what the player does^[1][2]. In this they show little evidence for agency. In contrast, the enemies in later games such as Braid, Gauntlet, the Doom series, the Left 4 Dead series, and the Super Mario series, adjust their actions in response to player actions. This is typically enhanced by them going from passive modes to active modes when first detecting the players.

Players' characters in roleplaying games such as Dungeons & Dragons and GURPS are examples of Agents since enacting or saying what the characters do is the way they can affect the game world. Likewise, all the non-player characters and monster controlled by game masters are Agents.

The computer programs underlying Façade and ELIZA give rise to characters that can be experienced as having personal goals and emotional states, and exploring these is the prime intention of the designs.

Using the pattern

Having Abilities to be able to affect one's environment is probably the most fundamental pattern needed to create the appearance of Agents in a game. This since without Abilities the agents have problems showing that they are working towards goals, and players encountering game entities without goals but with Abilities may wrongly conclude that they actually have goals and thereby are Agents (ELIZA is an example of this). The presence of humans controlling Avatars, Units or other Focus Loci in Game Worlds makes it very common to have Agents in games, and one could argue that unless human-controlled Agents exist in the design it is not a game. The description of Crobots, P-Robots, etc., as games show that others argue that this is not necessary and it should not be forgotten that many Card Games and Dice Games do not have any Game Worlds whatsoever. Avatars, Characters and Units are examples of how Agents can be created by being directed by humans (or only seem to be so due to actually being Mules or AI Players) but Player Characters need humans or AI Players that take control of Characters for them to become Agents. NPCs and Units controlled by Algorithmic Agents or Game Masters are examples of other common types of Agents. Giving several Agents Mutual Goals can make them Teams, and shows how Teams can consist both of humans and non-humans.

Agents as a pattern are often volatile, that is players tend to rationalize them to simpler mental constructs if possible (typically Converters, Containers, Obstacles, Self-Service Kiosks or Traps)^[3]. This in practice means that they are perceived as game elements which move and act with Predictable Consequences and where one doesn't have to assume they have intentional goals. The use of Game Masters, and for Multiplayer Games other players, can make it impossible to reduce Agents behavior to a set of rules since the people controlling them can have an Unpredictable Behavior, at least as long as the Game Masters and players aren't simply following a set of rules (they may not have an option unless the game has a Freedom of Choice or Randomness). For Agents built on Algorithmic Agents giving them Unpredictable Behavior is somewhat of a paradox since they are a set of rules fundamentally but several tricks exists, including Ambiguous Responses, Emotional Attachment, Memory of Important Events, Own Agenda, and Randomness.

Acting on one's perception is often tied to agency as well, and game designers can encourage players to see game elements as having agency by specifically implementing patterns tied to perception, e.g. Line of Sight. This is especially effective if the Agents move and act to maximize their amount of information about their environment.

Since Agents can have Opposing Goals to each other, e.g. if both sides of Capture, Eliminate, or Herd are Agents, it is possible to create Enemies through the use of this pattern (or just Preventing Goals in the case of Herd if those herded are non-aggressive), and this is necessary to create Conflicts in games since this requires at least two opposing parties. Even if Enemies do not have to be built upon Agents (they can display their hostility primarily through their diegetic presentations), unless players perceive their Enemies as having agency the gameplay will more likely be interpreted as Maneuvering or Puzzle Solving than Conflicts. Construction is an example of goals which can be Preventing Goals between Agents (if they are trying to use the same Construction area for different purposes) without necessarily making the Agents Enemies to each other. Agents can of course also be used to support Cooperation, e.g. through providing Companions of team mates in Teams (which also raises the question of providing opportunities to have Reserves). By using Agents for Conflicts or Cooperation game designers can create and modulate Challenging Gameplay, although they may also wish to have Difficulty Levels, Dynamic Difficulty Adjustment and Handicap Systems so that it can be adjusted for individual game sessions.

In Single-Player Games it is easy for players to forget to differentiate between their own goals and that of which the Focus Loci they are controlling. Characters can be used to provide an explicit statement of the goals of diegetic people and thereby promote Roleplaying which in turn makes the player act according to the Characters goals. Quests are especially suitable in this context, since they provide a way to make the goals of the Characters into both diegetic and gameplay concepts and allow Goal-Driven Personal Development. However, Goal-Driven Personal Development can be applied to any type of Agents, i.e. also on those not human. This is most easily achieved in games where one controls Avatars (e.g. the Fallout series), but the makers of the Europa Universalis series suggest that players can Roleplaying historical countries.

Some games are designed to allow the change of whom or what is controlling Agents. One example of this is Drop-In/Drop-Out games where that do not add or remove Avatars when people enter and exit but instead shift control between these and Algorithmic Agents. Another is when Game Masters take control of Player Characters due to them losing their self-control, e.g. when failing sanity checks in the Roleplaying game Call of Cthulhu. Games with Public Interfaces open up for players to switch between themselves or play simultaneously without the game system being able to detect this. In other cases, players willing give up some of their control to Aimbots^[4] or other types of Player Augmentations to achieve better results.

Adding the Roleplaying pattern (which indeed players themselves can do as a form of Meta Game to most games) is an explicit way of advocating that game elements should be controlled as if they had their own goals - if this is not done the players can be said to not be Roleplaying.

Diegetic Aspects

That Agents visually appear as being capable of thought and self-controlled actions is a first step to making players and Spectators approach a game element as something possessing intentionality and agency. Fortunately, this is quick easy to do since humans are prone to anthropomorphize, allowing for example zombies (the Left 4 Dead series), robots (Crobots), Lego figures enacting movie characters (Lego Star Wars series), and ball of tars (Gish) to be viewed as Agents.

To maintain Thematic Consistency it may be required to have Enforced Agent Behavior. This can be encoded in the rules of Algorithmic Agents easily or delegated to Dedicated Game Facilitators such as Game Masters for the control of NPCs (Enemies typically have such a Limited Set of Actions that they cannot break the Thematic Consistency). Game Masters can also ensure that Player Characters act believably as part of their role in setting up and running Self-Facilitated Games. Supporting Agents to have Context Dependent Reactions is another way to help them maintain Thematic Consistency.

Interface Aspects

It is quite self-evident that Agents controlled by humans need to consider how they should be able to initiate all the possible actions the Agents can do through their own actions using an interface. Somewhat less obvious is that games using Quests to provide Character goals may need Secondary Interfaces if a multitude of them are being undertaken at the same time.

Narrative Aspects

Agents are a vital component of any Storytelling since they provide anchors for Emotional Engrossment, and are typically the sources of Conflicts. As such Agents need to be designed in conjunction with a game Narration Structures; for Predetermined Story Structures this can include designing the actual Agents while in other case it can instead be how the Agents are supposed to be able to influence the Storytelling.

Consequences

Since Agents require players to consider what the Agents will do next, they promote Stimulated Planning. They can be used to create Conflicts, and when so used they typically also provide Tension since the players perceive that forces are working against them. They can also be used to add independent Movement to game elements that are to be herded, thereby making Herd goals more difficult.

As entities seen as having intentions about future events, Agents can also be members of Delayed Reciprocity relations.

Agents are not readily compatible with Moveable Tiles given that the latter represent game elements that do not move due to their own will.

Relations

Can Instantiate

Challenging Gameplay, Cooperation, Delayed Reciprocity, Emotional Engrossment, Multiplayer Games, Narration Structures, Predetermined Story Structures, Predictable Consequences, Stimulated Planning, Storytelling, Tension

with Capture, Eliminate, Herd, or Opposing Goals

Enemies

with Enemies

Conflicts

with Construction or Herd

Preventing Goals

with Mutual Goals

Teams

Can Modulate

Challenging Gameplay, Herd, Non-Player Characters, Spectators, Units

Can Be Instantiated By

Abilities, AI Players, Algorithmic Agents, Avatars, Characters, Companions, Dedicated Game Facilitators, Focus Loci, Game Masters, Mules, Player Characters, Players, Roleplaying

Can Be Modulated By

Ambiguous Responses, Context Dependent Reactions, Drop-In/Drop-Out, Emotional Attachment, Enforced Agent Behavior, Game Worlds, Goal-Driven Personal Development, Line of Sight, Memory of Important Events, Opposing Goals, Quests, Player Augmentations, Public Interfaces, Randomness, Roleplaying, Teams, Thematic Consistency, Unpredictable Behavior

Reserves in games with Teams

Possible Closure Effects

Converters, Containers, Maneuvering, Obstacles, Puzzle Solving, Self-Service Kiosks, Traps

Potentially Conflicting With

Moveable Tiles

History

New pattern created for this wiki by Staffan Björk.

References

1. ↑ Enemies section in the Wikipedia entry for Pac-Man.
2. ↑ Wikipedia entry for Space Invaders.
3. ↑ Lankoski, P. & Björk, S. (2007). Gameplay Design Patterns for Believable Non-Player Characters. DiGRA 2007 Conference.
4. ↑ Wikipedia entry for Aimbots.

Acknowledgments