At the Office


Home
On the Couch
At the Office
In the Library
Through the Mirror
Mailbox
Neighborhood
Pissing in the Sandbox: What's in an Analogy?

The term "sandbox" in relation to games is one of the most used and misunderstood in talk about games today.  Largely due to the commercial success and cultural high profile of Grand Theft Auto 3, "sandbox" has become on of those grail-like features: if only we can discover its whereabouts, gaming will achieve a glorious new future.  While it would be easy enough to rail against this phenomenon's status as the newest "bright, shiny object", I believe there is a more fundamental problem with the concept itself and the way it is being used.

One caveat before I get started: while the inspiration for this piece was a recent posting on the ludologist web-site, what I say here should not be taken as a critique of Dr. Juul's talk, the substance of which I have no access to.  The points I make here probably bear no relation whatsoever to what he actually has to say on the topic.  Now, back to your regularly scheduled program...

Games in general bear no relation to a sandbox.  On the surface, this is a trivial point.  After all, the dictionary definition of a term doesn't need to bear any relation to its meaning in another context, so long as that meaning is generally understood.  However, there are a number of key distinctions between game structures and other forms of interaction that get glossed over in this image, so it's worth delving a little further into this analogy.  The sandbox image's primary connotation is one of open play: a sandbox has a limit, but what is done within it is mostly the creation of the player.  The sandbox is a mutable medium, and it is the imagination, investment, and interaction of the player that gives it structure, meaning, and fun.  These are, I believe, the two key connections when talking about games as sandboxes: the game is flexible (placing loose rather than strict limits on player behavior) and allows for creative interactivity.

Now, let us look a little more closely at that comparison.  One of the critical elements of the literal sandbox is that the player gives it shape.  Sand as a medium tends to be flat as a signifier.  In building forms out of the sand or inscribing on its surface, the player creates a landscape that was not there before, investing the flat medium with a depth of signifying structures.  The imaginative characteristics of play in the sandbox are transformative: you need not imagine yourself to be at the beach or in the desert.  From the raw matter of the playing field, you can construct any number of landscapes.  This is the first point of divergence in the analogy of games as sandboxes.  While there are a few experiments in player-created spaces (Second Life, for example), with the commonly sited examples of sandbox games, the player does not in fact construct the world at all.  The world is a given, the background against which the gameplay experience unfolds.

In Grand Theft Auto 3, for example, while the player has a number of degrees of freedom, there is never an opportunity to exist outside of the urban space or detached from the avatar.  Rather than being flat signifiers, I would argue that these are in fact over-determined.  The urban space creates structures beyond the simple restrictions of space; it creates expectations about not only the shapes of the environment but the actions and activities that go on within it.  In spite of the fact that the player avatar never eats, there are restaurants; the impersonality of urban spaces (their stanadardization and mass-flow organization) carries over into the lack of interactive options with its inhabitants.  In a large city, after all, we expect different kinds of interactions than we would in a small village; in the latter, we would expect there to be a history between characters, a connectedness of shared experience, while in the former the generic "types" of characters reinforce the mass-culture logic of the urban space.  We are not concerned with interacting with the majority of the game's characters for the same reason we do not worry about not knowing most of the people around us in large cities.

In other words, the sandbox game, as opposed to the literal sandbox, has a defined shape already.  The player does not determine it; rather it has already been pre-determined (if not over-determined).  One might, of course, counter this argument by going farther afield and bringing up examples like SimCity or Rollercoaster Tycoon, where the player does affect the shape of the world.  However, this simply brings us to the next level of divergence between the two terms in the analogy.  For, while it is true that the player can shape the space of these games, this can only be done with pre-determined pieces.  In other words, these games are more like playing with blocks than playing in a sandbox.    Even if the player could warp the space within the game, redefining the nature and function of the blocks, distorting their shapes, or creating entirely new pieces, the rules of the game space are nevertheless set.  One of the key parts of any game (certainly from a designer's perspective) is the user interface.  UI is not simply the mechanism by which the game represents itself to the player; it includes the mechanisms of player input, in other words, the forms of interactivity allowed for by the game.  One could imagine, for example, a version of SimCity where the player can not only use existing zones but create entirely new ones; however, this does not take the game away from the mechanic of selecting and placing zones.  The player cannot alter the fundamental rules of the game and decide that zones can now overlap each other or that zones may randomly switch between functions, or that the mouse will suddenly function as a strange attractor for traffic flow within the game rather than a mechanism of selecting and placing pieces.  While in the literal sandbox, the player's options are similarly limited (one cannot, for example, change the sand into gold at will or create a towering vertical monolith), the player is engaged with the raw matter of the game world; in computer games, this raw matter is not the set of world-objects but the code that enables those world-objects to appear, to have behaviors, and to respond to the inputs of the player.  To argue that there should be a game where the player can interact with this raw matter as a primary form of play is nonsensical, at least in the commercial games arena; it takes teams of trained professionals years to get the code to work and do the things it does.  Giving players the ability to alter the material of the game would only produce a game of "crash and re-install".

Now, a clever theorist will, of course, look at this contrast and argue that it is a mis-reading of the analogy.  Surely, we understand that the term "sandbox" does not apply to the raw matter of the game but to the gameplay experience.  The forms of interactivity may be a given within a game world, and the shapes of that world may be determined, but the player's behavior within the world can still be a sandbox mechanism.  Players can invent their own goals, invest a game world with their own meanings, and be just as imaginatively creative within the game space as long as the world gives them enough degrees of freedom to do so.  But wait, there are some fundamental distinctions here.  The player is not creating their own imaginative space from a non-signifying medium; they are over-writing an existing system of signification with their own.  In other words, what is being described here is not the creation of a new play-space, but rather the appropriation of existing structures and their reorganization.  The underlying system provides a set of resistances (in the form of the pre-determined signifying system), that the player must (consciously or unconsciously) work against.  While there is certainly an interesting potential here, it bears a much closer relationship to the postmodern appropriation and redistribution of signifying elements than it does to the child-like invention of structures.

Furthermore, the game world can be appropriated only to the extent that it allows for such behaviors (intentionally or not).  If the game has a time-limit for performing an action, for example, the player cannot invent a new method of play in that space that involves not doing that action for an arbitrarily long period of time.  While this may seem like a minor point (and repetitive of the argument about the determination of worlds in game space), it actually gets at one of the core issues of the sandbox analogy.  When critics or players ask for more sandbox style gameplay, what this means in the context I have established here is not the ability to create out of whole cloth new game worlds, but rather loose and flexible limits rather than hard and fixed ones.  What is being called for is, in other words, more degrees of freedom.

Now, this is a critical point.  Going back to the distinction between shaping matter and shaping behaviors that are already built (out of the code that is the underlying substructure of the game), creating games that have more degrees of freedom means shaping the code in such a way that it takes into account a greater range of player behaviors.  This is not as simple as giving the player more options for doing things; it also means building the game world in such a way that it reacts to these options.  The UI loop from player input to world response has to be closed.  So, to create game worlds with more degrees of freedom, you have to not only give the player the structures within which they can perform these actions (you can't jump, after all, if there is no input mapped to it), but you also have to put into the code appropriate reactions to these behaviors (if the avatar doesn't actually jump when you tell it to, the action doesn't exist).  So, the question here is not one of structured vs. unstructured playing experiences, but rather simple structures opposed to complex and diverse ones.

[Of course, the more complex and diverse the game structures are, the more difficult it is to build the code, ensure that it works consistently, and make it stable.  On top of that, the more complex the gameplay structures are the more difficult it is to teach the player about them, to ensure that they get those experiences.  Issues of performance, hardware capabilities, and time to production all come in here, but that is a tangent.]

This is where the fundamental misunderstanding of the sandbox analogy lies.  Players experience Grand Theft Auto 3, and they get that feeling of creative interactivity that the sandbox image implies: exploring the limits of gameplay through imagining and then experimenting with different actions and the magical discovery that the game not only allows them to do such things but responds to their actions in interesting ways.  However, what they are experiencing is not a sandbox within which they can do whatever they want; rather it is a flexible simulation that takes into account a variety of actions beyond the simple "progress to goal along a specified play path" logic.  Take, for example, the taxi-driving mini-game.  It is not in the game by accident.  The fact that the player can spend hours driving around the city without ever doing a mission didn't "just happen".  The team that built the game put these diversions into the game world and its code.  It's not an unstructured play experience; rather it is a complex and diverse--but still highly structured--gameplay experience.  It's not a sandbox where you can do whatever you want, it's a mall where you have multiple activities and engagements available as long as you play within the rules that have been established.

At the end of The Matrix, the character Neo tells some off-stage representative of the powers-that-be that he is going to show them what it means to play in a "world without rules".  Of all the pseudo-philosophy within the movie, this has to be the worst, and it represents the same core misidentification that critics are making when they talk about sandbox games.  Without rules, there is no game, there is no play, there is no world.  Rules are not the limits of play, but its generators.  Far from inventing worlds out of some shapeless matter, what so-called sandbox games deliver are internally consistent and coherent worlds, sophisticated simulations that build on their over-determined materials to allow players the fantasy of inventiveness, when in fact every action has to have been accounted for or it simply wouldn't exist in the game space.  The desire represented by the term "sandbox" is not a return to a child-like stage of imagination and creation, but rather an engagement in a sophisticated, articulated, and complex game world that acknowledges the desire of the player to explore and experiment and rewards those behaviors by producing reactions that fall within the range of expected results.

We need to stop thinking and talking about these complex structures as sandboxes, so we can get on with the business of identifying the actual mechanics that create the opportunities for pleasure that we find within them.  The more we mystify these experiences with fantasies of childhood and false analogies, the more difficult it is to talk about the underlying structures and the critical ways in which they structure these worlds.

Home
On the Couch
At the Office
In the Library
Through the Mirror
Mailbox
Neighborhood