The Koster-Vogel Cube

At the 2001 Game Developers Conference, Raph Koster and Rich Vogel introduced ¹ their well-received storytelling cube as a model for describing the various ways by which players can experience virtual world event sequences as narratives. The following year, they explored it in more detail ². Their model classifies the elements that make up narratives along three binary dimensions:

• Control. Is the story told by the designer, or does it arise from player interactions?

• Context. Does the story arise more from real-world events or from the virtual world itself?

• Impact. Are the story’s effects short-lived or lasting?

These dimensions give eight possible combinations:

• Designer/in context/static. All content exists at the beginning, and players uncover it. Example: Majestic.

• Designer/in context/dynamic. Designer-created content is delivered episodically, possibly making major changes. Example: Asheron’s Call.

• Designer/out of context/static. This recognizes real-world events as part of its fiction—for example, Christmas holidays. Example: Ultima Online.

• Designer/out of context/dynamic. Designers attempt to weave any new features they add into a consistent narrative. Example: Achaea.

• Players/in context/static. Players role-play in the designers’ world. Example: EverQuest.

• Players/in context/dynamic. Players role-play in the designers’ world, and through their actions can change that world in major ways. The designers adjust the context to take account of the events. Example: Castle Marrach.

• Players/out of context/static. Things the players do for their own reasons, irrespective of what the context says they should do. They tell each other stories that make myth of their adventures; they go on themed jungle crawls. This is the default player activity.

• Players/out of context/dynamic. Players act outside of any designer-driven plot, but through their own actions can cause important and permanent changes to the virtual world. Example: Lineage.

The idea behind the Koster-Vogel cube is that it helps virtual world designers conceptualize what kind of virtual world they want; it lays bare the narrative choices that are open to designers. The cube has certainly proven effective in this regard, but I have a few misgivings about it.

My main complaint is that the dimensions seem to change their meanings. Players can role-play in Asheron’s Call as much as they can in EverQuest, so why is EQ classified as having player-driven narrative when AC isn’t? The main difference between the two in storytelling terms is that AC gives (the designers’) stories more impact than does EQ’s static “vision,” so what’s the justification for having differences in two dimensions instead of just one? Similarly, Ultima Online’s players role-play more than

EQ’s and arguably have a stronger ability to determine their own destinies, so why is UO designer/out of context/ static and not player/out of context/static? Perhaps EQ should be in the designer/in context/static sub-cube instead (or perhaps it should be acknowledged that virtual worlds can be in more than one sub-cube at the same time).

The context axis is especially tricky: I’d have thought the way Achaea strives to weave real-world events (such as the appointment of a new administrator) into its fiction would make it a more in-context world than one like EQ that presents the events with no in-context explanation at all. But if the “out of context” flag means “giving in-context meaning to real-world events,” how does that agree with the default player activity sub-cube? There, players are not making any reference to real-world events, just to the in-context events that occur through their own determination of their story. Personally, I think the context axis is weaker than the other two; it’s a distraction, and the model makes better sense if reduced to two dimensions (impact and control).

To illustrate my point, where would MUD1 fit in this scheme? Well it had real-world events programmed in, so that means it’s out-of-context (even though most of the time those events were dormant). It was reset-based, so any story impact was static. Control of narrative events was in the hands of the players, operating in the space of events made possible by the design. This suggests that MUD1 should be placed in the default player activity cube. This may well be correct: MUD1 had no backstory. What difference would it have made if it had one, though? Bear in mind that if I hadn’t added the real-world holiday stuff to MUD1, it would appear in the same sub-cube as EQ (which does have a backstory of sorts—more than MUD1 does, anyway).

Back in Chapter 1, “Introduction to Virtual Worlds,” I discussed persistence versus change as a way to categorize virtual worlds.

The Koster-Vogel cube’s impact dimension is roughly the same as persistence, and its control dimension is roughly the same as change (or player impact ³). There is a difference between in-context building and out-of-context building, but it’s not orthogonal to the other two dimensions; in particular, the groups of players able to build in-context in a world may differ from those that can build out-of-context in the same world.

The Koster-Vogel cube looks at virtual worlds as a medium for storytelling, and as such succeeds. It is, however, neutral concerning the likely success or otherwise of a virtual world that follows any particular classification. Designers using the cube should bear this in mind; in particular, virtual worlds with overarching designed-in storylines (as opposed to opt-in, opt-out mini-narratives) face a long struggle.