Author, Director, Performer, Audience

1 day ago 1

Last week saw the end of this year’s Interactive Fiction Competition. I sort of grew away from the interactive fiction world about ten years ago as my interests shifted to the kind of stuff I do here, but before that I was relatively active as a judge and critic, and towards the end I had put together a little theory that I had used to guide my evaluations as a judge. It was exactly ten years ago that I actually formally presented it:

My theory of interaction is that I treat the designer as taking on the roles of screenwriter and director, with the player as both audience and the performer. I consider video games to be at their best when they engage the player in both my roles. This does not actually translate to “games should adhere to traditional notions of ‘fun'”, though doing that well tends to meet this requirement easily. The noticeable case is when it backfires—while one can ideally challenge the actors to new heights of performance and inspire pathos in the audience, it’s very easy to end up reversing the results, and end up with an alienated audience and an uncooperative performer. A performer that is sufficiently irked may attempt to produce a subversive performance, producing a playthrough that alters or reverses the clearly intended themes of the writer. A truly excellent design can notice and account for this.

(I didn’t call it out specifically in that postmortem, but when I referred to “subversive performances” I was definitely thinking of Ultima IV.)

Some years later, I discovered a review of my prank game Annoyotron V: Atrocitron that observed that I had actually done a frontal assault on my own theory and basically tried to test it to destruction:

Reviewing comp games in 2015, Martin wrote that the best interactive works engage us in two ways, as audience and as performers. The Atrocitron exaggerates the contradiction between these roles, deliberately repulsing its audience to demand pure performance; a transcript would read to the uninitiated like a glitched speedrun of some bizarre prototype. It falls to the preface to do the work of the overture in establishing motivation: theory in place of story.

(If you wish to play Atrocitron, it can be conveniently played online here and should you find yourself wanting a copy of the binary, that is available on the IF Archive.

It’s been ten years. Some of my other reader-response theories of gameplay did not hold up so well under that time frame. How does this one do?

I think this one does a little better, and the changes I’d make to it nowadays end up generalizing it.

The Four Roles

The idea here is that there are four “roles” involved in any given session of gameplay:

Author. Game worlds and narratives must be created, and those creations must be realized within the game’s environment.
Director. Games evolve the state of play through sets of mechanics. These mechanics will gate progress within the game and adjudicate the results of actions within the game.
Performer. Games have players: beings whose interactions with the game interact with the mechanics, often via the control of some avatar character within the game world.
Audience. Games are multimedia experiences. Someone in this process needs to be consuming that media.

In the original theory, there are two neat dichotomies that emerge from this quartet of roles:

Individual. Explicitly, in the original formulation, the game designer is the Author and the Director, and the player is Performer and the Audience.
Conversational. The Author is writing for the Audience, and the Director is specifically directing the Performer. Classically, progress in the game is gated behind the Director deciding that the latest performance exceeded some par value, and the newly authored material is understood to be the “reward” for that feat.

We can try to attack these dichtomies, if we wish: the designer may be able to delegate parts of their roles to the player, or messages might be directed outside of their traditional recipients. The players might be able to alter their relationship to the roles as well.

Role Delegation

Once we move away from the specific genre I was interacting with in 2015, it’s pretty obvious that there are plenty of very popular gameplay modes that delegate design or authorship roles to the player. Minecraft relies on procedurally generated geography but the defining feature of the gameplay is that it is the players who shape the terrain and create what are ultimately its most important features. Programming or automation puzzles such as those which made Zachtronics famous invite the player to direct the processing of the game’s tokens in ways that are determined in advance and then left to run. The “factory builder” genre including Factorio and Satisfactory offer the player some of each of these.

Roles can also be heavily minimized. Abstract puzzle games offer little to the player from the Author role, inviting the player to interact with the mechanics as directly as possible. A player that chooses to interact that way can minimize the Author’s role in any game whatsoever—indeed, a speedrunner or competitive player will be obliged to take that attitude as a necessary first step to serious improvement.

Other games offer themselves up as a spectacle, with the Director and Performer roles both taking a backseat as the Audience role becomes front and center. Casual games like Bejeweled and Peggle rely on this sort of engagement, and more aesthetically sophisticated modern versions of Tetris often try to have some sort of multimedia symphony going on around the main gameplay as a core part of the experience.

It’s a little weirder to talk about eliminating the Audience role, but there’s two ways to make that make sense. In one way we can imagine that a player that has chosen to minimize the Author’s role has also ceded the Audience role along with it: for example, a player attempting to master a difficult song in a rhythm game is no longer aesthetically appreciating the song itself and is instead treating it as a set of cues and prompts for the Performer role. The review of Atrocitron that I quoted above suggests I actually accomplished it there, but this was not really my conscious goal: more on that later on.

The other way we can eliminate the Audience role for the player is to delegate it. Video game players can have audiences for their performances, and between e-sports and Twitch streamers there is no shortage of stages on which gameplay is performed.

(As an aside, if you are unfamiliar with how spectacular high-level rhythm game play is these days, the Games Done Quick marathon has had rhythm game showcases as a regular part of their fare in recent years. The mainline marathons this year included showcases of Beat Saber, Chunithm Luminous Plus, and Pump It Up Phoenix, and all are incredible to watch.)

Interestingly, there are also clearly some ways where a player can delegate roles besides Audience. There is a streaming tradition called “Crowd Control” where the player plays a modified version of the game, and spectators may bid to cause it to do things like remove the player’s inventory or reverse their controls for a time. This is a delegation of Director.

A Worked Example: the Atrocitron

I didn’t start out intending to play with the audience/performer distinction. My original intention had been to mess with the Zarfian Cruelty Scale—and a few years after initial release, it turns out that Zarf himself has revisited that old cruelty-scale theory and put it through its paces in the modern world. My goal at the time was “to break every tradition of playing fair with the player while still, in the end, actually playing fair.” This turns out to slot in extremely neatly with the challenges, corner cases, and refinements that Zarf spun out in that revisit! Let’s look at that briefly before actually looking at Atrocitron understands the audience role.

Atrocitron’s initial base warnings are threefold:

The solution is different each time you play.
The game can be rendered unwinnable without warning and without indication that you have done so. (This is a restatement of Zarf’s original “Cruel” rating.)
SAVE and UNDO are disabled.

The last looks like an unusually nasty trick, but it turns out to actually be a mercy. Zarf notes that “the Cruelty Scale was [always] about the player’s responsibility to manage game state.” The effect of the three warnings and the mechanics they imply are that no individual run matters. The instructions end with a clarification: “This isn’t a game with puzzles in it. The game itself is the puzzle.” Interaction strings that would correspond to a normal playthrough are not understood as a win or a loss but are the fundamental unit of interaction. The idea was that the player would bang around through the game world over a number of plays, get a sense of the rules of the world, and then figure out how to apply that knowledge with purpose. That also plays fairly nicely with Zarf’s note that games that expect heavy use of replay are most rewarding “if each run-through is relatively short and the variations cover a broad range of possibilities.”

This brings us back to the audience/performer distinction. The review I quoted up top described the messages I’ve discussed so far as the sole extent of its nod to the audience role: “theory in place of story.” I had actually considered this to be messages directed to the performer role; I was imposing unusual rules and the player would need to know them to interact with it properly. To the extent that I was considering the audience role, it was to explore the question of what the audience is supposed to be an audience to. You’re certainly not going to be enjoying the story, or the elegant prose. At most, you might notice that the game casually mentions that disassembly is fair play and appreciate that some responses to helpful actions are carefully phrased to not suggest the helpful action when read in a vacuum. But it describes itself as “a prop and a plaything”. The player, as an audience, is audience to the box of bytes that is the software artifact itself. To underline this, the final reply to a winning player can never be generated within a play session at all. As I mentioned in my commentary on its source code release, I structured the actual file binary such that the first thing a would-be reverse engineer would find is evidence that the game file is also an encrypted document of some kind. When the player is done, they do not stand before a ** You have won ** message that ends a narrative, but rather at their command terminal, with the final secrets of the binary file itself spilled before them.