WRT: Writer Response Theory

So far I’ve tried to talk about different types of frustration and different approaches to the study of frustration in interactive media. But the elephant in the room is participant expectation.

The experience of frustration is inextricable from the expectations of the participant.

Participants coming to the same work with different expectations may have very different experiences of frustration. For example, a participant may possess media literacy - a refined set expectations and skill relating to the genre or interface. A literate participant already has a mental model that approximates the limits of interaction (”implied code”) and thus can avoid interactions with a high likelihood of failure.

Likewise, all participants will tend towards increased familiarity over prolonged periods of interaction, gradually acquiring media literacy in the work and perhaps the genre. As their mental models of the system of interaction become more refined, they will tend to avoid interactions that are likely to result in failure. This naturally reduces the frustration of a piece over time.

It may sound obvious that literate participants are less frustrated, and that their experiences translate into a general literacy or set of expectations that reduces frustration. This is how accommodation works. There are however cases when accommodation does not work and frustration remains. While people can easily accommodate frustrations that remain predictable, inconsistency (that hobgoblin of User Interface Design) of available possibilities for agency makes it difficult for participants to accommodate by changing their expectations.

Like frustration, inconsistency is neither a good nor a bad thing in interactive art - it is merely a quality that affects the ability of the participant to form and make use of expectations. Inconsistency may be accidental or it may be designed in - with critics tending to prefer intentional, radical inconsistency to accidental lapses. In interactive art inconsistency takes two forms, which I’ll call “weak” and “strong.”

Weak inconsistency is design-based - it is arranged, iterated, or aggregated, such as the ability to interact with certain objects but not with others from a fixed collection of objects within a scope. While weak inconsistency inhibits the ability of the participant to gain literacy and form an “implied code” model of how the system will behave, one can over time and through trail-and-error learn a brute-force catalog of all the clickable and unclickable objects, thereby (slowly) minimizing frustration. In Interactive Fiction, the classic example of this brute-force-able frustration was the maze, a staple of IF design that was slowly replaced by more algorithmic or “elegant” formulations before eventually falling entirely out of favor.

Strong inconsistency is code-based - it is randomized, shuffled, or aggregated, such as the appearance of a door at unpredictable intervals upon entering any of a number of possible locations. This works in much the way Skinner boxes and one-armed bandits do, making haywire of our accommodation processes - an interaction succeeds sometimes and fails other times for no apparent reason, and this encourages compulsive repetition. Likewise, when we encounter selectively reinforced frustrations with no discernible pattern, our stress levels skyrocket, and they stay there.

The positive-reinforcement example of this is the loot-drop in MMOGs, which turns farming NPCs into a long slog at the one-armed bandit. My favorite negative-reinforcement example of of this is rush-hour traffic, which remains stressful for most commuters regardless of their exposure time in hours, months, or years.

Available evidence clearly shows that, even after long periods of adjustment, most people experience the task of navigating through heavy commuter traffic as stressful.

How Not to Buy Happiness - Robert H. Frank, Daedalus

In traffic, inconsistency is the moment-by-moment fluctuation of speeds that constantly regulates your ability to progress, often lacking any clear connection to accidents, closed lanes, particular locations, etc. Like traffic, strong inconsistency in interactive design generates frustation that is not surmountable by experience - not, at least, unless the random interaction is in fact pseudo-random, and a high-level pattern can be discerned. Here I’m thinking of Andrew Plotkin’s Shade, in which the actual algorithm that determines the location of the missing tickets is both elegant in conception and impressively pseudo-random in the experience of the interactor.

Last time I mentioned Media Specific Analysis (MSA). Because it is evident only in aggregate design, the exploration of weak inconsistency in interactive art (e.g. the clickable objects of Myst) requires a holistic approach of this kind. However, the exegesis of strong inconsistency (e.g. the missing tickets in Shade) is a priori, and requires careful examination of the arrangement and execution of source code, which seems to require Critical Code Studies (CCS).