Argax Project: Thesis: Draft Node

Node Status: COMPLETE

Conclusion

A work of interactive fiction (IF) can be defined as an explicitly-modelled narrative world with a text-based user interface. Interactive fiction games were once a dominant commercial game form, and new games are still produced by a small but active community. IF has also been applied to various "serious" non-entertainment goals, such as education and interactive drama research.

Skald is a new interface for IF games. Skald is still text-based, both for output and user input. However, it is menu-driven, rather than offering free-form input on a command line.

As hypothesized, Skald has shown a number of interesting effects on game-play with real users. It successfully eliminates syntactically invalid user input, and it makes object and action affordances clearer--though not crystal clear for all players. Skald did not significantly change players' game speed or the time they spent playing a game, but it did lead to players using with a wider range of both objects and actions. In particular, they did much more examining of game world objects.

Skald was rated as easier to use than the traditional IF by 70% of participants in this study. However, this did not correspond to a significant increase in world-level agency, which suggests that certain features of the player also impacts this feeling of agency. For example, it was found that previous command line interface experience corresponded with a higher sense of world-level agency, although previous IF experience did not.

Skald had no significant effect on users' reported experience of the story--the structure of the story, the challenge of the puzzles, the impact of their choices, or their overall enjoyment. However, Skald did have some subtle effects on the story structure, increasing the distribution of possible endings that player's reached.

Despite Skalds advantages and it being the preferred UI for a slight majority of users, the traditional IF interface was still preferred by a substantial minority of players.

Future Work

The findings here suggest at least two avenues for future work.

First, Skald should be expanded to also support a command line input alternative, possibly revisiting the "real-time parsing" design considered in Chapter 4. As some participants suggested in comments, a hybrid design may still be valuable. As with Legend Entertainment's design, this may offer some blending of the best of both worlds, depending on player preferences. As currently architected, this will be a bit of a challenge, given that the TADS parser runs on the server and was largely bypassed in the current implementation.

Secondly, it seems that the interactive drama poetics could use some further thought, particularly around the notion of player agency. It was encouraging to see that the two different stories used in this study did lead to significantly different levels of story-level agency. However, this story-level agency, as reported by users, did not correlate positively or negatively with world-level agency, as the poetics suggests that it might. Even more importantly, Skald significantly improved the ease with which users could recognize the affordances of the game world and construct commands. Yet this improvement in the "material cause" available to their choices did not have a significant impact on their feelings of world-level agency. It seems that some other variable having to do with the nature of players themselves may be at work here.