finitely win eventually. In reality, the risk is minimal, and the probabilistic nature of attack success simply functions to provide disappointment when an attack does less damage than expected, and thrill when it exceeds expectations.

The issue of mimicry is even more problematic - the moniker role-playing game' has more to do with the fact that the core game mechanics are inspired by Dungeons and Dragons and less to do with any actual role-playing engaged in by the players. Mimicry involves making decisions about your behaviour that match what you believe a given character would do. Children's games of Mums and Dads' are games of mimicry, and tabletop role-playing games such as Dungeons and Dragons require that players narrate their imagined behaviour in the voice of their character, for example, `I swing my Damascus steel sword at the dragon, screaming ‘this ends now!’'

In a Dungeons and Dragons game, players must roll dice to determine whether their action succeeds or fails, and how much damage an attack deals. In Skies of Arcadia and other digital RPGs, the computer calculates this using a set of `random numbers', but players are not required to choose actions based on what they believe the character would choose to do at the time. They must simply choose the best strategy to win the battle. The same goes for behaviour in town maps; players do not choose what to say to non-player characters based on what they believe their character would say, but they simply select what will make Vyse appear as heroic as possible. If anything, agency for playing a role is located in the game software itself, not in the players.

Taxonomic categorisations mask this kind of distributed agency, relying on an understanding of games as objects played by human agents, while the reality is less stable. Caillois' notion of mimicry suggests that players role-play within the boundaries set by Skies of Arcadia, but it is in fact known as a role-playing game because of a network effect - its relationship to a heritage of tabletop role-playing games. In spite of this ambivalence, I find it useful to take from Caillois’s typology the broader categories of paidia and ludus. Certain game mechanics, such as collecting `discoveries' in the sky map, encourage paidia play, while others, such as battles, engender ludic play.

This focus on play styles rather than game types is more successfully met by Richard Bartle’s taxonomy, first presented in 1990. Bartle presents four player types, which, like personality types identied by psychologists, are all present to differing degrees in all people: killers enjoy being better than everybody else and defeating all opponents; explorers like to know everything about the game world and find rare artefacts; achievers like to own everything in the game world, or to complete checklists of tasks; and finally, socialisers like to interact with other people, and are motivated to play by the possibility of social interaction in a game.

Bartle’s taxonomy was intended for online multi-user domains (MUDs), the text-based precursor to the contemporary massively multiplayer online role-playing game (MMORPGs), and he has objected to its application to other genres of video game. Nevertheless, many game designers and theorists find his taxonomy useful for describing the play styles encouraged by different game mechanics, and I will be referring to Bartle’s player types in my description of Skies of Arcadia space design.

Nevertheless, many of the pleasures of RPG play are left out of the taxonomies presented so far. In Toward a Ludic Architecture Stefaan Walz describes the Ph.D dissertation of Jurgen Fritz, which has not been translated from the German, and offers it as a more thorough list of play activities; contesting, risk-taking, leaving it to chance, amusing, pursuing vertigo, meditating, collecting, role-playing, savouring, creating, and problem-solving. This provides a better description of the kinds of pleasures that players might take from Skies of Arcadia: it is amusing, because the storyline and spatial features are often comedic; it enables collecting, particularly in the sky map; savouring, by enjoying the way the storyline unfolds; and problem-solving, by succeeding in battle and responding to spatial puzzles in dungeons.

In addition, Walz offers a typology of perception, pointing out that sound affects the spatiotemporal experience of game-space through sonic space, while tactition (the sense of touch) and equilibrioception (the sense of balance) have hitherto unexplored eects on players' experiences of a game. The voices of fans featured in chapter one will show how the 3D game space aected players' senses of balance, leading to sickness for some.

It should be noted that Skies of Arcadia is compatible with `rumble packs' that give tactile feedback to players through vibration - for example, in the opening sequence the approach of a large ship is annotated by forceful vibrations. My own rumble pack was not functioning for most of my research period, so I will not be able to comment on its effects - however, this is an opportunity to note the eect that unpredictable hardware components have on player experience, and that therefore human-computer interaction extends beyond the design intentions of software engineers. Games are not static structures, but network e]>