What follows is a summary of some thoughts I was mulling over today; it somewhat follows on from ideas presented earlier by Clint Hocking and others.
To be the player of a game is to take on a role, or to assume an identity. This identity may take on different forms depending on how the role is conveyed through the mechanics of the play experience and the visual/auditory presentation of the role itself.
A player's role and the presentation of this role may interact in subtle and complex ways. Certain roles suggest certain presentation styles; for a player to fill the boots of a space-marine, a first-person perspective is apt, whereas taking a third-person perspective may suggest more of a puppet-master persona, which may itself be suited towards exposition of the interactions between personalities or characters, as in the case of the Sims. These are by no means hard-and-fast rules, however.
Another kind of affordance exists between plot - or to frame things a little more generally, the ideas that a designer/designers wish to convey - and the play mechanics that constitute the game experience.
In order to provide a consistent game experience, the ideas a designer wishes to convey, and the play mechanics used to embody these ideas should complement eachother.
An example - from a game that I have done some of the programming for - of where this guideline has not been followed - and I contend suffers for - is Casebook.
This game uses as its narrative foundation the classic detective story. You, the player, take on the role of the assistant to Detective James Burton. The primary affordance of the classic detective/mystery narrative is the interpreter's (speaking generally - could be reader's, player's, or viewer's) ongoing struggle to get to the bottom of the case - taking the collected evidence, formulating hypotheses about what really happened, and continually refining these as new evidence comes to the fore. Classic scientific method stuff.
Where I believe Casebook falls down is the dissonance that divides its classic detective narrative and the tasks that you, the assistant, are presented with. While the player is comtemplating the game's numerous cut-scenes, pondering on the perpetrator, they are also... collecting evidence, and analysing it.
There is no doubt that collecting evidence and analysing it are imperative to solving any case. But they are not what people enjoy about detective stories - so why on earth design a game's main mechanics around it? Curiosity and hypothesising is the detective story's raison d'etre, it should be so with the detective game's mechanics. Evidence collection, at least the part of it unrelated to these dynamics, should be be treated on the sidelines, if at all.
A large part of a game designer's job is to proceduralise ideas. To visually and aurally convey one idea and procedurally present another distinct idea is to create disharmony.
I should note that there are some very good reasons for the design of the game the way it was; unfortunately that doesn't make it good design.
