The following section goes after 'Complements' and before 'Dysfunction' in the Interaction Model. If you haven't read the longer essay yet, go read it here. I'll add this to the longer article once it's been hashed out.
Now that we've got all the pieces on the table and what each piece 'does', we can talk about how the whole thing works together. As stated before, the function of roleplaying is to reconcile and develop the three aspects in the players' heads. The model represents a self-reinforcing gestalt which self-corrects, manipulates, and adds to the aspects through interactions performed by the players.
First off, I should point out the limitations of my glorious illustration. While the illustration does make it look like there is one Imagined, one System, and one Goal, in reality there is one of these for each player. If there are five players at the table, there are five private Imagined, five private Systems, and five private Goals at work. They are not identical, and never will be, but the process of reconcile-and-develop works to make them more similar. The interactions are the only elements in the model which can be public. Articulation leads the pack -- usually spoken aloud, this interaction is very common to all the players' experience, and can easily be misidentified as 'being' roleplaying all by itself. On the other end, Imbuing is relatively private, and only rarely and barely made public -- and usually by the other players's deduction rather than being explicitly declared.
Now, the public interactions are what align the private aspects. Not only does the GM's Articulation manipulate his own conception of elements in the Imagined, but it also manipulates the other players' Imagined. When the GM describes the smoky bar, the other players add a smoky bar to their private Imagined. I tried to illustrate this; I failed. It would involve the Articulation arrow leaving the GM's System, splitting up into five arrows, and pointing at the GM's Imagined as well as the other four players' Imagined. Multiply this by six interactions, and then by five players. That's a whole lot of arrows, and the illustration quickly turned into spaghetti. With just two players, there are twenty-four arrows; with five players, there are one hundred and fifty. Guess what? Roleplaying is really complex!
When a player performs an interaction, it may be (mostly) public (such as declaring your action in combat -- steering) or (mostly) private (estimating your character's chances at making a good impression with the local potentate -- contextualization). If it's private, it will only update the player's own aspects; if it's public, it will provide some input for the other players' aspects as well. Of course it's not a one-or-the-other public/private distinction, but a spectrum between 'mostly public' through 'a little public, a little private' and down to 'mostly private.'
Interactions may or may not affect the aspects they connect to due to gatekeeper processes. This is especially the case in Steering and Fuel (into System), less a case in Validation and Contextualization (into Goal), and almost never the case in Articulation and Imbuing (into Imagined). System has the strongest gatekeepers because it is the most regimented in terms of access and input; in many games, for instance, only the GM has access to the steering interaction of changing the scene. Other players may want to change the scene to something else, but they are not allowed to and must wait until the GM decides to. Contrariwise, the game may dictate that only the non-GM players are able to stipulate when their character's merits and flaws manifest (Fuel). A player's Goal may have some weaker gatekeepers in terms of disregarding Validation and Contextualization interactions; the player may decide that "that failure didn't mean anything" or "that imagined detail has no relevance to my story." As far as I can tell, the Imagined has no gatekeeper processes -- what is Articulated is true (at least for the moment) and what is Imbued is significant.
In facile terms, once an interaction passes gatekeeper processes, its content is added to the understanding of the aspect -- someone narrates "Ed falls into the water" and everyone's Imagined adds a wet Ed. However, this is only the case when the interaction's content is new information instead of old data. When a player announces that his dagger does +9 damage to ogres, this is not new information added to the others' System; everybody knows that he's got a dagger and everybody knows it's good for attacking ogres with. No one's understanding of the System is changed. However, if the player announces that he will use an Imagined element in a clever way, such as using his dagger to cut a rope that drops a load of lumber on the ogre, this may be something that the other players had not considered before, and that information (daggers can do more than poke ogres) will be added to the other players' Systems. Over time, the addition of new information common to multiple players brings those players' conceptions of the aspects into closer alignment, since they comprise similar data.
I should make pains at this juncture, however, to underscore that interactions which contain old data instead of new information are not without use. These interactions confirm details in the players' aspects. The more times that a player character's kung-fu overcomes incredible odds (validation), the more the other players believe that said player character is capable in combat. They already knew that the charater had impressive combat stats; in fact, they probably already knew that he would win the fight. Nothing new is added, but existing content is strengthened and underscored. Similarily, if multiple scenes take place in the characters' secret hideout and the hideout is described in any detail each time, that hideout will become more and more real and tangible in the players' minds -- their Imagined aspect is heavily reinforced.
Remember the "Bang Bang! You're Dead!" kids way up top? That example only has the Imagined and the System, no Goal. Lots of roleplaying tries to work this way, heavily emphasizing the Fuel and Articulation interactions and giving short shrift to the rest. Player initiative arises from the Goal, however, and so while this aspect cannot be totally removed, its de-emphasis can result in a pretty lifeless game. Recognizing the importance of player Goals and making the interactions that connect to them more prominent and explicit invigorates the whole game, as has been patently demonstrated by the good work at the Forge. The inclusion and 'equal' standing of the Goal ensures that not only are the aspects convergent, but that said aspects and their development are inextricably bound up in what the players care about. In simple terms, not only is everybody thinking similar things, but they are having fun doing it.
Incomplete and Self-Contradictory Aspects
It is (relatively) easy to see the model working between players who have a solid understanding of the game's rules and clear goals for their roleplaying. Especially at the start of a new game, however, players are often shaky on the procedures of the System, have incomplete conceptions of the Imagined setting, and have fuzzy or very skeletal Goals. Roleplaying's reconcile-and-develop function will eventually amend this, allowing the players to synthesize their understandings, enriching everyone's aspects. The players will learn the rules and rituals of the game until they are second nature; they will learn and develop the world in which they roleplay; they will even, over time, come to recognize their own goals and their fellow players' goals, working them together into a conception of 'why we play this game' (as opposed to 'why I play this game').
However (and this is a big 'however'), this will only happen if the gestalt gels together and works. Players working under very sketchy aspects, or even with self-contradictory aspects (often the case in Goals), may perform interactions based on those aspects in erratic ways. It's important to note that there are no 'wrong' ways to perform interactions, but some methods will be more effective in some situations than others. Long-winded narration making frequent allusions to snippets of poetry to set the scene is great when playing by candlelight after dinner with friends; it will be completely lost when playing with thirteen year old new gamers. It's not that the teens are bored, it's that they do not understand the information being presented to them. This articulation 'misses the target' and adds nothing, underscores nothing, in their Imagined aspects.
Alternately, if a player believes that she should get a big effectiveness boost (Fuel) against General Nogoodnik if she uses the Sword of Damocles, she may go to great lengths to ensure that her character has the Sword, that it's sharpened, that the General is out in the middle of the battle, and so on. The player will be very disappointed if it turns out that the others' conceptions of the System affords no special bonuses for special weapons. While all the prior narration may have certainly been interesting, to some extent the player is going to feel as if her efforts were wasted and ineffectual. While the player's conception of System will be closer aligned with the others' after the exchange, this is a disappointing and rather brutal way to go about it.
Roleplaying is a self-correcting process, and players can utilize that function to avoid such problems. It is a simple matter to send out 'test' interactions and watch how they are received and used by the other players. After the first blank-eyed stare from the thirteen-year-olds, the belabored poetic narration can be toned down. The Sword of Damocles can be 'tested out' on the General's lieutenants, displaying how everyone around the table believes it should function in terms of Fuel. Simply watching what characters a certain player gets attached to (imbuing) or how a player responds to system failures (validation) can reveal a great deal. Openly and honestly communicating around the table also circumvents issues. Simply sharing enthusiasm by saying, "I need that Sword! That'll give me a big boost against General Nogoodnik!" can open a conversation about what kind of a boost that sword will give.
Next up -- Emotive Content in the Interaction Model