mardi 26 juillet 2011

The goals...

A lot of fantasies are anticipated when we start to talk about interactive storytelling. This is why I think it’s important to clarify the goals of such an approach. Here is my personal opinion on the subject. Let’s start by what I think are the goals to avoid:
  • Interactive story telling is not about player expressiveness. It doesn’t have to insure that the player will be able to do a lot of different things, as in real life.
  • Interactive story telling is not about player free will, it's not about open ended game play and it’s not about multiple possible endings.


Even if expressiveness and player free will are needed to support an interactive story, they are means and not goals.  This relaxes a lot of the pressure often put on those features. This allows the designer to include just the minimum of player expressiveness and player free will to support the true goals.

I think that the most important goals of interactive storytelling are:
  • Having the players feel ”first person emotions”.
  • Having the players freely explore a given theme.


By first person emotions, I mean emotions that the player feels because of his own acts and his own decisions. Emotions that are generated thanks to non-playing characters that adapt their behaviors around the values that the player exhibit when he play the experience.  Off course, the point is to have the player makes decisions as often as he can, not at given forking points in a predefined plot.

First person emotions are in opposition to emotions generated by cinematics because prerecorded cinematics freeze the causes of the emotions and by doing so; put the player out of the causality chain and ultimately out of the experience.  Prerecorded scenes only offer emotions that we can feel by empathy.  In my opinion, “1st person emotions” are far more powerful than empathic emotions and if we consider that emotions generation is what can qualifies an art form, we have to focus our efforts there. 
Here is a good essay on the different level of emotions www.lostgarden.com/2011/07/shadow-emotions-and-primary-emotions.htm

The second goal that i propose is about allowing the player to interactively explore all the facets of a given theme. That’s means that the elements of the story world have to allow the player to make choices that will push him to think about a given theme. He has to be able to think about the chosen theme thanks to the consequences of his own acts.  In that sense, an interactive story is a way of storing an interactive discussion between the author and the players (has holograms store a 3d image). This discussion between the author and the players doesn’t specially have to be moralizing, but it can. As in other Medias, it’s up the author to choose his position on the subject.

Both those two goals complement each others very well. Indeed, emotions are a way to give feedback to the player to allow him to feel the consequence of his act. And exploring a theme also serves as a fertilizer for emotion generation.

You will have maybe noticed that in this post, I often use the word “play”. This is because I think that, in the end, an interactive storytelling experience must be designed as a game.

Designing an interactive storytelling experience as a game allows us to use proven design principles. It also gives us an important ingredient: what is sometime named a “story engine”. That is to say, something that motivates the player in the long run and that will drive him during the experience.  
The story engine can be seen as what Dramatica (www.dramatica.com) names the “Overall Story Throughline”; or what McKee (mckeestory.com) names the “Central Plot”.  The story engine is the support that fixes other story “threads”, where emotions generation and theme exploration can be fully implemented. 

The story engine also offers many decision points, to allow the player to exhibits his values (the system has to progressively make a model of him). All those decision points (for instances, the way he build something over the course of the game) are exactly what we need to implement 1st person emotions and theme exploration.

Some of the previous attempts to create interactive storytelling systems shared the same goals as the ones I listed but they failed.  In my opinion, it’s maybe because they were attempting to do something else than a game.

This vision now explains the term “Dramatic games” that I really prefer over the term “interactive storytelling” that embed too much paradox to my taste.

Now, even if it’s about games, it’s about games with a very different pallet of emotions and themes than usual, and that still is a long way to go…

27 commentaires:

  1. Hello!

    I think a better term than first person emotion is "emotional agency". Actually it contain both of your goal.

    But let's try to analyze that in a pragmatic way. So far you made statement of intention with no real clue about "how" and where we are already.

    1st question: Do we need IS as a mean to achieve this goal? If the goal are more important that the mean then the answer is no, the participatory nature of game put us in the first person. We totally could achieve the same thing through script sequence with branching, parallel or even linear structure. But is the appellation IS really appropriate I could easily argue something simple: branching narrative are already IS, it's interactive as we make decision and the game answer them. Maybe a better term is generative storytelling.

    RépondreSupprimer
  2. Oh I mistakenly publish the first part ...

    ON branching narrative, they don't have to force decision at forking point, it might be the decision of the player that carry on on a "story corridor", Mass effect and heavy rain have branching narrative where the main narrative is complemented by and interact with lower scale narrative (character arc) and gameplay (upgrade, assignation and performance). For exemple if the player is careless and do not upgrade his ship, he loose important character he have relation with, loyalty mission is also a way to deepen relationship with a character and explore his psyche a bit more, but the loyalty meter may also decrease (hence the character affinity with the main player) depending on the player's action with other character or main events and create (trigger) new events themselves. etc...

    Sometimes linear storytelling is just good enough to give the illusion. FF7 is fondly remember by many gamer because it use every mean to tell a story, gameplay, cutscene and navigation. Here side quest are the main storytelling mean to deepen character arc, while it is not interactive beyond the choice to pursue them or not the impact is still there as it create empathy, explain the character motive and give more information which might shift the emotional understanding of the player. The game Nier use this trick with great effect too, when the player is offered to replay the game after finishing him, he is offered the evil looking foe perspective and share their struggle and hope which definitely shift the meaning of the first playthrough into something more ambivalent and less clear cut, making the player question his own motivation and action that where resulting of ignorance. The famous sequence of aeris death is also arguably a great piece of interactive story TELLING, the reason the sequence wass so impactful is not becuse it was the peak of cinematic experience but it was the embodiment of the notion of FATE that cinematic totally convey. During the begining of the game, the player is offered many seemingly inconsequential choice to interact with character and among them aeris, while the consequence does not carry deeply it's a fantastic build up for the cinematic, because just before that cinematics the player is task to kill the character aeris upon new knowledge he know, a character the game had build empathy all along, the character have no choice so his the player (the world is at stake), when the player move toward the character to kill it the cinematic kicks in and rob the player of agency as another character kill aeris, and the player is still too far to prevent it, cinematic or not. The cinematic here parallel the carefully laid out feeling of FATE, it heighten it, and it became the most famous cinematic moment of gaming, and it's in direct continuity of an action the player was doing (movement toward killing aeris). On the same note that's through gameplay that the main vilain is characterize, at the begining it's not a villain, it's the role model of the main character and the game build up so the player feel the same, at one moment you play along this character (sephiroth) and through game mechanics the game tell us is awesome, in combat he does full quadruple digit damage while you only do 2, effectively the player is worthless aside this character and made the connection the main character he control had made, respect. This building up is very efficient because when the role model shift side, it effectively create an emotional impact and sense of tension, both emotional and gameplay. This trick made the character memorable. Too bad square enix start using cinematics more conventionally, to tell story without the gameplay or along a gameplay movement.

    So this is effectivelt interactive storyTELLING, and it's already very effective. But what we want is not TELLING, but DOING.

    continuing next post

    RépondreSupprimer
  3. Oh ben il a manger mon post lol

    Basically I was deconstructing a bit more IS to get at the core, general idea:

    Game already does this, but we are not satisfied, the reason is more about the narrow range of emotion video games deal with. To get to generative story, we need to get gameplay for those type of emotion. Mostly because all interaction in game are about the physical world which game answer with physical behavior. Now if we achieve emotional (moral and psychological) behavior, we still miss the focused quality of story instead of randomly generated event even on a causal basis (see the sims). SO we need a constrain along dramatic quality, but all the tool are there, just use in the wrong sense.

    I was providing example and many deconstruction of game like the sims, tokimeki memorial and princess maker as procedural editing, what they achieve correctly and what are their pitfall and how we can rebound on it. I was also outlining how it can fit existing production model and how a simple shift on presentation of information could achieve procedurally the same effect has the strong script event outline in the post above. Exemple with game like the path, the void and pathologic.

    Basically I was going made and writing an essay :°

    RépondreSupprimer
  4. Hello Timmy (Neoshaman),

    Thanks for your comments.

    About the term « 1st person emotions » or "emotional agency", I am not sure that naming is what is important provided that we can easily relate the name to his concept.

    About the statements I made for whish you think I have no clue and about the long list of examples you provided: I am sorry if you thought that my goal was to be the first one to do what I want to do.

    I don’t want to be the first; I want to push further in the direction I explained. Anyway, it’s pointless to try to innovate completely.

    You can always trace back previous work that also goes in the direction you want to go. Even in Outcast, 13 years ago, I made a kind of ‘player modeling” for the player reputation, to allow the NPC to react accordingly. All that said, I think we can really go forward and do a lot more than what have been done already.

    When you say : “So this is effectivelt interactive storyTELLING, and it's already very effective. But what we want is not TELLING, but DOING. », i completly agree with you. This is exactly the shift I want to contribute to.

    When you say:

    “Mostly because all interaction in game are about the physical world which game answer with physical behavior. Now if we achieve emotional (moral and psychological) behavior, we still miss the focused quality of story instead of randomly generated event even on a causal basis (see the sims). SO we need a constrain along dramatic quality, but all the tool are there, just use in the wrong sense.”,

    I also totally agree with you, this is why I wrote at the end of my post :

    “Now, even if it’s about games, it’s about games with a very different pallet of emotions and themes than usual, and that still is a long way to go…”

    RépondreSupprimer
  5. Oh! hum I'm sorry you thought I'm saying you had no clue and want to be the first, that wasn't the goal at all, I surely have overdone with explanation.

    I wanted to trace back the statement to his origin to better define the problem and show what are already be done.

    For example My conclusion is that even when we achieve some sort of complete system that effectively generate story, it is still very unfocus and it let the player filter element. We need something that filter event right before, and here that relate to the previous discussion on twitter, purposeful "spawning". Instead of simulate the world we only instantiate useful process base on the perceive context of the game, which game already do with Dynamic difficulty (left 4 dead). SO far I bet we are on the same wave length.

    But my concern (and the focus of my research so far) have been about this story constrain and how it relate to gameplay (aka player direction, agency and boundaries). Profiling is one step for the system but its not enough. I see little reference on how you will solve that problem (aka story focus), not that you didn't thought about it. What I mean, is that one scene in a movie try to convey one information only and paralel thread are edit sequentially, only when the first bit is "played" that the second can fire, which mean they compete for attention resource and story beat resource, so it may be discard because there is not enough resource to develop some idea (which is a process of story writing).

    If you answer that concern you will effectively be ahead of all thing I know have been done :) I'm sure you thought about it!

    RépondreSupprimer
  6. Hi Timmy,

    Don’t worry; we are indeed on the same wave length 

    I am not sure that I understood your concern about what you call story focus. But if I understand it well: you wonder how we insure that the player stay focused on actions and decisions that serve the purpose of the story (emotions and theme). If that’s the case, I think that the story engine is the reply.

    The story engine (the gameplay part) goal is to provide a drive to the player. The emotion generation is made in the additional story threads. Actually, I think that I avoid the problem by separating the player primary motivations from the interactive story part. I don’t try to mix them. Look at TV series as Dr House for instance; their story engine is about curing patient. This drives the doctors to have their primary motivations. On top of that, a lot of story threads are plugged to give rise to the entire thought provoking themes we know. As explained in Dramatica, the Overall Story Throughline is in the objective domain and not the subjective domain.

    I exaggerate a bit because, for Outcast2 and The Incident (2 past aborted projects), I was thinking about mixing them but in a controlled way. The theme exploration part would have been anchored in the story engine, but not the emotion generation part.

    Have I understood your concern? If not, could you clarify it a bit?

    RépondreSupprimer
  7. I can complement a bit my explanations on the story engine as an objective-domain anchor for the subjective-domain interactive story threads: I think that, to be a good anchor, the story engine should also be based on character relationship and not on warfare or on a building metaphor. Actually, for a good story engine, we must create a game play that involve relationship in-between characters (NPC<->NPC and NPC<->player). For instance, a good gameplay candidate would be a murder investigation where the goal of the player would be getting information on the past actions of different suspects NPCs, to discover who the murderer is.

    The game would consist of an economy of information. The player actions, to get more information would be: negotiation (information trading, intimidation…)/stealing/finding/deduction/corruption etc… The gameplay would be based on an information economy where information, opinions and knowledge would be “first class objects” (imagine them represented as cards in the game) and where the opposition would be the AI of the murderer that tries to hide himself and to cheat to have other accused. This would be like a chess game where the terrain is the information network and the moves are your negotiation abilities...

    This would be the story engine that gives a goal to the player: find the murderer. The theme could be: “does end always justify means ?” Themes must be expressed as a question. This theme would require the player to be able to use different means to achieve his goal of finding the murderer. Some of those means would be quite controversial. Other means would be more fair play but more difficult in the short term (if we want the story to be moralizing, we can create a difficulty bias toward one reply to the thematic question). The player would be able to see the consequence of his choices through the reactions of some key NPCs that would help him (information gifts) of hinder him (information retention), depending of the player model (profile). A series of subjective story threads would involve the emotional relationship (love, hate, admiration, fear) of the player with some NPCs (reactive AI). This would be the emotion generation part.

    It’s quite difficult to give an example in a comment. I should do a post just on the story engine structure and his relation to the emotion generating story threads. Oh, but before I need to really understand what I am proposing…

    RépondreSupprimer
  8. Oh, i was forgetting to add that quantum representation help a lot to implement all that ;-)

    RépondreSupprimer
  9. One last think, in my vision, the author become a philosopher, not a writer (btw: a lot of writer are philosophers too, to write interesting stories). Indeed, the author exposes all the facets to a theme. And give enough depth to the experience to allow the player to deeply explore the theme, to discover some meaning, some take-aways. This is what i was naming: pre-recorded interactive conversation... This is how dramatic games have the potential to become a new powerful media, even if the experience is bounded by the author will.

    RépondreSupprimer
  10. Actually I was being a bit of jest to challenge you :) I had already solve all those problem, and I can summarize the solution as a neat single word: Stake :)

    Stake tie information, meaning and emotion to a concrete problem, it allow the manipulation and classification of world affordance as a story dynamics.

    Game are no different than a story, game might generate story but they only change the details of the overall stake, the system propose a number of dot that are dynamically link by the game with the player help. A classic story just have those dot already link. If you change the stake you basically change the game. Stake are also fractal, most game are a collection of stake that branch or are parallel to each other in hierarchical fashion.

    Stake decide the consequence, the goal, the mean of the progression, the object of resolution and the progression. Those element further define the remaining story function and allow great details about manipulating the experience.

    For example let study the premise of romeo and juliet:

    Romeo is from faction A
    Juliet Is from faction B
    Juliet love romeo
    Romeo love juliet
    Faction A hate faction B
    Faction B hate faction A

    The stake here is that there is a conflict between the love of the character and the hate the faction they belongs have for each other.

    Remove one fact and it have meaning but no emotional impact. There is no drama. By deciding on which element to place the main stake and the desired resolution you effectively create a strong place to decide what the story is generate. I had further details about how value (which constitute theme) balance the story movement through simple mean derive from stake, so basically I have a mean to concretely implement whatever theme in story/gameplay term very strongly and without ambiguity.

    Stake is very powerful because it give context and meaning to very action.

    Let's take a simple scene: A killer is following some woman as his prey, it's a stake of survival. Chit chat may be just a casual useless action normally but in this context it became an opportunity for the killer to close on her victim if she engage with it, also information have meaning as we know the prey is unaware and she face a consequence of death, the progression can be clearly measure to the goal, it's the simple physical distance if the killer kill in close combat, but we also know that a crowd is likely to negate this progression, of course if the victim engage a dark alley it close further the killer from his goal, if the victim is aware she will try to fly, now we have to conflicting goal and each are sub stake of the over all stake, the victim have a goal to find a safe spot (as define by a stake of survival) while the killer have somewhat the stake of not letting her flee away (as it will prevent his goal of killing). We can clearly see how each affordance have a meaning, if the killer acquire a gun he is suddenly closer, if the victim broke her leg, it's an opportunity (less mean to evade) etc... and the closeness to goal define the tension in a very measurable way. Replace the stake by something like love and suddenly all those action have a very different meaning!

    The dramatic potential entirely rely on stake. I have already define stake in a definitive dramatic grammar that allow seamlessly to go from an idea to implementation and it's so effective that it work with card and token and retain full agency and emotional impact. I will confess that I find that there is ONLY 3 type of dramatic stake from which you may derive all possible dramatic arc.

    Actually you can see hint of this in those analysis of Totems I send you and helpt me to refine the theory into some pragmatical approach. My in depth analysis was based on the theoretical model I had just built prior working on it.

    RépondreSupprimer
  11. On quantum theory, it's better to see it as an information theory, superposition of phase is due to the fact we can't mesure the inside the box, collapsing of the state is simply when decide to remove the uncertainty. The reason it's so weird at quantic level is that information is convey by energie (we see because light impact the retina, we heard because air vibrate) and at that level energy and matter are on the same scale, therefore observing a state is altering the data, at human scale this interaction is meaningless because information scale better than energy. Information being just change through time between energy state, and most information being noise, we only need little energy to infer the relevant state (which is a large scale state too). So we see red because light transmit this information that escape the object, and the object is not significantly modified by this small energy. I tried to kept it short, I'm not there to make an essay. I already think I overstay my welcome because I have this bad habit to explain everything all the time. I try to control myself :(

    It's not quantum theory that is important just information theory indeed.

    RépondreSupprimer
  12. Notice when dealing with stake I didn't need to talk about the player role, all agent in stake space is on the same level, which allow it to be a strong universal model of story and dramatic interaction. Profiling is indeed a strong tool to communicate the player action to the system and stake help define what's really important in that profiling. I have further details on this model I keep for myself right now.

    RépondreSupprimer
  13. Timmy,

    Thanks for your contribution.

    When i read your explanations, i can't help to think that stakes are conflicts. What do you think of that ?

    RépondreSupprimer
  14. Not really, conflict is a specific instance FROM the stake. For example: "save the world" is a common trope which is stake alike, conflict will be more like " but the government prevent me from reaching that rocket who will save the world by blowing this meteor that threaten us". A further distinction is from thematic argument (more precise than just theme) which is like "end justify mean" you proposed and encode the balance of action value.

    RépondreSupprimer
  15. So, stakes are character goals that generate conflicts ?

    RépondreSupprimer
  16. No not at all, a goal is more specific, it's a "what" stake is the "why", it define why some goal is chosen and important to attain and why the failure to attain that goal is to be avoid.

    It define progression toward success and failure. I said that "save the world" was "stake like" (the world is save or doomed as consequence), I choosed this example for clarity and avoiding too much world.

    While stake is an important concept to grasp is not as useful without the complete model I have.

    RépondreSupprimer
  17. Stake is more like how a global situation put pressure on character. The example with romeo and juliet show that's not a goal at all and it's from the character but the situation. But as this situation matter to character through pressure (love and hate) it force them into action, the only way to them to remove the stake is to remove one fact: they stop loving themselves, they stop their faction to hate the other or they stop being affiliate to their faction. Two of these solution are not possible, they don't have nor desire such emotional agency, they don't have the political power to change their faction, what left is the possibility to left their faction through a trick: juliet choose to fake her death so she can effectively free herself from her family. But because romeo thought she was dead, and still want to be unite with her follow her in her perceive state which is effectively a solution that collapse the fate, the rest is history.

    Stake can be local (internal) to a character yes, but stake is not tied necessary to character, it's more like an abstract consideration of a situation. It's also universal and can be share among agent even if it's internal to one character.

    RépondreSupprimer
  18. All that is quite interesting. What you are describing is a kind of story atoms. Composed of different particles that are emotions, goals, conflicts and so on...

    Those atoms can be assembled (hierarchically) to grow as story molecules.

    Sorry for my metaphors :-p

    RépondreSupprimer
  19. I have to think about that in the context of my backard causality chain (present create past causes in the backstage). But maybe it can be an usefull representation ? I just have to check that it doesn't imply too much forward planning of causalities which is the trap i want to avoid.

    I will do a post on my backward causality idea. But i still have to grasp it completely because it imply to have a drama manager that is experience driven instead of being causality driven.

    RépondreSupprimer
  20. Your stake proposition should also take place in what i call the story engine. And i see simpler things there. More gameplay oriented (procedural monotone opposition), as with the information economy i explained. In your stake system, the opposition seems embedded there and this means that the action vocabulary of the player would be very heterogeneous. I am not yet sure about what i am saying here. Just provoking thoughts !

    RépondreSupprimer
  21. Just to be clear : only the story engine would have forward causality to allow the player to play (this would be combined with backward causality).

    In contrast, the other story threads that implement emotion generation would be mostly in backward causality.

    RépondreSupprimer
  22. I would say stake is more like a frame. It let you define a story space in which you could span event on the perceive event horizon. Think about Dynamic Difficulties, you spawn enemy that enter the realm of perception based on how the player fare. It's nearly the same, look at left for dead. You can implement event as a narrative rubber band very like mario kart since you know where the player is in the story space, basically that mean instancing an item based on position that challenge the player.

    The newfound power is that my complete model (what you call the storyengine) tell you all information you need to know and that causality is implicit.

    Implicit as if you only have a position and momentum, and from that you can infer the next and previous move (because potential of movement are limited by inertial constrain within that space define by the rules).

    RépondreSupprimer
  23. I think we don’t exactly have the same goal. It seems that:

    - You want to do a story with interactivity as in games

    - I want to do a game with drama as in story

    That’s said, you ideas are quite interesting.

    RépondreSupprimer
  24. How about having both? ;)
    I can that's what I explain

    RépondreSupprimer
  25. Maybe it's better to walk before running ?

    RépondreSupprimer