Chr1spy - glad you like the model, I wasn't sure whether it was going to seem excessively unusual or overly complicated.

But I must disagree with the statement that all artists/writers instinctively dislike formalization. I know an awful lot of artists and writers, and while perhaps 2/3s of them think rules stifle creativity, the other 1/3 (including me [wink] ) like systems and orderliness and rules of thumb and theories. It's more a matter of personality - some people just hate rules, while other people just like them. Me, I think it's a more fun challenge trying to discover the rules, work within them, bend them, etc. that it is to be in a freeform environment with no rules at all. Probably the same reason I like playing games.

Trapper Zoid - Excellent point about adventure games: "You need to pick up everything that's not nailed down as you just know that you will be using it later." This is exactly the point I was trying to get at when I was arguing that players should not be allowed to go outside the wedge or take actions which do not advance the plot. A story should contain all the things and only those things which are necessary to fully tell the story. And teleology is the determination of what things are necessary to fully tell the story so that they can be made available to the player. [smile]


Anyway, onward with teh Sunandshadow's explication of plot! [wink]

Every plot must have, at bare minimum:
- one character (in our case, the PC) who has
- a motivation to achieve
- a goal, and
- an obstacle which stands in the way of achieving the goal, and
- some action which the character can take to overcome the obstacle.
- (goals and obstacles may consist of character states or object states)

So, the Island engine must be able to understand the previously-mentioned levels and rooms, plus characters (PC or NPC), the character property of motivation, character states, objects, object states, actions, and how to generate characters and objects to fulfill the plot roles of goals and obstacles.

What is a character state? Character states include: character's current emotional state; character's relationship with another character; character has/lacks posession of an object or piece of information; where the character physically is. (These in turn should determine what behavior an NPC is currently carrying out.)

Object states are even simpler: an object simple has to know how it ought to react to any object or character action. In many cases it won't react at all (for example, swearing at a wall would have no effect on the wall). In other cases the object may toggle a behavior or rotate through a cycle of behaviors (e.g. power on/off), or it may change into another object (e.g. Lighter used on kindling results in fire).

So, now we have all the pieces of our basic system. When a level/plot segment begins, an initial incident/thematic question is chosen either randomly or based previous input. A room is populated with characters and obstacles and one or more goals. In general, the goal is to be able to leave the room and proceed to the next room, which is directly equivalent to advancing the plot, but there may be sub goals like solving a puzzle to acquire a tool, which will be needed to solve the next puzzle, which must be solved in order to unlock the door and leave the room.

Now this could be done in a strictly linear way, but that's not our goal since it wouldn't allow the player to make any choices, so we have to decide how often our player should get to make a decision which directly affects which direction the thematic argument should go. Certainly there should be more than one of these decisions per plot wedge. One per room? That might be both more and less often than necessary. You might want to have rooms where there are only fragmentary clues, not enough to make a decision on. Or you might want to have a conversation with an NPC where every dialogue choice affects the plot a little bit. I'll get to how characters and objects should be generated and arranged to provide opportunities for the player to explore the theme next time.

In the meanwhile, would one of you programmer-types be willing to create a proposal/description/whatever for the Island engine, taking into account the elements and procedures I've described so far?

That's a fair bit to digest. But here's a stab at at creating a proposal for it.

First we have character and object states similar to what sunandshadow and I mentioned earlier. These states would contain properties like the ones sunandshadow mentions: emotions, inventory, etc. Which properties are included would depend on the particular game, story setting, and what kinds of choices you want to present. When the character initially enters a room, the system picks a motivation and a goal based on the character and object states. This might also depend on the particular game, because you may want some continuity with previous themes, but also some randomness, so that it's not always the same theme. Potentially this is a list of all the themes you want to include and then one is selected at random, perhaps with some restrictions such as not using the same theme twice in a row (or maybe always doing that), or certain themes not following other themes. As for actions, I say they should be exposed by the objects and NPCs, dependent on character and object states. For example, an NPC might offer the "talk" action, unless that NPCs emotional state is such that s/he's mad at you. And that leaves the creation and placement of objects and characters, which sunandshadow has yet to do. Obviously, there is somewhat incomplete but it offers a starting point and I think as much as possible we want to avoid implementation details until the system is more complete.

tj963

I'm enjoying LOST a lot. It's basically the only thing on TV. I started thinking about it as a model for an interactive story system a couple months ago, and something clicked that seems like it'd be relevant to this discussion.

Each character has their own peculiar convictions and baggage, and these are the things that dictate how they'll act on the island and what challenges they'll face. Charlie is a reformed junkie, so 'the island' provides a crashed drug dealer's plane.

Each character that is focused on delving into a mystery, solving some problem, or actively pursuing and creating their own destinies seem to have constant pitfalls put before them, while those who submit to the natural course of events (such as Hugo) get a payoff (in the season2 'everybody hates Hugo' episode).

This isn't a hard and fast rule by any means, but it seems like a nice mechanism for advancement of plot. I dunno if any of you have heard of Dirk Scheuring, but he mentioned something to me that he's using in his implementation, which is that the player-initiated moments that seem like they present the most difficult decisions to the auto-plotter are the best opportunity to spawn some new obstacle.

- OK, to summarise and make my point; the auto-plotter recognises the focus (the target objects/characters) of the player(s)'s activities and the type of activity they favour, and responds by intensifying the challenges involved in that type of activity.

- Players are rewarded (with objects, currency, abilities) by submitting to events and going along with suggestions provided by the game world.

The effect of the combination of these two mechanisms is that players of different types can be accomodated - (I) people who play for a specific challenge get progressively more involved in one type of activity (II) Players who explore get rewarded with various objects and different paths of advancement.

....
BTW, This forum is probably the best place I've found for letting off interactive storytelling steam. There's precious few. Thanks guys!
waiting for S&Ss next post...

Quote:

Original post by Chr1spy - OK, to summarise and make my point; the auto-plotter recognises the focus (the target objects/characters) of the player(s)'s activities and the type of activity they favour, and responds by intensifying the challenges involved in that type of activity. - Players are rewarded (with objects, currency, abilities) by submitting to events and going along with suggestions provided by the game world. The effect of the combination of these two mechanisms is that players of different types can be accomodated - (I) people who play for a specific challenge get progressively more involved in one type of activity (II) Players who explore get rewarded with various objects and different paths of advancement.

This sounds more like a method for generating interactive gameplay than story. Obviously, the two go hand in hand, but they are different. If you look in sunandshadow's last post, your method doesn't address the theme and motivation. While it sounds like a decent start on a method of generating interactive gameplay, it doesn't address some of the requirements of a story.

tj963

Actually, Chr1spy's approach is very similar to the one that I would take, with a few tweaks. It's a pretty good idea for interactively choosing gameplay types, but it can also work for the storyline as well. (I'm enjoying this topic too, as it's unlocked the memory of the research I did into this a few years back; the idea seems more feasible when I discuss it).

I agree with sunandshadow's list of plot elements: a plot needs a focal character (i.e. the PC), with a motivation, a goal, an obstacle in the way of the goal, and action/s to overcome the obstacle.

The tricky part is the fact that our main character is also the PC, which means she is not in our control*. We also cannot completely control the mindspace of our heroine, as her thoughts are controlled by the player. This is problematic since this is the central character, and the way that we deal with this will heavily shape the types of stories we can tell, and the approach to designing the system.

One of the things I feel uneasy about is immediately deciding what the goal and particularly the motivation of the heroine will be. This is mostly because the reason I want to implement interactive storytelling is to give the player the freedom to enact different types of characters and have the storyworld react to their decisions. I can understand locking in goals due to limitations in designing a truly reactive system (and art assets, I guess). But motivations are within the realm of the player's mind, and are tricky to impose upon a player (unless you are a very convincing writer).

For example, if the goal has been decided to be "defeat the Pirate King" because of the motivation "you are the champion of all that is just", that won't work if the player/heroine would prefer to spend her time getting to know that cute young fisherman a bit better rather than fighting a horde of pirates. However, the same goal can still work if the system can detect that, and create the situation of the pirates kidnapping the PC's boyfriend. The goal is stil the same, but the motivation has changed.

My favoured approach (if I remember my research [smile]) was to leave as much of the story undefined as we can. We can pencil in that there will be a big climatic scene at the end of the chapter, but we won't know what that scene will be until the player makes some decisions. As the story unfolds the story will have to become more defined, and we will be able to lock the player in to more and more story elements.

There needs to be an exploratory phase, where the heroine is left to explore the world with no definitive (i.e. goal or motivation locking) story events occuring. This phase is designed to test the player in order to figure out what kind of heroine they are playing. This involves the introduction of "plot seeds"; little events that help mould the player, give some insight into their character, but also can be used as the basis of further story events later on once the player's character is better defined.

This is getting a bit too long, and I've probably spent too long writing this already, so I'll stop here for now. I can see how this approach might be incompatible with other approaches to interactive storytelling though, although I think it is broadly compatible with the approach we are describing here.

* I hate writing he/she or (s)he for gender indeterminate characters, so I flipped a coin for the gender of our PC.

I agree with that general idea, but it sounds to me like you ask the player a few questions to see what kind of story she wants and then present that story in a fairly linear fashion. It does bring out a good point though. We could choose practically any goal so long as we pick the right motivation. And even if we get attempt to "read" the player and figure our what she wants (or think she wants), we still may be wrong.

Perhaps a way around this might be to somehow allow the player to choose their own goal. Since they chose the goal they would have some motivation, though I guess it would still be tough to create a story that meshes with their motivation. And maybe this is what I see your method doing. The player seems to be killing lots of pirates so we pick a goal that involes killing pirates (or something). Or if the player talks to a lot of NPCs, we pick a goal that involves lots of talking to NPCS. But I don't see how that creates a motivation.

I guess I'm just kind of rambling and don't really have much to add at the moment. Maybe that'll at least spark somebody else's brain.

tj963

The purpose (well, one of the purposes) of presenting the initial incident to the player is for it to act as a test, allowing the player to pick the PC's motivation. We don't want to impose a motivation on the player, we want to impose a disruptive situation on them and let their reaction determine their motivation and choice of immediate goal.


(This NaNoWriMo is killing me - I'm only at 500 words of my 2,000 word goal so far today. [bawling] Although, I shouldn't complain - I've got 5,000 more words written then I did before...)

Quote:

Original post by sunandshadow The purpose (well, one of the purposes) of presenting the initial incident to the player is for it to act as a test, allowing the player to pick the PC's motivation. We don't want to impose a motivation on the player, we want to impose a disruptive situation on them and let their reaction determine their motivation and choice of immediate goal.

Okay, then it sounds as if we're in agreement over that point then. That's one of the ways I think could work.

As far as I've considered, there are a number of approaches that could be taken, depending on what you want the storytelling experience to be like.

---

The first would be a series of questions to explicitly ask the player what she wants. Admittedly, in its simplest form it might be a bit ridiculous:

STORYTELLER: Stop! Who would play this game would answer me these questions three, ere the end he see.
ROBIN: Ask me the questions, Storyteller, I'm not afraid.
STORYTELLER: What... is your name?
ROBIN: Sir Robin of Camelot.
STORYTELLER: What... is your quest?
ROBIN: To seek the Holy Grail.
STORYTELLER: What... is your motivation?
ROBIN: I don't know that! Aaaaaaaargh!

However, at least with this approach the player will get exactly what she asks for; there's no ambiguity as to what she selects. And with good enough writing for the questions it won't be as hokey as the example I've provided (but it mightn't be as amusing).

---

The second approach is a variant of the first, and involves testing the player with a series of situations. In this case the actions of the player determine the answers to the questions that were asked in the first situation. I'm assuming that there will be a series of predefined goals and motivations that have been determined by the writer. The A.I. would be similar to those 20 questions A.Is., where the actions of the player will increase the predicted likelihood of which predefined motivation they are following; if they spend all their time trying to gain favour of NPCs, they are more likely to be motivated by character based options, but if they spend all their time collected loot, greed is more likely to be a consideration.

There's always the chance that the system will get it wrong, but with a well-designed system I don't think this will be much of a problem. This can always be combined with a "Bridgekeeper/Storyteller" event where the player is explicitly asked a question that locks in the motivation and goal (i.e. "Tell me truthfully; you're only in this for the money, aren't you?").

This approach is the one that I like the best.

---

The third approach is to only implicitly reference the motivation. Several story-based games do this already, to good effect. For example, in one of my favourite RPGs, Baldur's Gate II, the game prologue starts with the PC being tortured by the villain who then promises that he will unlock great untapped power within the PC. The prologue ends with the villain being whisked away to a wizard's prison together with the PC's best friend. The PCs first goal is to reach this prison, but the reason could be one of wanting to rescue their friend, gain revenge against the villain, or to learn the secret to gain power. It didn't really matter which because the goal is still the same.

The main problem with this approach, as I see it, is that you are giving away the opportunity to tailor the game experience to the player, as you don't know what makes them tick. This still can work if you offer all opportunities to the player (i.e. assume that any possible motivation can apply at one time), but this weakens the strength of the story as the hero remains a nebulous undefined character throughout the story arc. However this approach might work best if you are wanting to extend the traditional linear game story with a bit more interactivity.

---

Any other approaches that I've left out?

[Edited by - Trapper Zoid on November 3, 2005 9:17:56 PM]

I don't like 3 at all. Here are some variations of 1 and 2 that I do like:

1.5. The game Tender Loving Care gives the player a series of psychological tests as the game progresses, but taking the tests is part of the game, and the psychologist is a character in the game story.

2.5. The character creation activities/tutorial level of the game uses the player's actions to assess their preferences before the actual game starts.

Xenallure uses a mixture of measuring player actions and directly asking the player their opinions through internlal monologue (dialogue) choices.

Quote:

Original post by sunandshadow I don't like 3 at all.

For the purposes of interactive storytelling, I agree; it's a bit of a cop out.

Quote:

1.5. The game Tender Loving Care gives the player a series of psychological tests as the game progresses, but taking the tests is part of the game, and the psychologist is a character in the game story.

That's a really good way to do that. I was thinking along the lines of a conversation with a close confidant such as a friend or mentor, and a psychologist is a good way of doing that. I suppose a priest could be a good alternative if you're using a classic fantasy setting.

Quote:

2.5. The character creation activities/tutorial level of the game uses the player's actions to assess their preferences before the actual game starts.

That's a simpler subset of option 2, but one that I think would work quite well (as it's a lot easier to implement). I was thinking of possibly doing just that in a prologue (maybe set much earlier, such as in the PC's childhood), where the character could be refined in relative safety.

Edit: Oh yes, the internal monologue is a good method too! Although somehow I have the feeling that this works best as a conversation, although a conversation with your own conscience would be a really good way to do this.

[Edited by - Trapper Zoid on November 3, 2005 9:31:52 PM]