Alright, so I sign on today over a dozen posts behind...

Quote:

...the current thread in the writing forum...

I actually did read most of it. Some of it I didn't really understand and arguements like the one about themes is the reason I hated English back in high school so I skipped a bit. Also, I have basically no background in literature other than reading books, so most of these theories and stuff you guys mention I have to look up on Google.

Quote:

Let's call this game Island...

I also agree that this is a good idea.

I still need to look through that journal entry as well.

EDIT: Alright, so I looked at the journal entry. It was interesting and I think I agree with the story as a circle idea as well.

Quote:

...Indigo Prophecy...

I haven't played it, but after looking around the net, the story didn't seem that interactive, but it's pretty tough to say based on what I read.

tj963

[Edited by - tj963 on October 28, 2005 10:34:21 AM]

Quote:

Original post by Trapper Zoid Oh, but there is such a thing as "a game story"!. That's why you can get sites like The Grand List Of Console Role Playing Game Clichés. For example, if I can play a game like the RPG "Tales of Symphonia", and correctly guess both secret identifies of the character Kratos (which they keep "secret" for about fifty hours of gameplay) there's obviously some rules that go behind game story writing. I've haven't been shocked by a plot twist in a game for years. I know from a game writer's perspective that isn't really a good thing; RPGs are one of the most cliche forms out there, but that means it's becoming an established genre by itself.

I'm sorry, but [sick]! "From a writer's perspective that isn't really a good thing" is a massive understatement. Aside from the fact that games come in a whole spectrum of genres (hero's journey, action, crime, mystery, comedy, etc.), even if you want to call the RPG plot a genre, nobody LIKES those stories, so generating them would be inherently a failure and interactive games of that type would not be an improvement on existing games at all.

Quote:

But it makes it a lot easier for someone like me who wants to automate these stories in an interactive system. I'm not sure if there is a universal theory of plot; they'll always be exceptions. However I'd be happy with a well-defined subset, such as "The Hero's Journey"; there's a lot that can be done with that (if handled correctly). The problem I have with the more universal theories of plot is that in order to become more broad they also have to become slightly more vague. While great for humans, this makes them a lot harder to translate into predicate logic suitable for entering into a machine. That's the reason why Propp is so beloved of computer scientists; since he managed to represent Russian folk-tales in the form of a grammar, it was written in a language that they understood.

You missed my point. I am saying, I believe my approach is a universal theory of plot which is not to vague to be implemented as a program. I personally would NOT be happy with a well-defined subset of stories, ESPECIALLY if it's the hero's journey or cliche game story subset. So could you let me present my approach to a universal theory of plot before you decide it's impractical to try to implement?

Quote:

With your model, would the island be potentially extended by adding more rooms if necessary?

Not by adding more rooms, but by adding another island. Or to put it in a more realist setting, say instead of an island our initial setting was one floor of a skyscraper. This floor functions as a complete level. When the player completes the level, the game has to generate a new level. The game could then choose to have the player take the stairs to the next floor of the skyscraper, or take the elevator to the city street outside the skyscraper, or even have a helicopter arrive at the window and take the player to some other building miles away.

Quote:

I agree with starting with a background exposition, the PC (with a set of attributes to define them), a cast of NPCs and possible starting objects. I'm not sure about having a 3x3 grid of rooms; I'd probably abstract the space away to simple "accessable areas" and "non-accessable areas" to be a bit more general. The initial incident would probably happen after a few choices, to give the player some time to adjust themselves to the world and to give the system some time to assess the type of player.

"Rooms" here are not literal rooms with walls and stuff, they are the area of game which fits on one screen (including scrolling). Basically whatever area the player can explore without going through a doorway or something to get to a different screen.

I don't believe it's possible to have the initial incident happen after the start of the game, because the initial incident is the first thing our plot-engine is generating, and it will in turn determine the setting of the game including the contents of the room and possibly aspects of the PC. If you want to test the player through gameplay let's do that in a tutorial level which is part of the character creation phase and not part of the game itself. So please consider the actual start of the game to be when our plot engine generates an initial incident and an initial setting to go with it. It is not necessary to tell the player about the initial incident immediately, but the initial setting will inevitably communicate something about what it is/will be.

tj963 - Glad you like the idea and good job getting caught up and doing the reading, I know that having that much reading to wade through intimidates me sometimes. But, no background in literary theory and you want to generate stories? o.O; Well, I'll try to explain my theory in simple terms, but just ask if I don't explain something enough.

Quote:

Original post by sunandshadow I'm sorry, but [sick]! "From a writer's perspective that isn't really a good thing" is a massive understatement. Aside from the fact that games come in a whole spectrum of genres (hero's journey, action, crime, mystery, comedy, etc.), even if you want to call the RPG plot a genre, nobody LIKES those stories, so generating them would be inherently a failure and interactive games of that type would not be an improvement on existing games at all.

This is getting a bit off-topic, but I think I do understand your opinion here; frankly I don't particularly like a lot of the stories that are presented in todays RPGs, and I'd think a writer might be even more annoyed at what passes for narrative in today's games (however someone must be liking them, otherwise they wouldn't still be making and selling them). I just find it interesting that the same plot elements keep coming up again and again in the games that I play. I guess it's because most game writers, being fans of games, like the elements they see in the games that they play, and so include them again and again in their own stories. And so genres of game storytelling are born. I just find it fascinating from a cultural analysis perspective how these things form.

Now I'd better get back on topic [smile].

Quote:

You missed my point. I am saying, I believe my approach is a universal theory of plot which is not to vague to be implemented as a program. I personally would NOT be happy with a well-defined subset of stories, ESPECIALLY if it's the hero's journey or cliche game story subset. So could you let me present my approach to a universal theory of plot before you decide it's impractical to try to implement?

I do understand why, from a writer's perspective, generating those kinds of stories would be considered a failure. I've had discussions with other writers, and I've noticed there's a huge opposition to any over-formalised theory of plot structure. From what they've said, I think it's because those structures (Propp, Campbell etc.) put too many restrictions on the types of stories they rely to (please correct me if I'm wrong in your case). And it's not really furthering the scope of narrative if a story engine is just generating tons of stories of a type that have already been written.

But after trying to think of methods to make a model that can handle deep dramatic stories, these days I don't think I can do it; at least not as a first attempt. I think I have to aim a lot lower. Since the stories in games are fairly cliche, I think I could create a system that generates "game stories". The advantage of choosing that as a medium is the result will be regognised as "game-like" by the audience. To put it bluntly, I'm hoping the low expectations for game stories will help people ignore the flaws in an interactive storytelling system that's geared towards games. Once I've managed to solve the technical problems, then I can consider how to apply that to creating better forms of drama. From my perspective, it would be better to get working interactive stories on the dramatic level of Pokemon than keep failing at generating interactive Shakespeare [smile].

But I'm certainly not wanting to dismiss your theory of game plot (and I'm sorry if I gave that impression)! I'm also not wanting to limit the scope of interactive storytelling to simply game stories or the "hero's journey" epic types (in fact I'm not sure I'll follow that model, mainly because it's also quite complex [smile]; however, your story-as-a-circle element is still viable for the domains I'm considering). I was just stating that for me, I think I'll have to pick a very limited model for my first attempt. I regognise that other people will have different aims and approaches, and I'm happy to discuss interactive storytelling in general without closing my mind to other people's opinions. And of course it's possible that with your insights into writing models it might solve some of the problems I've had with infusing interactive storytelling with some proper dramatic techniques. So I really would like to read your (and everyone else's) opinion!

Quote:

Quote: With your model, would the island be potentially extended by adding more rooms if necessary? Not by adding more rooms, but by adding another island. Or to put it in a more realist setting, say instead of an island our initial setting was one floor of a skyscraper. This floor functions as a complete level. When the player completes the level, the game has to generate a new level. The game could then choose to have the player take the stairs to the next floor of the skyscraper, or take the elevator to the city street outside the skyscraper, or even have a helicopter arrive at the window and take the player to some other building miles away.

Yes, I definitely agree with that. So to put it in slightly more poetic terms, the island is the fixed portion of reality in a swirling sea of potentiality, and the player's choice will force a new island to be formed out of the possible set of potential islands. This is pretty much what I think needs to be done.

Quote:

"Rooms" here are not literal rooms with walls and stuff, they are the area of game which fits on one screen (including scrolling). Basically whatever area the player can explore without going through a doorway or something to get to a different screen.

I'm not sure about including implementation specific details such as whatever can fit on one screen in the model. In my model, I included everything that the player could get to without difficulty as one "room". So if the player could walk from one side of a continent to the other, everything they passed would be considered in the same "room" with regards to the model, but if they had to solve a puzzle to enter a building, the interior of that building would be a separate "room". However, if travelling is a burden on the player (through random encounters, or just takes a long time), then it might be better to split the world up as you've done.

Quote:

I don't believe it's possible to have the initial incident happen after the start of the game, because the initial incident is the first thing our plot-engine is generating, and it will in turn determine the setting of the game including the contents of the room and possibly aspects of the PC. If you want to test the player through gameplay let's do that in a tutorial level which is part of the character creation phase and not part of the game itself. So please consider the actual start of the game to be when our plot engine generates an initial incident and an initial setting to go with it. It is not necessary to tell the player about the initial incident immediately, but the initial setting will inevitably communicate something about what it is/will be.

I'm still not sure whether I agree with this; in my model I'd probably have a "state zero" that just consists of the background information, with the initial incident causing the world to move to "state one". That way I could put in some choices before the creation of the initial incident.

However, for the purposes of discussing this model, I'm happy to include the initial incident in the start, as it's only a minor detail and won't affect discussing the rest of the system (I don't particuarly want the model to get stuck on this point [smile]).

I was afraid that after not checking over the weekend there'd be be several pages worth of messages to read, but I guess not. I'm really impressed by Civ IV and took up a lot of my time. Anyways...

Quote:

From my perspective, it would be better to get working interactive stories on the dramatic level of Pokemon than keep failing at generating interactive Shakespeare.

I agree here. Not that I want to be limited, but a small success would be better than a big failure.

Quote:

However, for the purposes of discussing this model, I'm happy to include the initial incident in the start, as it's only a minor detail and won't affect discussing the rest of the system

I agree again and whichever is more correct conceptually, it's not a big deal. I'd lean more towards Trapper Zoid's view, but sunandshadow's point is also pretty valid so let's just go with it.

So, what comes next? :)

tj963

NaNoWriMo started today and will hopefully (since that means I'll be getting my novel written) continue to eat all my freetime for the rest of the month. But I'll try to find some scraps of time to continue with this.


Okay, so we've described the story as being a circle. This isn't quite compatable with the familiar Freytag's pyramid, which depicts the dramatic tension of a story as gradually increasing until the climax, wereupon it drops off sharply. But they are similar concepts, so let's see if we can combine them. In point of fact, let's change our circle into a moebius strip.

Why? Because a circle implies that the end is literally the same as the beginning and you could keep going around forever. But this isn't quite right - the purpose of the story is to make a thematic argument, to change something in the audience's mind by conveying a meme through the protagonist's change over the course of the story. So, in our moebius strip the twist represents what changes - the things which are different between the stable state before the initial incident and the stable state the story returns two ofter the climax. In other words, the twist is the way the climax presents the thematic conclusion by rearranging the apparent chaos of the argument into a new pattern.

Thus, let's imagine a story as a journey where we start at the bottom of the circle, climb up the left side and over the top, then suddenly hit the climactic twist and slide back to the bottom again.

So, the shape of our moebius strip is determined by several things: time, which is progression around the loop; dramatic tension, which is distance from the bottom; thematic range, which is distance from one side of the strip to the other (or the width of our wedge as descrived in previous posts); and causality and teleology, which are the essential 'loopiness' which keeps all the little segments of time connected in a linear order rather than letting them collapse into a pile of unrelated bits. Causality keeps each segment hooked to the ones which came before, while teleology keeps each segment hooked to the ones which come after.

Now causality is fairly easy to model with a computer program, especially if it's object oriented programming. All each object has to know is what to do if it interacts with another object. Like, "I am a stick of dynamite. If I interact with fire, I will explode." No problem. But how the hell do we model teleology? That is the secret of computer-generation of stories. And I'll talk about it in my next post. [wink]

I've always like to compare the story to a rollercoaster, with a well planned series of calm periods and thrills that eventually returns to the starting point [smile].

To clarify your description, I know that "causality" is the state of the world being based on past actions, such as the reason being the gallant hero has a shiny sword being he picked it up on his way out the door of his castle, but is "teleology" the reason[i/] why he has the sword, as in: because the hero picked up a sword, he will be involved in a swordfight? I'm thinking this is not really the same as a hero picking up a sword because he wants to be in a fight, but the mere act of having a sword means that fights will come to him?

Heck, this isn't really clear; maybe I'll just wait for your next post describing how you plan on implementing it [smile].

It's more like, the sword existed for the hero to pick up because the plot will require him to have a sword later. Teleology always works backwards in time: X must exist/happen because it is necessary to fully argue the theme/get to the ending.

Quote:

Original post by sunandshadow It's more like, the sword existed for the hero to pick up because the plot will require him to have a sword later. Teleology always works backwards in time: X must exist/happen because it is necessary to fully argue the theme/get to the ending.

Okay, I understand that; it's like in adventure games where you need to pick up everything that's not nailed down as you just know that you will be using it later in the game [smile].

I've tried to think of a good way to put teleology in my system. I think the best approach I was considering was to do what I described in my previous post; using post-hoc reasoning. If the hero gets a sword, then make up a reason for having the sword later. Now I have a word to describe that problem [smile].

Sunandshadow, cheers! the moebius strip is a great way of thinking about it. I just stapled together a little paper one to toy with at work. I drew an arrow labelled 'time' going one way around the strip, and one labelled 'destiny' going in the opposite direction.

To implement destiny in a game I think it'd be nice to have an explicit representation of it in your game world. Like for example in this analysis of the StarWars cycle that I read today, The Force is a direct representation of the demands of plot.

_________________________

I hate to use the word post-modern, but I think anyone working on this IS stuff is aware of how a natural writer's (or any artist worth their salt) instinct is to rail against any formalisation of how they do their job. Thus when Propp (or whoever else) comes up with his general model, everyone immediately writes or finds a story that doesn't fit in with it, to prove it wrong. It doesn't make the model any less valid. I reckon a good implementation of any story model you choose in an automatic plotter could have equally good results.

It's all about quality of execution if you want to touch and entertain people, whereas to interest people who are already working on the same sort of technical problem requires a lot less polish.