next »

[gimymblert]

I'll try a partial implementation of doing generative story, basically I postulate that story speak of human need and use a simplified maslow hierarchy of need to generate theme (survival, social, aspiration). Then I can run/generate character with variable that represent those need. When a character as "theme" need low enough he do not try to fulfill it, he just give feedback about it by emote, behaviour and reaction. Each cycle the system choose a NPC whose need has to be fulfill and then generate a goal with several step and obstacle to reach. The character then as to find a mean to reach the goal and go through the step. With this I will validate a first step toward my idea of generating story and then had dynamism by throwing the player in it, but that will be after I hijack the system to generate quest for regular game format.

[Untrustedlife]

Interesting, im developing a roguelike right now that randomly generates plot-lines for the game (Basicaly Short Stories), and i just finished up with about the first 3 quarters of it, ill go ahead and post a story it generates on here later so you can see. The stroies of course in this case really effect the game so its alot of randomly determined variables.Its definetly worth checking out though.

[Graham-]

I'll try a partial implementation of doing generative story, basically I postulate that story speak of human need and use a simplified maslow hierarchy of need to generate theme (survival, social, aspiration). Then I can run/generate character with variable that represent those need. When a character as "theme" need low enough he do not try to fulfill it, he just give feedback about it by emote, behaviour and reaction. Each cycle the system choose a NPC whose need has to be fulfill and then generate a goal with several step and obstacle to reach. The character then as to find a mean to reach the goal and go through the step. With this I will validate a first step toward my idea of generating story and then had dynamism by throwing the player in it, but that will be after I hijack the system to generate quest for regular game format.

Yeah a major quality to build into any story, like linear ones, is to explicitly define the needs of every character. Tension exists between conflicting needs of characters (or characters and abstract forces). Pacing is built by shifts in needs - or could say the progress towards fulfillment (of a need).

All of my characters function according to needs as well. "Internal economy," that you were talking about before, is well handled by defining characters' needs in a way to play well off each other then activating them with plot devices that you (the system) control(s).

[Graham-]

Interesting, im developing a roguelike right now that randomly generates plot-lines for the game (Basicaly Short Stories), and i just finished up with about the first 3 quarters of it, ill go ahead and post a story it generates on here later so you can see. The stroies of course in this case really effect the game so its alot of randomly determined variables.Its definetly worth checking out though.

Good luck with it.

[Graham-]

Since Oblivion, I"ve been thinking about worlds that exist without the player, and AI interactions.  Oblivion's Radiant AI really excited me.  Hearing about incidents like NPCs killing each other (over food, I think) was really awesome.  But they ended up dumbing the system down, instead of making it more complex so that things like that weren't screwy. 

So for these years, I've been thinking about it.  I've been thinking about a world where the NPCs had needs and wants, and went about fulfilling them through things that humans do daily.  It would start with simple reactions like going from being hungry to looking for food.  Then expand to buying the food, if there were none nearby.  Eventually, it would have to handle food shortages in the village, and some NPCs would have to be spurred into fixing that shortage.  Certainly, the baker would already be working on it because that's how they make money anyhow.  But others might butcher a cow, or buy from a trader.  Or even become a trader.

Yeah people try this approach every so often. You hear about it a lot at least in the dev diary stuff when the devs talk about the original vision of the game. By the time they get to the actual release much of it has been removed. Then you get a game sort-of built inside a functioning - and severely limited - world.

fyi My game has this stuff planned. The world just goes and goes, forever and around again.

And while I find that world really compelling and interesting technically, I recently realized that it wouldn't be interesting as a story.  Without a pre-programmed plot happening, the events aren't significant.

Yes this is true. Creating a story has to be done inside the world. The generative stuff is more like a backdrop. It builds atmosphere. It is like a play-pen that the real story occasionally drops the player in, returning for him later.

The other use for it is as a way to inspire ideas. Jon Blow *drop* says games are as much a tool for exploring a design as a product. I agree with this idea. A classic AI amateur mistake is to not understand this. You don't build AIs, you grow them.

Now, I'm still really interested in such a world, but only as a backdrop.  It can be used to flavor a story, but won't really produce compelling stories on its own.  At least, not often enough.  Like monkeys with typewriters, you can eventually produce compelling stories randomly.  But I think that producing them reliably would mean stifling the system to such a point that they wouldn't be very varied.

Not necessarily. It is very challenging to bend a world to suit a story but nowhere near impossible. You just have to design your world in such a way to facilitate the kind of story you want to tell then develop constructs to control the player.

I talk a little about how to do it here: http://coldtheory.blogspot.ca/2012/09/player-control-over-narrative-these-are.html (same post I plugged before). It's very dense but covers the basics for what we're talking about. Extrapolating it to suit purely generative worlds is more work but isn't a stretch. I will say that given the state of AI in games today doing what we're suggesting is a big step ahead of what's currently being done.

Of course, there's room for all kinds of games.  I could have a lot of fun in a game that had no story, and the interactions were purely AI.  I could also have a lot of fun in a game that had minimal or no AI, and everything was pre-programmed.  But I think the best games will combine them.

Yes. Hopefully they will be combined to such a degree that the difference between something generated and something man-made is indistinguishable. They should be woven together so often and tightly that the mixture is like saltwater. You know there is both salt and water but to tell which is which requires a special device to separate them.

[stephanduq]

Now, I'm still really interested in such a world, but only as a backdrop.  It can be used to flavor a story, but won't really produce compelling stories on its own.  At least, not often enough.  Like monkeys with typewriters, you can eventually produce compelling stories randomly.  But I think that producing them reliably would mean stifling the system to such a point that they wouldn't be very varied.

All that you have to do is have the right filters. Every day in the world millions of conflicts happen, and people change in what they belief. The stories we know and like are the ones that suit certain patterns we enjoy reading about. All that you have to do in such a system is use a character that can actually have the kind of story happen to him that we would enjoy. I have designed a software tool that does this, and the output is incredibly varied and interesting. It just comes down to looking in the right place.

The problem with procedural generated stories is not generating the start for it. A storyworld acts as a one-dimensional layer for a character that represent a theme or conflict. This is all good, and works amazingly to generate a world filled with interested elements and needy people. The problem is that for a story you need multi-dimensional characters, and one story can only hold one of those. Now you can generate the multi-dimensional character, but you will most likely want several stories, so you need several multi-dimensional characters, and sometimes in a story one character can change its own dimensionality (like in Game of Thrones). And this is when the data becomes unreliable to use in stories.

On the creation of grand narrative based on the players action: A storyworld on his own has its own very simple story that is based on prinicipal definition of story. A story is a value changed by conflict. A storyworld essentially has two variables that bind all the characters and other elements in the world to it, a grand conflict and a thematic subject. The bigger the change of value is within either of these variables during the story, the more impact the actions of the character has on the world around him, and the characters in it that align to these variables. The only thing that seperates a grand narrative from a normal narrative, is that the story of the storyworld is so big that everything in it is affected by it

[Graham-]

I don't know what you are talking about.

Why can't you have several multi-dimensional characters in a single world? You can define needs in multiple dimensions. That's what the writers of a good show would do.

[stephanduq]

Imagine you are creating two stories, set within one world. One of a stewardess, and one of a cab driver. In the story of the stewardess, she is incredibly deep, and well developed. In the story of the cabdriver, he is. Now we write a third story where they both meet. If we strive to keep both characters as interesting and deep as in their own story, the audience will get confused as they will not know with who they should associate their own emotions with. The result is that the story will feel flat and confusing. So the result is that we have to decide who is leading, and the one that is isn't will be deprived of dimensions until the character has 1 or 2 dimensions left that compliment the leading one.

What writers of good shows do is that they switch the multi-dimensional character around every episode. So you get the know each one, in a short story that is made for them to lead in, and where the other characters step back a little.

The multi-dimensionality is not created by the world, its created by the storytellers when they interpertate the world they are presented with. The world itself is a very simply thing, that every character shares, and audiences can explain in one sentence.

[Graham-]

Ok I see what you're saying but you're making a mistake.

Writers don't "remove" dimensions. In good writing the dimensions are always there. If they are not there the character will feel flat. The mark of good writing is the allusion to dimensionality with small moments.

If you want to follow your example you can swap around who takes the lead. You can even do this every line. A story generator would take multi-dimensional characters and put them in situations where the needs of one are highlighted over the other.

I don't see how this affects generation. If it can be done with linear media it can be done with procedural.

edit:

The multi-dimensionality is not created by the world, its created by the storytellers when they interpertate the world they are presented with. The world itself is a very simply thing, that every character shares, and audiences can explain in one sentence.

Says who? And even if it were true it wouldn't matter.

edit:

I've solved this problem so if you keep explaining your roadblock I might be able to explain away your issue. I just don't understand it yet.

[stephanduq]

Ok I see what you're saying, but you're making a mistake.

Writers don't "remove" dimensions. In good writing the dimensions are always there. If they are not there the character will feel flat. The mark of good writing is the allusion to dimensionality with small moments.

Yes, but not for every character in one story. Robert McKee has an entire chapter dedicated to it in his book. There is always only one character that is not flat, the other ones enforce or weaken his traits. An easy example, does Hulk in the avenger movies show all the dimensions we have seen in his own series?

If you want to follow your example you can swap around who takes the lead. You can even do this every line. A story generator would take multi-dimensional characters and put them in situations where the needs of one are highlighted over the other.

Its not about wether you can or not, its about your audience. Write a story where you do this, and test it on an audience. 

I don't see how this affects generation. If it can be done with linear media it can be done with procedural.

Here is the thing, its not being done in linear media. The structures you would be using for generation are an abstraction of the original stories. Try  abstracting a good story to its structure, and use that structure to create a new story.  Again, write a story in the same way your engine would do it, and test it out. If you are right, the new story would be just as good

Creating these systems are excpetionally hard, it took Chris Crawford 11 years before he had something working. And for good reason.

Says who?

If you are still at this knowledge level about storyworlds I would suggest you to start reading some books on developing storyworlds for transmedia storytelling. Henry Jenkins is a good start

[Graham-]

Ok I see what you're saying, but you're making a mistake.

Writers don't "remove" dimensions. In good writing the dimensions are always there. If they are not there the character will feel flat. The mark of good writing is the allusion to dimensionality with small moments.

Yes, but not for every character in one story. Robert McKee has an entire chapter dedicated to it in his book. There is always only one character that is not flat, the other ones enforce or weaken his traits. An easy example, does Hulk in the avenger movies show all the dimensions we have seen in his own series?

Uhh.... I think you're getting confused. A single character gets focus, and others may take the sides, but the narrator can switch focus whenever he wants. A well written scene will expose dimension in several characters by the end of it. Robert McKee sounds like a guy who knows things, but I think you may be taking something out of context i.e. too literal.

I haven't seen the Avengers yet. Long story.

There are no restrictions on scene structure. You can tell whatever story you want. If your characters have needs and they act in the pursuit of those needs and are engaged in the activities of the scene then they have to be displaying their "high dimensionality." That may not be obvious.

The art of truly excellent writing is the ability to give depth to many elements simultaneously, the various components of the world, its history, each character, their pasts, etc, then bring them in concert together. Writing side-characters to contrast your main character is one thing. Writing rich side-characters is something else. The latter is just more difficult.

If you want to follow your example you can swap around who takes the lead. You can even do this every line. A story generator would take multi-dimensional characters and put them in situations where the needs of one are highlighted over the other.

Its not about wether you can or not, its about your audience. Write a story where you do this, and test it on an audience.  

So you're saying it can't be done? Go watch the Godfather. Dimensions of every present character are explored in every scene. We see more of one character over another each scene but we learn a great deal about everyone. In fact, behaviour in the small parts is critical to the painting of a character's depth.

I don't see how this affects generation. If it can be done with linear media it can be done with procedural.

Here is the thing, its not being done in linear media. The structures you would be using for generation are an abstraction of the original stories. Try  abstracting a good story to its structure, and use that structure to create a new story.  Again, write a story in the same way your engine would do it, and test it out. If you are right, the new story would be just as good

No shit.
 
This stuff is being done in linear media. Have you never read a piece of literary criticism.... Rich stories have rich sequences. When single events expose many layers of meaning that is the mark of good writing. If the story manages to do that consistently and link everything into one whole that story is a classic (nearly).

Creating these systems are excpetionally hard, it took Chris Crawford 11 years before he had something working. And for good reason.

It is hard.

Says who?

If you are still at this knowledge level about storyworlds I would suggest you to start reading some books on developing storyworlds for transmedia storytelling. Henry Jenkins is a good start

You're splitting hairs.

[stephanduq]

Uhh.... I think you're getting confused. A single character gets focus, and others may take the sides, but the narrator can switch focus whenever he wants. A well written scene will expose dimension in several characters by the end of it. Robert McKee sounds like a guy who knows things, but I think you may be taking something out of context i.e. too literal.

I haven't seen the Avengers yet. Long story.

There are no restrictions on scene structure. You can tell whatever story you want. If your characters have needs and they act in the pursuit of those needs and are engaged in the activities of the scene then they have to be displaying their "high dimensionality." That may not be obvious.

The art of truly excellent writing is the ability to give depth to many elements simultaneously, the various components of the world, its history, each character, their pasts, etc, then bring them in concert together. Writing side-characters to contrast your main character is one thing. Writing rich side-characters is something else. The latter is just more difficult.

So you're saying it can't be done? Go watch the Godfather. Dimensions of every present character are explored in every scene. We see more of one character over another each scene but we learn a great deal about everyone. In fact, behaviour in the small parts is critical to the painting of a character's depth.

I'm tackling these at once, since my reply is similar. In the Godfather only Michael is not a flat character. We are suggested about the depth in other characters because we can see the effects of their pursuits in Michael. For example, the part with Kay and the abortion, if you remove Michael from that scene Kay would still be flat, but in Michael we see an aesthetic emotion that we reflect on, and as a result we create our own image of how Kay must be and what has driven her decision for the abortion. Interview several people about Kay, and you will find that each will give her different strengths and weaknesses, interview them about Michael and everyone gives the exact same answer. All the characters are believable in their setting, and Michael is an amazing deep character. This combined with good storytelling, allows the audience to fill in the blanks of all the flat characters.

Its not creating rich side characters that is complicated, thats just going through a workflow and keeping all your data structured. Its good storytelling that is the hard part.

This stuff is being done in linear media. Have you never read a piece of literary criticism.... Rich stories have rich sequences. When single events expose many layers of meaning that is the mark of good writing. If the story manages to do that consistently and link everything into one whole that story is a classic (nearly).

When I started working on my system I did. But I found out it is all rubbish. The result of using that knowledge in my system created boring, 13 in a dozen stories. You can compare it too food. Imagine reading a review of a Michelin star restaurant. It goes into detail about the ingredients used, the recipe, and the balance of the taste, etc. That still does not mean that you can create the dish as good as the chef did. So I threw the literary criticism and narratology books aside and started reading interviews with storytellers, and emailing them questions. The funny thing is that both Propp and Campbell warn for this in their books, their structures should not be used for creating.

If you follow the literary criticism path you are always one step behind. Like I said earlier, use the algoritms you have made so far, and create stories with them (by hand) and test the stories on audiences. Your system needs to empower players, not pass a checklist of a critic or scholar.

You're splitting hairs.

If you are working on a system like this, one can expect you to be up to date on the research that is available and the methodologies used in the field. Again, look up Henry Jenkins and his work on storyworlds and storytelling

[Graham-]

You're focusing an awful lot on semantics. Be careful of that. The internet spurs that sort of miscommunication.

"Dimensionality" is subjective. This already should shadow any argument about the bounds of dimensionality in any given piece of media, whether a scene or an entire story or whatever. A single moment can communicate a world if constructed correctly. There are tools to deconstructs stories, certainly, and you are using them, but you have to remember that they are only tools. Stories themselves have no rules. All there is is the listener, and he is an incredibly complex thing to understand.

People will most certainly disagree over Micheal. Why did he volunteer to commit the assassination? Out of duty? Certainly. He also wants to contribute. Maybe there is a little brother mentality. Maybe he wants to prove himself. Maybe he has a special connection to his father. Maybe he wants to run away from the question of whether he'll be consumed by the gangster life or not, by doing something that forces his hand forever more.

Have a conversation with anyone who appreciates the movie and you will quickly see opinions diverge. None of those questions have clear answers. I bet even the actors in the movie didn't have clear answers. They say actors read their lines differently each time, at least great ones, including Al Pacino, and then it all gets put together in the editing room. That's where the movie is made.

Maybe on the surface people agree, because Micheal's qualities are under the microscope, but to find the divergence in opinion you just have to dig deeper. Good art is always left open to interpretation. That is almost the definition of art (as opposed to science).

If I have to explain how this argument applies to Kay then you don't understand it. But you can ask a question and I'll reply. Welcome to an adult conversation.
-----

Its not creating rich side characters that is complicated, thats just going through a workflow and keeping all your data structured. Its good storytelling that is the hard part.

I meant creating rich side characters that served the story. When you have side characters they serve a function. They contrast with the main character, or set up a plot device, or whatever. But they also have personalities.

Creating rich personalities but still allowing for their characters to serve their functions is what is hard. It is the combination of rich character design and story telling that is challenging. When you separate them you push yourself into the "all my side characters have minimal depth" box.

This stuff is being done in linear media. Have you never read a piece of literary criticism.... Rich stories have rich sequences. When single events expose many layers of meaning that is the mark of good writing. If the story manages to do that consistently and link everything into one whole that story is a classic (nearly).

When I started working on my system I did. But I found out it is all rubbish. The result of using that knowledge in my system created boring, 13 in a dozen stories. You can compare it too food. Imagine reading a review of a Michelin star restaurant. It goes into detail about the ingredients used, the recipe, and the balance of the taste, etc. That still does not mean that you can create the dish as good as the chef did. So I threw the literary criticism and narratology books aside and started reading interviews with storytellers, and emailing them questions. The funny thing is that both Propp and Campbell warn for this in their books, their structures should not be used for creating.

That's because your system sucked. You can't reveal random qualities of a character. You have to reveal the ones that make them interesting. That was implied.

No one said anything about just using narrative deconstructions. Your generators have to take into account all the nuance of the story teller's craft. Hence the discussion, and why bullshit like Chris Crawford's system fails.

You're splitting hairs.

If you are working on a system like this, one can expect you to be up to date on the research that is available and the methodologies used in the field. Again, look up Henry Jenkins and his work on storyworlds and storytelling

You didn't understand my point.

[thunderhead.hierophant]

I'm gonna drag in some economic theory.  Now, you may or may not agree with this as economics, but hear me out.

A couple of summers ago I read Mises' "Human Action."  If you want the book in one sentence, it would be "Since the scarcity of time makes all choices economic, the proper domain of economic study is all human behavior."

Now, it's the first clause that's important.  He talks in the book about how humans have a ranked list of demands (read:  priorities / motivations) that need to be satisfied.  Once the most pressing demand is satiated, others can be paid attention.

This probably seems obvious.  But what if you gave NPC's a good list of demands?

On a strictly mercantile level, this would make dealing with vendors more realistic.  Let's say a player has been dragging around five greatswords.  The vendor wants greatswords, but after acquiring two, his demand is met.  He will pay less for each successive greatsword.

Now, the vendor wants to stay in business.  So when a thief breaks in and ransacks the till, this demand is intensified.  After earning his trust, he gives the player a quest.

What's more interesting is what motivated the thief.  Maybe his gf left him, leading him to meet the "alleviate sadness" demand with drink.  He loses his job, becomes desperate... Maybe he gives the player a "win back my gf" quest.

So anyway, ranked demands to emergent behavior to generated quests.  Thoughts?

[Charlie Sheen]

I would like to state that pretty much every game has some sort of story. This is not a revelation, of course, as it has been stated many times before, but it's worth repeating just so that the rest of what I'm going to say does not come off as way too strange. Basically, any kind of activity creates, as a byproduct, events which in turn form a story, and since games are activities, they too create stories. The biggest acitivity ever is life itself (which creates the biggest story ever called history) and, in a way, games are mini-lives, small lives within one big life.

But when we live a life we are not, you see, creating stories. The story of our life is a byproduct of our life, not its goal. When I live, my aim is not to generate a series of events that have an "arc" or some other dramatic shit lol; my aim is to exercise power (oh hello there, Nietzsche!). This is why stories created by lives (more precisely: stories that signify real events) are on average far worse than stories made up by people (more precisely: stories that do not signify any real events).

All of this applies equally well to games since, remember, games are lives too. And as such all games create stories: Tetris does, Super Mario Bros. does, Civilization does etc. But what they don't do is create interesting stories. And they shouldn't! As I said, on average, life creates bad stories, and we're talking about life here which is far more meaningful than some random game about weirdly-shaped blocks falling. But life, at least, has potential to create interesting, meaningful stories, whereas games like Tetris have none, or very little. And this, why is this? Is it simply because we're dealing with systems that generate meaningless events? not because we don't have a "drama manager" or, I don't know, cutscenes?

If so, in order to make a game that will have at least some potential to create interesting stories, you simply have to simulate systems that generate meaningful events, and if you think a little bit further, you will realize that.. that simply means simulating people lol. The more complex their behavior is, the more meaningful the events will be, the more interesting stories the game will generate. Is there really anything other than that to it? Is the secret to "stories in games" pretty much psychological simulation?

So what's the point of "drama managers" and "artificial gamemasters"?

I used to flirt with drama managers before, you see, but I can't remember what was so cool about them anymore lol. What I know is that they are boring! Why do I find them boring!? Is it because they will never work? Is it because they require human-level artificial intelligence to work properly? Is it because they fuck up the challenge in order to chain a series of interesting events? I can't really tell. I just know that my instincts are screaming "FUCK THESE DRAMA MORONS!".

[gimymblert]

Drama manager is a fancy name for good game rules ... i mean every game create an engaging experience by reacting to player action through events ...

Therefore the probleme was never stories as procedurally generated, just coherant narratives.

... and one easy solution is stakes!

Imagine, you have a girl unknowingly follwed at a distance by a serial killer (that the audiance know) ... SUDDENLY any meaningless action takes meaning! That the girl stop to chit chat mean that the killer get closer, that the killer slip on a banana and then distance is increase between him and his target ... distance is the measure of dramatic tension ... will the girl escape or the killer will satisfy is sadistic tendency? Any single events participate because stakes give context.

This is a simple example of stakes on a physical level, but you can have it on a morale level (will he finally cheat his wife with after so many temptations) or psychological (will he have enough faith to follow is father's path?) whatever "space" you can define them in.

What about procedurally generated stakes through procedurally generated theme?

[Graham-]

I'm gonna drag in some economic theory.  Now, you may or may not agree with this as economics, but hear me out.

...

So anyway, ranked demands to emergent behavior to generated quests.  Thoughts?

Yeah, so, all AIs should work this way. Often, the first fix you should make to any AI design - assuming it is poor, and it probably is - is to redefine its behaviour as a series of desires, abilities and perception.

So instead of having a shooter combatant (an AI-controlled one) who sees the whole world and tries to pick a play pattern that is deemed "interesting," he should have a desire to kill, one to kill efficiently (etc), the ability to see in front of him and remember the basics of where he has been in the past, a set of details about what the human player did last, and so on. It is far, far easier to build an AI as a set of demands and abilities - and let it produce its own behvaiour - than to build one as a machine of perfect knowledge that "generates" a personality.

Anyone who would try to make me make an AI the second way would receive a funny stare, laughter and so on. If he then tried to touch the code without changing his ways I would literally give one warning, then never, ever work with him again (unless he changed his ways). That is how fucking terrible AI design done like that is. That's how most game AIs are built. Stunning.

[Graham-]

@ Charlie

I don't think the stories generated in Tetris are boring. The events that a player goes through when playing a game he enjoys are by definition interesting, because I just said the player enjoyed them. Life can create wonderful stories. In fact, the most powerful stories in our lives are the ones we participated in, not the ones we were told. What we are told just plays a significant role, but it is when we are the starring character that we are most likely to remember them.

Those stories may not be as interesting to other people but they are interesting to us.

In a sense, the rules that balance difficulty in Super Mario Bros 3 are a "drama manager," as Gimmy says. They are carefully crafted so that the spontaneously created stories by players have arcs. They build tension, and climax, and have releases. Often, a good way to hone mechanics is to consider the likely stories to be created when interacting with them (as a player), then tweaking them so that those stories follow models of good stories.

Here is a simple drama manager. There is a level in which the player must learn a critical skill, develop it largely on his own, explore its applications, then uncover the correct way to use it to overcome some challenge. The player's journey is roughly accentuated by likely events in the level.

There is a general sense of challenge increasing and a climax at the 3/4 mark. There are 3 small spikes of difficulty. There are 2 walls that when up against the player will likely have no idea what to do next for at least a few minutes. There are 2 false positives: seemingly good pathways to a goal that turn out to be illusions. Just the placement of the things that create these experiences, whether static or not, are part of some kind of "drama managing" strategy.

Now we add one single dynamic rule. If the player gets stuck in one of the several - let's say 4 - places he might stuck a little bunny pops out of his hat and rudely offers advice. We justify this behaviour by saying that the bunny is a travelling companion who has limited patience and doesn't want to travel with a "loser." The player is not berated for taking too long, but when the bunny helps him out his relationship with it changes in someway; he loses respect as a competent explorer.

We have created a decent drama manger. In fact, many games already do both of these things, though they normally do the second thing - the bunny - poorly. Navi is the bunny. The free pass through a level after many failed attempts in recent Mario games is the bunny. Those player-aiding constructions are primitive but exist to control the player's drama. Even the amount of health a player gets is part of some drama-managing construction.

You don't need human level intelligence to make smart drama managing decisions. You don't even need dynamic rules. Chess, at least played by competent players, has drama built all over it. Tension, and the kinds of problems players solve, are distributed in a nice way by the emerging properties of the game. That's elegant drama design.

Essentially, to build a good drama manager, you want an intelligence that does the most with the least. You want to control the key points in a generated story and let the player fill in many of the blanks.

Most failed drama managers fail because their creators miss these points. Their creators try to do too much. They try to get their system to understand every element of their game. But that isn't necessary. Is there proper pacing of action? Is there proper balance of challenge? Is there enough confusion? Is there a proper ratio of exploration to puzzle solving? Is there enough rest? etc etc. These are the questions managers should focus on answering. Then they should solve related problems by fucking with the rules of the game as little as possible.

Here is a good example. Let's say you have a town in an RPG. The town has some information the player needs. Some of the info is hard to get. Some isn't. Some is necessary. Some isn't. What the designer wants is for the player to:
  . not waste too much time in the town
  . to get at least some of the optional information
  . to not do all the town things in one burst (i.e. take breaks and fight etc)
  . to only do all the things in the town if it isn't boring
So he creates 2 rules - drama-constructing rules.

1. There is a companion to the player's character, in his party. The companion has an itch to fight. Bad things happen in increasing amounts if the companion's needs aren't tended to. Effectively the player is punished by not maintaining his fighting quota.

The designer can further control the player by changing the companion character based on setting. Maybe in one particular town the companion character is more subdued, because that town has "a girl he likes," effectively allowing the player more town-time per visit.

2. There is a natural "pull" in the player's character to investigate certain issues. He has an intuition that "develops" based on what happens in the game. As he uncovers things he develops a desire to investigate whatever he has an intuition about. If he doesn't investigate he gets "tired," or "bored," or "distracted," etc. These states can affect how other characters perceive him, or how he behaves in battle.

There are signals that alert the player to what his character wishes to investigate, like inner monologues that read like, "I need some closure on this shit with Rita soon," or a "mental arousal" meter that goes up in situations that have potentially compelling outcomes, driving the player in the right direction.

Neither of these rules have complex understandings of the player's state. In fact, they are enforced in a very trivial way. It is not hard to predict their impact, they are easy to implement, and they blend with the world. In fact, they enhance it. That's good drama management.

[Charlie Sheen]

So what are you saying? That good games lead to good stories? Aren't you debasing the concept of "story in games" this way? Wouldn't that mean that the best way to create stories in games is by creating a challenging game?

And what about this situation: you start playing SMB and you die from Goomba. Would you call that a good story? Are you going to say "well, he enjoyed it, so yeah, it's a good story!"?

As for the rest of the post: I can't really understand it.

[Graham-]

Hunh?

No, there is a distinction between "story" in games, like the plot in FF7, and the "user story" in games, like a sequence of events in my head when playing Tetris.

The difference between the two is largely semantic. I think the point was that you can manage a regular story the same way you can manage a user story.

---

The best way to create stories in games is make an engaging game. Challenge and engagement are not interchangeable. Though I know Kierkegaard likes to pretend they are. I know you're a fan of his so I'm mentioning it.

--

What part don't you understand, if you're curious?