Storyteller Design Problem
I have a big problem with Storyteller’s design right now…
There are just too many rules.
The first stories are simple and most people have no problem going through them, but as more and more characters are introduced, it becomes harder to predict what is going to happen when you place the lover next to the angry villain that just secretly killed his wife out of jealousy.
Examples of Rules
I fall in love with the closest character if:
- I am compatible (nobles like rich people, heterosexuals need opposite sex, etc.)
- I don’t feel remorse about something I did to him (lovers don’t care)
- I don’t hate or dislike her (lovers don’t care)
- I am not already in love and as far as I know my partner didn’t die (lovers don’t care)
- I am not envious of this person (lovers don’t care)
I spent the past two months rewriting, debugging and adjusting the rules that govern character behavior, hoping that it would fix the problem, but I got stuck going back and forth without making significant progress. Some designer friends suggested I should proceed to make the whole game with the current mechanics, hoping the process of producing the actual game would reveal the proper mechanics to use.
The build I am preparing for PAX has 25 stories in it and I started noticing a structure underlying the whole mess of character-defining rules. When placing characters in frames, you are trying to create scenes (i.e. theft, murder, crush, abandonment, etc.) that satisfy what the story is asking of you. Could I implement a higher-level mechanic that deals with whole frames instead of individual characters?
I will consider this when I come back from Seattle.
Interesting changes I made to the game:
- Each character only has one nature, making things simpler than the old version
- Max three characters are allowed per frame, this is easier both for players and the code
- Some things can only happen in quiet frames: you used to be able to fall in love during a bloody murder. Not anymore.
- Introduced the Gift and Mina the Vampire
Do you need lots of mechanics for variety, and the unpredictability is an unintentional consequence, or do you need lots of mechanics so that there are sufficient interesting interactions (which kind of goes hand in hand with unpredictibility), or for some other reason I’m not imagining?
If the first (variety), one option would be to try splitting up the mechanics: to explain clearly, imagine if you made three or four different Storyteller games, each with different sets of mechanics, but small enough sets that each game becomes less unpredictable. Then put those four games back into one overarching Storyteller game, as four ‘worlds’ or ‘episodes’ or ‘genres’ or something. Then make it clear to people that they should forget the rules/expectations from the other worlds/genres.
By Sean Barrett on August 18
It seems to me that the real charm and appeal of this game are the interesting interactions and the illusion of “awareness” that the characters are able to demonstrate through their understanding of players choices.
Structuring the system so that each “frame” or story contains its own set of interactions and potential outcomes for players actions seems immediately less complicated. But it also could lead to a decrease in unexpected outcomes and surprise for the player (or a lot of work for you to add unexpected paths through each story).
I don’t know much about the game or how you structured the code, but instinctively I want to say that there has to be a way to store and process large amounts of data per character. Something akin to a database filled with actions and reactions, with a key that would be generated based on the parameters required to trigger the matching response.
I’d love to find out more about how you’ve been handling the AI. Other than Facade, this is the most unusual implementation of NPC AI I’ve ever seen.
Good luck!
By Sam Batista on August 18
@sean
All of it.
What happens when playing Storyteller is that your brain can make sensible interpretations of most setups of actors because our brains seem to be wired to do so. So whenever I add a new element into the game, I need to consider a lot of sensible uses for it. When I introduced the treasure chest, I discovered that a character and a treasure chest looked like the guy became rich, when adding a third character it could be stealing, or giving away money or maybe just one character losing their money to another. And then I have to be careful about the unintended interactions… sometimes they make sense and are funny but most of the time they don’t (ie: you can’t have someone giving away money in a bloody murder frame… it looks stupid and broken).
So in order for an element to be properly integrated I need many rules. Just having a treasure opened up many interesting stories, but I need more elements to make a lengthy game and each element or concept I add interacts with all the other rules. I spend most of my time in the gameplay logic fixing corner cases.
It often happens when I add a new element with all the associated rules that it interacts with stuff already there, allowing new kinds of stories to be made… it would be sad to compartmentalize them and lose all the interesting combinations! I would do this if I end up with many elements and I don’t figure out a way to make the game predictable enough for players to feel they are learning anything…
@sam
Yeah, it would become a different game too. I tried a variation of this experiment where you didn’t put actors into the frames but prebuild frames themselves and then the engine solved everything at the actor level. The interesting thing about it is that there are not that many possible frames you can build! (enough to not fit all of them in the same screen, though).
I decided I am going to try this at least in paper after I come back from the trip.
I am considering making a post about that soon…
By Daniel Benmergui on August 18
It might not be a problem that situations are hard to predict if why things happen is made clear to the player. Why not explore ways to make these large sets of rules expressed in the visuals. For example, is there a visual that expresses “I am envious of you, but otherwise would fall in love with you” the pixel art aesthetic makes this more difficult to pull off.
By Marc ten Bosch on August 20
The underlying assumption being that humans can subconsciously grasp (learn patterns about) complex systems successfully. But the more feedback the better.
By Marc ten Bosch on August 20
Maybe not the direction you want to go (more complexity), but any thoughts on adding interactions when certain frames follow others? I’m thinking of a simple Markov chain or something, to enable some rudimentary story arc structure.
By Alan Au on August 20
@alan:
Most interactions in Storyteller happen between frames
. i.e. Having two lovers on frame 1 and only one in frame 2 means he was abandoned, although individual frames could not know it.
@marc:
The captions in the game help figure out what’s going on in story… if you really want to know, 95% of the information is there. But it’s always forensics; predicting the outcome of your next move is hard right now and that’s unsatisfying to me. I am suspecting the game has too many “mechanics”, and I am trying to figure out if there’s a way to keep the richness with less rules…
By Daniel Benmergui on August 20
It’s true that removing rules is always good.
However, I guess it depends on what you mean by “predicting the outcome of your next move is hard” since that depends on the complexity of the level. Simple levels should be easy to predict, and Complex levels hard to predict at first. It makes sense that prediction is hard, since that is the whole point of the game. You want the player to be surprised by the outcomes, a lot of the fun comes from that, as long as after a while players can internalize the basic ruleset.
But you know all that…
By Marc ten Bosch on August 25