Mutual Storytellers

The best thing about the old tabletop role-playing games is the Dungeon Master concept. The Dungeon Master is a person who creates and presents challenges to the other players. Since the DM can essentially do anything that a fictional world could do, he can spin, together with the players, emergent stories of incredible richness and variety. The breadth of experience creatable by a good DM is astronomically greater than in any other form of gaming.

The system works because it depends on the players making things up to entertain each other in real time. We don’t even need to think of it as a game, really. One could easily imagine it as a structured mutual storytelling activity. This mutual storytelling concept is what I want to talk about.

In some sense, all multiplayer games use the mutual storytelling. Some people say that Chess is a series of puzzles that players pose to one another. Games like Team Fortress 2 or Counter-Strike essentially consist of two teams both trying to present the most interesting challenge possible to the other team. In Chess or TF2, the back-and-forth of challenges and solutions creates a story.

Still, we haven’t seen computer games that really focused on mutual storytelling. Left 4 Dead pushed this a bit further than before (as I mentioned in my Design Technologies 2008 article) by placing two totally asymmetrical teams in a rich narrative framework. The Infected team isn’t that far off from a Dungeon Master in D&D. The main difference is that their goal is still to kill the survivors as fast as possible, not to entertain them like a DM. But the idea of mutual storytelling has still never formed the core conceptual framework of a game.

I want to figure out if it’s possible to create a digital game where mutual storytelling is the core idea that drives the entire design. Is it possible to design a game where:

  • Players construct dynamic experiences for each other in real time
  • Players get points for constructing better experiences. Victory by force isn’t the goal, but victory by superior storytelling

The first mental model that comes to mind is to simply take Left 4 Dead and give points to the Infected team based not on how fast they kill the Survivors, but on how profound an experience they create for the survivor players. This reveals the core difficulty of this story trading game concept: how do we judge story effectiveness?

The first option is to create an AI story judge that will rate the storytellers on the effectiveness of their story. This is problematic. It starts to get into the hard AI problem of having an AI understand real-world human concepts. I think we could get an AI to rate action pacing, for example, but what about all of those more human storytelling elements? If the storytellers can place corpses or write things on the wall to imply history to an area, for example, how does an AI rate the effectiveness of this? Say that a message scratched in a wall foreshadows a challenge which the storyteller will present later. Computer can’t read or decode language, so they couldn’t judge the effectiveness of the foreshadowing device.

We could try to step around the problem by only including storytelling devices that the computer is capable of judging. This is a pretty narrow range of tools, though. And under this system, it’s not even clear that an AI judge judging a human storyteller wouldn’t be better than just using an advanced AI Director like Left 4 Dead. I think the AI story judge idea is dead; we’d be better off just investing more resources in a more advanced AI Director.

The second option is to allow the players who experience the story to rate how cool they thought it was. This can only work if they don’t have any other incentives to rate the storytellers low or high. For example, it won’t work if they are in direct or implied competition with the storytellers. This is difficult to pull off because multiplayer gamers tend to be so competitive. I think it might be doable, though.

Here are a couple possible models:

  1. MMO Framework: The first way is to have players compete not just with the people in the current game, but within a larger framework. Imagine an MMO in which you gain experience in two ways: first, through the traditional method of playing with your character, except that now the challenges are presented to you by other players. Second, you gain experience by creating great story experiences for others, who will then judge you highly, thus giving you experience points. This doesn’t necessarily need to be tied to character advancement either. Players could simply be assigned a permanent, persistent storyteller score which is very visible to everyone, like an eBay seller’s rating. Nobody will want to play with someone who has a poor storyteller rating.
  2. Forced Points-Giving: The second way is to force players to give a fixed amount of points to dole out. Imagine a game with six players. At any time, four are storyplayers and two storytellers. The game will proceed in three rounds. Each player is a storyteller for one round and a storyplayer for two. At the end of the game, players are forced to vote on which of the two stories they played was more enjoyable. This clear choice between two experiences may be workable for most players. The “voting up your enemy” problem still exists, though. You might be afraid that the better story will be voted better than yours and thus vote for the worse one. In order for this system to work, the culture and presentation of the game would have to be very non-competitive.
  3. Eliminate Storyteller vs Storyplayer Competition: The third way is to only have storytellers in competition with storytellers, and storyplayers in competition with other storyplayers. Imagine a game with one set of storyplayers, who are always storyplayers and who, within this game, only play stories and rate them. There are also a set of storytellers who are in competition with each other, and who get to take turns telling stories to the storyplayers. Whoever pleases the storyplayers the most wins. Since the storytellers and storyplayers aren’t in competition with each other at all, storyplayers are more likely to rate their enjoyment of the stories honestly.

I’m really not sure exactly how to do this. There are still many questions to be resolved. I want to discuss this concept with anyone and everyone. So consider this an open call for discussion. Please, if you’re reading this and you have an opinion or idea, write it in the comments or email me!


  1. Is there any precendent for multiplayer games that are scored based on players judging each other instead of objective criteria?
  2. Is there any significant proportion of players capable of or interested in creating an interesting storyline or accurately judging one presented to them?
  3. Is it possible to do this non-competitively while maintaining player interest? Could this fit into a larger noncompetitive framework, a la The Sims or Second Life?
  4. How much power can we give storytellers? If we give them unlimited power will they simply abuse it? How many tools can they handle?
  5. Can normal people really fathom the concept of constructing an experience for someone else? Is it possible to train them to do it using various tutorials and tooltips or an AI advisor?
  6. Do we need more time than a normal game session to create a really compelling story? Does this mean we need a broader framework which can spread games out over several days?

10 thoughts on “Mutual Storytellers

  1. Rafael

    About question 4, I think that a good measure to prevent storyteller’s abuse is to limit their story with ‘story points’ (sp). Each event that the storyteller bring to the game must cost sp, so if their sp is zero, nothing more can be done. When a event is resolved, the storyteller get their sp back, in the same rate wich was spended. So, if the storyteller bought “HUGE boss encounter”, she couldn’t also buy ‘disable supply spawn ‘ and ‘wave of 100000 enemies’. This would prevent just some kinds of abuse, but maybe this idea could be expanded.

    Another idea that I have is to not hinder the game just because the storyteller haven’t created nothing. The game should progress in a very linear fashion, bringing encounters to storyplayers and keeping them busy, giving time to the storyteller. So when she selects some events to the story, this ‘autoteller’ mode would be temporary disabled.

  2. Leo

    You’re essentially talking about leveraging a very small portion of your player base to generate content for you — and in this case it happens to be story. The immediate problems are:

    1. Most players don’t want to generate story (they want to be entertained)
    2. Most players will generate CRAP

    To solve this problem, you need:

    1. Strong content filtration and voting
    2. A very LARGE community (because 90%+ will not be storytellers)
    3. A very easy-to-use “story editor”

    It’s probably doable, but story is such a tricky thing that even professional, college-educated writers can get it wrong. Practically speaking, you’d be appealing to a very narrow fan base.

    If you’re limiting this only to “intrinsic narrative”, then I’d argue many MMO’s already do this. For example, a guildleader in WoW — who plans raids, etc. — is like a Dungeon Master in light of “intrinsic narrative”.

  3. Rafael

    I don’t think that the amount of storytellers would be so low, because there is a lot of tabletop RPG Dungeon Masters (I prefer the term Storyteller), me included. The problem is: to be a Storyteller would be so hard as is to be a game designer? Because if it is, very few players will became Storytellers. But if storytelling in this game was something easy to do, if it alllows the storytellers to focus on the story and not in the system, I think that a lot of players would became storytellers.

    For Storytellers, usability would be a huge issue…

  4. Tynan Sylvester Post author


    A Source mod called Zombie Master did something similar to the “story points” concept. But, once again, that was an openly competitive game in which the storyteller was simply trying to kill the storyplayers as quickly as possible. I’m trying to figure out if it’s possible at all to get people to work together to create a collaborative experience.

    We would definitely have to include some limits on storyteller power, at least for performance reasons. I would actually want to look at including at least an option to give the storyteller as much power as possible. This way, they can create a wider variety of experiences.

    You also have an interesting idea with the “AI Assistant” storyteller, if that’s what you mean. I thought about solving the problem by making multiple storytellers work together, but simply having an AI Assistant who will lay down basic stuff for you to keep the player busy while you come up with something cool might be better. I was worried about two storytellers being unable to create a coherent experience.


    Indeed, you’re correct. This is why you would probably have to make the game so that one or two storytellers can entertain 4-16 storyplayers. It feels strange to me that not everyone would want to be the storyteller, but I am a professional game designer after all. Maybe my views on this are different than normal.

    Your comments on content filtration made me think of Spore, actually. It tackled a lot of the same problems with player-generated content and I think it did an OK job. And we have it even easier, because only a minority of players need to be storytellers for the game to work.

    Also note that when I say “story” it doesn’t really have to be a thematic, narrative, verbal experience with characters and relationships. This could essentially be one long action scene, maybe with a narrative interlude in the middle. L4D-ish, I mean.

  5. Rafael

    I understood that we are talking about a colaborative experience. My idea of story points are meant to prevent abuse from storytellers. In fact, the players could agree on the amount of sp that the storytellers will have. If they know that they are playing with an “storykiller”, they could limit the sp available to easy things (at least here in my city, I came across a lot of Game Masters whose goal was solely punish her players). On the other hand, if they know that the storyteller just want to make a good story, and will not abuse her power, so give her infinity sp.

  6. Nels Anderson

    When discussing this idea, I think it’s very important to avoid the notions of dialog, writing, etc. (Not saying you were, just being emphatic) I feel he terms DM/GM (while inherited from tabletop RPGs) are much better than “storyteller” as storyteller evokes narration in the traditional sense. We have to avoiding traditional notions of narrative (and make it clear this isn’t the intent) because a) the vast majority of people are quite poor at it and b) improving these skills have little to do with video games.

    Versus mode in L4D works well because the players learn the special Infected’s abilities and tactics as they’re on the receiving end as Survivors. You learn by playing. For what your proposing to be successful, players need to be able to become better GMs/storytellers by playing the game. I think this model needs to be based around creating emergent/procedural narrative through gameplay experiences.

    As for what this would look like and more importantly how it would be made playable, I’ve got to think about it a lot more.

    As an aside, there are some conceptually related projects to look at. Players creating new levels with mod tools (and Little Big Planet) are attempting to accomplish the same thing here, only static environments. Spore does the same with lightweight 3d model creation. The question this becomes, how do we do the same for gameplay and what tools would be needed to support it?

    You going to the GDC this year Tynan? This is a conversation that might need to be continued over beer.

  7. Tynan Sylvester Post author


    I can definitely see at least two game modes: “trusted storyteller” and “public storyteller”. Public storytellers are limited by some system, trusted storytellers are limited only by the performance of the computer.


    You’re correct that very few people can write a “story” worth anything in the traditional sense. So I agree, the narrative elements they’d be throwing at the player would mostly be just various forms of action challenges. So DM (or GM, which I think it better and more general) are better terms to use.

    I have another concern now. If the GM is just throwing gameplay at the players (because real people suck at making “story”), would the players really feel gratified about defeating that gameplay? Does it seem like a GM who balances the challenges well against his players’ abilities is “handholding” them? I wonder if the knowledge that the experience is fundamentally artificial would make it less powerful. Of course, you could say that the AI Director in L4D would have the same problem but it seems to work fine.

    “players need to be able to become better GMs/storytellers by playing the game.” I think you definitely could simply by watching what good GMs do to you and learning from what they do that works.

    LBP and Spore would be really important references, I think. The main difference between those and what I’m describing is that the GM experience is in real time. You’re reacting dynamically to what players do, as opposed to composing something (a level, a creature) and just farting it out onto the Internet. The closest reference I can think of to exactly what I’m talking about is the Zombie Master mod for Steam, but it’s not a very good implementation. It’s hard to learn from because it’s got so many design and tech problems.

    Fraid I won’t be at GDC, 2K Boston doesn’t love me enough to send me down there. Yet. We could simply get drunk and blog to each other if you want to have this conversation over beer though ;)

    I’m gonna have to let this whole thing simmer for a while I think since my mindspace was just suddenly filled up by some work stuff. But I definitely want to come back to this and try to come up with a few actual game models to think about. We’re getting to the limit of discussion without more specifics.

  8. Nels Anderson


    Drunk blogging? I’ll get the hooch! ;) I’m definitely going to be thinking about this a lot more too.

    You’re definitely right in that one major consideration here is how things will be when the GM is running things live(-ish), rather than totally offline e.g. LBP and Spore.

    I think the ideal GM/player interaction would be when the players feel that they’ve survived just by the skin of their teeth and if they had performed poorly, they would have failed (and indeed, failure should be a real possibility). That’s where I try to get all the tabletop games I’ve GMed and played in to be and it seems to work well.

    That’s the ideal difficulty space for single player games as well. I guess the big question is now well that will translate into situations that involve other players, where the expectations are different.

    Maybe the best way to start is to consider a handful of specific things that a human GM would be good at that a more sophisticated AI Director wouldn’t be (because given the choice, there’s a lot of more consistency of quality in the AI Director). Building up some rough drafts of gameplay that build upon those things might be easier from there.

  9. David R. Lorentz

    Why the focus on establishing a point system to rate stories? I don’t think a judgmental atmosphere would be the best way to foster interesting stories.

    In tabletop RPGs, the DM is not rewarded for telling a good story. What motivates him, I think, is mainly social pressure – he wants to entertain and impress his friends. The same type of thing can probably be established in a video game if the game enables social groups to connect and remain tightly connected, to spend all their play time together – like a really heavy guild system, I guess.

    Storytelling needs to be a cooperative experience, and it needs to be a social experience, otherwise no storyteller’s going to give a damn. I don’t think good stories could ever emerge from the anonymous atmosphere typical to online games.

  10. Moni

    Interesting idea! I’d love to see more DMing in games, to help with adaptive pacing and such.

    One experience I’d like to relate here is with L4D. I see most people mention the VS mode, playing the infected. With one of my friend, we actually used the debug mode while playing on his own server, which allowed him to die as a player and fly around as a ghost as long as we didn’t rescue him from the “lockers”. From there, he used debug commands that allow one to spawn the special infecteds where you’re looking at, and also to launch the “panic” waves of zombies. The regular Director was turned off for everything except the regular random zombies spawned here and there.

    What this meant is that the Director did provide a basic layer of regular zombies, which allowed my friend to take his time to find the best places to put special infecteds, and to observe the best time to launch big waves. One great thing about this is that placement could be way more intelligent than what the Director does (when you look at it in debug mode, you realize how *simple* it is… and you start seeing the patterns after a while), and pacing could be slightly less predictable.

    The bad side of it is that having all that control leads people to some exageration. My friend loved sending big waves… a lot :). It ended up being frustrating at times, and this is where a “point” system would be interesting : you’d have to follow some pre-established pacing rules if you’re starting as a DM, and the more thumbs-up you’d get from players, the more freedom you would have. This should ensure DMs don’t abuse it… If you want power, you have to be responsible about it, players being the judges about this.

Comments are closed.