Recommendation request: Simulation and Emergent Narrative in Interactive Fiction

As I’ve mentioned elsewhere on the forums I have pretty much no experience playing traditional IF games and draw a lot of my inspiration to want to make IF from non-IF games. Two large influences I would say have been Dwarf Fortress and Deus Ex and both of these games have gotten me interested in immersive simulations, not just as a genre of game often associated with games such as Deus Ex, but as a form of design philosophy. A large thing that has attracted me to inform or parser IF generally has been the idea of truly open ended game design in both terms of gameplay and narrative. I feel these two games showcase these two ideas, where Dwarf Fortress is an example of great levels of simulation which lead to emergent narratives, and Deus Ex an example of open ended gameplay where players can approach the world and the narrative in several ways.

As someone who is rather unfamiliar with traditional IF (non-AIF) I was curious if there were IF games that relied heavily on simulation or emergent narratives. Parser based games are primarily what I am looking for but I would also be open to any games that use these sorts of design philosophies regardless of system. Thanks in advance.


Ascension of Limbs from the 2020 IFComp is largely a resource management game, but with an emergent story.

1 Like

I think Inkle’s Heaven’s Vault could fit the bill of emergent narrative. It’s not parser based, but it relies on procedural generation. It uses Ink and there are some talks on youtube from the creators.

1 Like

Kerkerkruip (Kerkerkruip - Details) is an Inform roguelike. The story is therefore highly dynamic depending on the random map and the player’s choices going through it. Admittedly, that’s dynamic within the roguelike domain of “you fight a bunch of monsters in rooms”.

First of all, I’d have to say that pretty much everything over at Fallen London is based on emergent narrative. Their QBN format encourages it.

Second, open-based game play in a parser world isn’t too hard to design, and I’m sure some of the more experienced folks will chime in with a list of games like that.

But open-ended with both narrative and gameplay? That’s… a bit tricky. Open-ended game play means you wander around and do what you want, figuring out any “puzzles” or overcoming obstacles as you see fit. Call it a “sandbox game” if you like.

But open-ended narrative? What is that, exactly? Procedurally generated? Some kind of AI Dungeon thing with GPL algorithms generating the text? Or some poor sod of an author writing ten thousand small chunks of text that the player encounters in random order?

At some point, the goals of both open-ended play and open-ended narrative are going to collide. Wandering around a game world where everything is random, nothing matters, and there is no narrative cohesion sounds like a surrealist hell. Why are you even in that world?

I guess when I think of a game that is both open ended in terms of gameplay and narrative I’m thinking mainly of (but I’m sure other types of games could also be possible) a life sim sort of game, where the setting and other aspects of the “initial state” of the world may be hand crafted (procedural generation could also possibly work but then I think you get more into the territory of AI dungeon where things don’t make sense for long, or No Man’s Sky where everything becomes very repetitive and shallow) but after that initial state it is then the open ended where the simulation not only takes care of emergent gameplay such as the simulation of windows in Deus Ex to create unique/creative solutions to the designed world, but to also effect narrative elements which are simulated.

This talk from GDC delves into a bit of what I’m talking about. Furthering the example given in this talk with the dove and the ant, let’s say the player controls the dove. An emergent gameplay of this game would then be the possibility of the dove to push the ant into the stream, or if the ant has fallen in on their own the player’s ability to help the ant or not (or to even be completely unaware of the situation if they aren’t near the stream). Emergent narrative then comes in the form of how these systems respond to the players choice, for example if the player saves the ant, and then needs saving themself, the ant may then save the dove/player.

Speaking entirely personally, and definitely not on behalf of anyone else here, my gut reaction is to say “but that’s not really a narrative, no matter how hard you try to push the words in that direction.” I tend to think that “emergent narrative” is necessarily a contradiction in terms, because, for a sequence of events to be “a narrative,” there has to be a meaningful connection between them. A series of random events is by definition not meaningful, so it can’t be a narrative.

So when I pick up my copy of Les Miserables and re-read it, which I’ve been meaning to do for quite a while now, there’s a narrative there: it’s the story of Jean Valjean, how his circumstances formed him and how his soul was saved, and what happens after that. It’s also the story of Inspector Javert, and how his mechanical notion of justice misses important things about Valjean, and how his commitment to a soulless notion of blind adherence to an institution destroys him; and it’s the story of Valjean’s adopted daughter, and of her love, and of the French Revolution, and of how the Paris sewers were built, and of many many other things besides; but it’s a narrative specifically because all of these things are connected in some meaningful way.

Similarly, David Mitchel’s novel Cloud Atlas (and the rather regrettable Tom Hanks movie that was made based on it) is a series of stories in different times and in very different places, but the stories are all interlinked in meaningful ways: what happens on the 18th century Pacific voyage gives insights into what happens at the California nuclear power plant in the 1970s and to the actions of a tribe of low-technology people living in postapocalyptic Hawaii. Each of those six or seven segments has both internal narrative coherence because the actions are meaningfully aligned with each other; and each of them “appropriately belongs with” the other segments because they have meaningful causal and thematic links between the segments. But it’s meaningfully a narrative because there’s a structured coherence between the parts.

But when you’re talking about a series of events that occur while playing The Sims, for instance, you’re talking about a series of events that, all in all, lack meaningful coherence; they’re more or less randomly determined. They could have gone in a different direction. There’s no writer-imposed higher-meaning interpretation possible of the sequence of random events to detect; though the player can make one up as they go along.

Now, you might be able to fake narrative on these terms: the human brain is a pattern-seeking device that looks for patterns and sometimes finds them, even falsely detecting intentionality where there is none: that pattern of stars was probably not put up in the sky in those exactly places because it’s the immortal remains of the hunter Orion. Rather, those particular stars happen to be visible from this particular location in this arm of the Milky Way galaxy, and humans produced a story about why it looks that way that falsely attributes a kind of causality that doesn’t actually explain the real causes of the position of the stars. It’s a more interesting story, but not an accurate one, and it doesn’t meaningfully interpret the facts of the case.

Part of sitting down to read a novel, or watch a play or a movie, or listen to your friend tell a story, involves willingly engaging the pattern-seeking part of your brain that constructs explanations for questions about how different parts of a narrative are connected. In Les Miserables or Cloud Atlas, those connections occur at multiple levels – causal, thematic, ontological, at the level of plot structure, psychological, etc. – whereas when you’re talking about a sequence of events in The Sims, they’re largely random, and the narratives that a player builds in their head don’t match up in a meaningful way with any mechanism of meaning constructed the author during the process of writing the game.

So I tend to think that any “narrative” that’s genuinely emergent, and not just loosely structured, isn’t “really narrative.”

Just my two cents’ worth.


FL isn’t what I think of as emergent narrative. There are choices here and there that bar other choices, but for the most part, everyone goes through all the same stuff and gets to the same place with characters that end up pretty much alike. Now I’m a fan of FL. A sig I’ve seen on an RPG forum is “No one complains about being railroaded if it’s a railroad to awesometown and the scenery is excellent.” With a fun world to explore and 15 zillion paths to take (and a feeling of more just around the corner), that most of those paths are internally fixed doesn’t get in the way of having a lot of fun.


I definitely feel like a difficult at times can be the spectrum between linear narratives and less/non-linear ones. I’ve never played Fallen London or iirc I think sorcery and 80 days were either the same company or similar design philosophies, but have heard about them a bit especially through GDC. I do think they offer interesting lessons on the less linear side of the spectrum of games (the GDC talk on Sorcery! was interesting in terms of some of what I’d say is closer to emergent narrative relating to the different knowledge states). But i think there is a difference between a sort of simulation focus (which can still be graphed and shown as a branching narrative in ways such as the video I linked above) and a game with a zillion paths. I think a big distinction comes down to design, with simulation having a focus on systemic design (the systems are what allow the narratives to emerge rather then choices being pre-written) but I’m hesitant to draw any hard line (and I definitely still appreciate the recommendation, it makes me want to take the time to actually play these games not just hear about them).

I think the main place I disagree with your reasoning is (or agree with your reasoning but disagree with the conclusion) is that emergent can’t/doesn’t have meaningful causal and thematic links. At least my own thoughts when it comes to systemic design in terms of narrative is that instead of these causal/thematic links being hard written they instead follow rules or systems.

The reason the games like the sims, or dwarf fortress work is because they aren’t random, they do have cohesion. If your sims mood was completely random, players wouldn’t be able to form narratives, or if they did they would be rather flimsy ones, but instead a sims mood is influenced by a lot of things that create both the personality of that sim, and the way it interacts with the world and other sims, creating a cohesive narrative. If you have a sim for example that is a shut in, they only go to work and take care of their other needs, eventually your sim is going to be upset, perhaps even spiral into depressive episodes you can’t help them through, and this isn’t because of randomness but because of the systems of the game creating a story about someone who’s so focused on work and their basic needs they don’t make time for anything else.

Similarly with Dwarf Fortress the moods of your dawrves aren’t random but instead influenced by all of the things happening within your fort, dwares relationships are simulated where someone dying influences the relationship of their friends, someone throwing a tantrum influences the moods of those around them, the prosperity of your fort influences their moods, not to mention the histories the game generates with stories of long forgotten beasts which may rearth to attack you, or ancient settlements which may be repurposed over time. A lot of people who like DF don’t even play the game themselves but instead engage with the stories other people have with the game, such as youtubers like kruggsmash. And these stories often times even have common themes, such as the hubris of digging too deep. It’s just instead of the author writing a predetermined theme or narrative, the author creates rules to allow these different themes and narrative emerge depending on what the player does.

For me a large reason I like this approach is that I feel it makes the narrative itself interactive, opening up choices even more through systems that respond to player choice rather then the author having to pre-define all of the players choices. Which isn’t to say these games are completely open ended or simulate everything (though DF seems to have that goal), in DF you are still constrained to playing as a group of dwarves creating fortress, or if you only pay attention to the history as narrative you are constrained to this particular setting. But things are open ended enough where its not just the author telling a story, but the player telling a story with the author through their interaction with the various systems.

I think one thing that separates AI dungeon from the kinds of games I’m talking about is simulation. AI dungeon is an interesting example of open-ended narrative, akin to ai novel generation (I don’t know much about the technology of AI dungeon so it could be these are actually the same thing). But i think the main difference I see is that AI dungeon does not have a level of simulation, it’s purely randomly/procedurally generated, without any rules for internal narrative/gameplay cohesion. I would say it lies more in the realm of what @patrick_mooney is talking about, where connections between narrative points often being flimsy to non-sensical.

So if AI dungeon is kind of at the extreme of open ended narrative I’m talking about something that is a bit more on the linear side of the spectrum from it which still has an underlying simulation, and logic, from which the narrative emerges. (This can be faked a little in AI dungeon by giving it a lot of information to work with but even then the AI eventually loses cohesion due to the innate randomness and lack of simulation/state).

So I guess really where these two things intersect, emergent gameplay, and emergent narrative, is around state, which is simulated. So returning to examples I’ve used, being able to break a window because all windows are simulated to be breakable, would be the emergent gameplay, other characters reacting to that broken window, and even using it because this is simulated behavior is then emergent narrative.

This is a very interesting thread, and I definitely am enjoying all the various thoughts and ideas people have shared. In fact, I do believe I love talking about games more than I do playing them.

However, I could not help but think that, if you build a big enough emergent gameplay world that also has an emergent narrative based on rational responses to environmental changes, well shit… you’re just recreating what happens when you walk out of your front door :sun_with_face: And at that point, it’s time to switch off the computer.

1 Like

I think Failbetter described it as “fires in the desert” - similar but different. On a desert landscape at night there are narrative events represented by multiple glowing fires. A player can go toward any, and it kind of glosses over how they get there and why, leaving it to the player’s imagination to fill in what happens during the “after several hours of walking” part.

A better example might be a haphazard encounter when I was playing Fable 2:

mild adult shenanigans

Fable 2 has NPC behavior and relationship systems. I was in a seedy ramshackle village, and got a tattoo from an NPC and was conversing and therefore building up relationship points with him. One of the local “business ladies” wandered in and - long story short - transactions occurred. Since I had been chatting up the tattooist and he was nearby and we were friends he also got involved (this is a system built into the game… I suppose it was locational - we were in his store, and I believe the game asked for confirmation from me to allow this to happen). Scene fades to black and then back up. Next thing I know the tattooist is attacking me and the lady has run off and I had to defend myself. What the heck went wrong in that scenario! Had they both conspired to waylay me and steal my gold? Was he secretly in love with her and jealous? Was he actually in love with me? Was there a consent issue that had fatal repercussions? All unscripted emergent narrative I inferred from witnessing the systems interact.


I think Inkle’s Overboard! is a bit like this. There are a small number of well developed characters who can be in various places at various times depending on the player’s actions, and Inkle’s twitter account has reported some narrative effects that they didn’t expect.

“Can things happen that the developer didn’t expect?” is a good benchmark for emergence, I think.

ETA: To a lesser extent, 80 Days by the same company also qualifies by that benchmark – I think the authors have said they’ve heard of players beating it in MUCH fewer days than they can.


A thing about games, and really all stories, is that they are about things which one does experience “by switching off the computer” but often because of perspective, or circumstance or what have you, there are experiences which people don’t have and possibly can’t have. I read an article some time ago about the death of the player,, which talks about a few different games(content warning for one AIF game mentioned in the essay which contains nsfw language). One game created by the author of the essay Mattie Brice, EAT is very much a simulation game, perhaps one of the most realistic simulations games ever made, since it uses reality and not a computer to simulate it’s systems.

I think EAT, despite being a game about switching off your computer and walking out the front door literally, demonstrates the value of these realistic narratives because even though these experiences exist for some people by walking outside, it allows authors to put the player in that position when they wouldn’t be able to experience those sorts of things in real life. While I do think EAT would lose something being a traditional video game (it’d be easier for the player to game the system, and might not be as resonate as actually experiencing it) as Brice mentions actually physically interacting with the game isn’t necessary to playing it and understanding it, but I do think there is something games, and in particular these emergent simulated games can tell us about experiences we have never and could never truly experience.

The reason I think simulation/emergence is important here is that while hearing/being told/experiencing someone else’s story through a more authored experiencing can still be moving, there is something different about experiencing it for oneself which I think these systems accomplish.

But all of this isn’t to say that these sorts of simulated/emergent narratives have to be realistic exactly (that’s just my personal interest). A lot can be said, or added to a game by also changing these rules, or recontexualizing things. Genre is a great example of this such as the cyberpunk genre which has a lot to say about technology, capitalism and what it means to be human, all very real things, but which also contains many unrealistic elements such as the nature of that technology itself.


I love it when people share links to discussions about games and game theory. But I’m gonna be honest, I couldn’t make heads or tale out of what Mattie Brice was trying to say or do or even what the rules of this EAT game were supposed to be, other than you have to go exactly 3.5 miles from your house, sit around a school all day, and budget your money.

I guess at some point I just get lost.